Authors: Koushik Chowdhury
Abstract: This paper examines the vulnerabilities of convolutional neural networks (CNNs) to adversarial attacks and explores a method for their safeguarding. In this study, CNNs were implemented on four of the most common image datasets, namely CIFAR-10, ImageNet, MNIST, and Fashion-MNIST, and achieved high baseline accuracy. To assess the strength of these models, the Fast Gradient Sign Method was used, which is a type of exploit on the model that is used to bring down the models accuracies by adding a very minimal perturbation to the input image. To counter the FGSM attack, a safeguarding approach went through, which includes retraining the models on clear and pollutant or adversarial images to increase their resistance ability. The next step involves applying FGSM again, but this time to the adversarially trained models, to see how much the accuracy of the models has gone down and evaluate the effectiveness of the defense. It appears that while most level of robustness is achieved against the models after adversarial training, there are still a few losses in the performance of these models against adversarial perturbations. This work emphasizes the need to create better defenses for models deployed in real-world scenarios against adversaries.
Authors: Weihua Du, Yiming Yang, Sean Welleck
Abstract: Multi-sample aggregation strategies, such as majority voting and best-of-N sampling, are widely used in contemporary large language models (LLMs) to enhance predictive accuracy across various tasks. A key challenge in this process is temperature selection, which significantly impacts model performance. Existing approaches either rely on a fixed default temperature or require labeled validation data for tuning, which are often scarce and difficult to obtain. This paper addresses the challenge of automatically identifying the (near)-optimal temperature for different LLMs using multi-sample aggregation strategies, without relying on task-specific validation data. We provide a comprehensive analysis of temperature's role in performance optimization, considering variations in model architectures, datasets, task types, model sizes, and predictive accuracy. Furthermore, we propose a novel entropy-based metric for automated temperature optimization, which consistently outperforms fixed-temperature baselines. Additionally, we incorporate a stochastic process model to enhance interpretability, offering deeper insights into the relationship between temperature and model performance.
Authors: Sarwar Saif, MD Abdullah Al Nasim, Parag Biswas, Abdur Rashid, MD Mahim Anjum Haque, Md. Zihad Bin Jahangir
Abstract: Federated learning, also known as FL, is a machine learning framework in which a significant amount of clients (such as mobile devices or whole enterprises) collaborate to collaboratively train a model while keeping decentralized training data, all overseen by a central server (such as a service provider). There are advantages in terms of privacy, security, regulations, and economy with this decentralized approach to model training. FL is not impervious to the flaws that plague conventional machine learning models, despite its seeming promise. This study offers a thorough analysis of the fundamental ideas and elements of federated learning architectures, emphasizing five important areas: communication architectures, machine learning models, data partitioning, privacy methods, and system heterogeneity. We additionally address the difficulties and potential paths for future study in the area. Furthermore, based on a comprehensive review of the literature, we present a collection of architectural patterns for federated learning systems. This analysis will help to understand the basic of Federated learning, the primary components of FL, and also about several architectural details.
Authors: Lucas Langlade, Julien Ferry, Gabriel Laberge, Thibaut Vidal
Abstract: We introduce an enumeration-free method based on mathematical programming to precisely characterize various properties such as fairness or sparsity within the set of "good models", known as Rashomon set. This approach is generically applicable to any hypothesis class, provided that a mathematical formulation of the model learning task exists. It offers a structured framework to define the notion of business necessity and evaluate how fairness can be improved or degraded towards a specific protected group, while remaining within the Rashomon set and maintaining any desired sparsity level. We apply our approach to two hypothesis classes: scoring systems and decision diagrams, leveraging recent mathematical programming formulations for training such models. As seen in our experiments, the method comprehensively and certifiably quantifies trade-offs between predictive performance, sparsity, and fairness. We observe that a wide range of fairness values are attainable, ranging from highly favorable to significantly unfavorable for a protected group, while staying within less than 1% of the best possible training accuracy for the hypothesis class. Additionally, we observe that sparsity constraints limit these trade-offs and may disproportionately harm specific subgroups. As we evidenced, thoroughly characterizing the tensions between these key aspects is critical for an informed and accountable selection of models.
Authors: Miruna Oprescu, David K. Park, Xihaier Luo, Shinjae Yoo, Nathan Kallus
Abstract: Estimating causal effects from spatiotemporal data is a key challenge in fields such as public health, social policy, and environmental science, where controlled experiments are often infeasible. However, existing causal inference methods relying on observational data face significant limitations: they depend on strong structural assumptions to address spatiotemporal challenges $\unicode{x2013}$ such as interference, spatial confounding, and temporal carryover effects $\unicode{x2013}$ or fail to account for $\textit{time-varying confounders}$. These confounders, influenced by past treatments and outcomes, can themselves shape future treatments and outcomes, creating feedback loops that complicate traditional adjustment strategies. To address these challenges, we introduce the $\textbf{GST-UNet}$ ($\textbf{G}$-computation $\textbf{S}$patio-$\textbf{T}$emporal $\textbf{UNet}$), a novel end-to-end neural network framework designed to estimate treatment effects in complex spatial and temporal settings. The GST-UNet leverages regression-based iterative G-computation to explicitly adjust for time-varying confounders, providing valid estimates of potential outcomes and treatment effects. To the best of our knowledge, the GST-UNet is the first neural model to account for complex, non-linear dynamics and time-varying confounders in spatiotemporal interventions. We demonstrate the effectiveness of the GST-UNet through extensive simulation studies and showcase its practical utility with a real-world analysis of the impact of wildfire smoke on respiratory hospitalizations during the 2018 California Camp Fire. Our results highlight the potential of GST-UNet to advance spatiotemporal causal inference across a wide range of policy-driven and scientific applications.
Authors: Liu Ziyin, Yizhou Xu, Tomaso Poggio, Isaac Chuang
Abstract: The dynamics of learning in modern large AI systems is hierarchical, often characterized by abrupt, qualitative shifts akin to phase transitions observed in physical systems. While these phenomena hold promise for uncovering the mechanisms behind neural networks and language models, existing theories remain fragmented, addressing specific cases. In this paper, we posit that parameter symmetry breaking and restoration serve as a unifying mechanism underlying these behaviors. We synthesize prior observations and show how this mechanism explains three distinct hierarchies in neural networks: learning dynamics, model complexity, and representation formation. By connecting these hierarchies, we highlight symmetry -- a cornerstone of theoretical physics -- as a potential fundamental principle in modern AI.
Authors: Fernando Llorente, Daniel Waxman, Petar M. Djuri\'c
Abstract: Flexible and scalable decentralized learning solutions are fundamentally important in the application of multi-agent systems. While several recent approaches introduce (ensembles of) kernel machines in the distributed setting, Bayesian solutions are much more limited. We introduce a fully decentralized, asymptotically exact solution to computing the random feature approximation of Gaussian processes. We further address the choice of hyperparameters by introducing an ensembling scheme for Bayesian multiple kernel learning based on online Bayesian model averaging. The resulting algorithm is tested against Bayesian and frequentist methods on simulated and real-world datasets.
Authors: Alice Gorg\'e, Julien Ferry, S\'ebastien Gambs, Thibaut Vidal
Abstract: Recent research has shown that machine learning models are vulnerable to privacy attacks targeting their training data. Differential privacy (DP) has become a widely adopted countermeasure, as it offers rigorous privacy protections. In this paper, we introduce a reconstruction attack targeting state-of-the-art $\varepsilon$-DP random forests. By leveraging a constraint programming model that incorporates knowledge of the forest's structure and DP mechanism characteristics, our approach formally reconstructs the most likely dataset that could have produced a given forest. Through extensive computational experiments, we examine the interplay between model utility, privacy guarantees, and reconstruction accuracy across various configurations. Our results reveal that random forests trained with meaningful DP guarantees can still leak substantial portions of their training data. Specifically, while DP reduces the success of reconstruction attacks, the only forests fully robust to our attack exhibit predictive performance no better than a constant classifier. Building on these insights, we provide practical recommendations for the construction of DP random forests that are more resilient to reconstruction attacks and maintain non-trivial predictive performance.
Authors: Elaheh Jafarigol, Behnoud Alaghband, Azadeh Gilanpour, Saeid Hosseinipoor, Mirhamed Mirmozafari
Abstract: We present a review of high-performance automatic modulation recognition (AMR) models proposed in the literature to classify various Radio Frequency (RF) modulation schemes. We replicated these models and compared their performance in terms of accuracy across a range of signal-to-noise ratios. To ensure a fair comparison, we used the same dataset (RadioML-2016A), the same hardware, and a consistent definition of test accuracy as the evaluation metric, thereby providing a benchmark for future AMR studies. The hyperparameters were selected based on the authors' suggestions in the associated references to achieve results as close as possible to the originals. The replicated models are publicly accessible for further analysis of AMR models. We also present the test accuracies of the selected models versus their number of parameters, indicating their complexities. Building on this comparative analysis, we identify strategies to enhance these models' performance. Finally, we present potential opportunities for improvement, whether through novel architectures, data processing techniques, or training strategies, to further advance the capabilities of AMR models.
Authors: Kevin Han Huang, Ni Zhan, Elif Ertekin, Peter Orbanz, Ryan P. Adams
Abstract: Incorporating group symmetries into neural networks has been a cornerstone of success in many AI-for-science applications. Diagonal groups of isometries, which describe the invariance under a simultaneous movement of multiple objects, arise naturally in many-body quantum problems. Despite their importance, diagonal groups have received relatively little attention, as they lack a natural choice of invariant maps except in special cases. We study different ways of incorporating diagonal invariance in neural network ans\"atze trained via variational Monte Carlo methods, and consider specifically data augmentation, group averaging and canonicalization. We show that, contrary to standard ML setups, in-training symmetrization destabilizes training and can lead to worse performance. Our theoretical and numerical results indicate that this unexpected behavior may arise from a unique computational-statistical tradeoff not found in standard ML analyses of symmetrization. Meanwhile, we demonstrate that post hoc averaging is less sensitive to such tradeoffs and emerges as a simple, flexible and effective method for improving neural network solvers.
Authors: Asaph Matheus Barbosa, Thao Vy Nhat Ngo, Elaheh Jafarigol, Theodore B. Trafalis, Emuobosa P. Ojoboh
Abstract: The goal of this paper is to predict the Remaining Useful Life (RUL) of turbine jet engines using a federated machine learning framework. Federated Learning enables multiple edge devices/nodes or servers to collaboratively train a shared model without sharing sensitive data, thus preserving data privacy and security. By implementing a nonlinear model, the system aims to capture complex relationships and patterns in the engine data to enhance the accuracy of RUL predictions. This approach leverages decentralized computation, allowing models to be trained locally at each device before aggregating the learned weights at a central server. By predicting the RUL of jet engines accurately, maintenance schedules can be optimized, downtime reduced, and operational efficiency improved, ultimately leading to cost savings and enhanced performance in the aviation industry. Computational results are provided by using the C-MAPSS dataset which is publicly available on the NASA website and is a valuable resource for studying and analyzing engine degradation behaviors in various operational scenarios.
Authors: Awa Khouna, Julien Ferry, Thibaut Vidal
Abstract: The advent of Machine Learning as a Service (MLaaS) has heightened the trade-off between model explainability and security. In particular, explainability techniques, such as counterfactual explanations, inadvertently increase the risk of model extraction attacks, enabling unauthorized replication of proprietary models. In this paper, we formalize and characterize the risks and inherent complexity of model reconstruction, focusing on the "oracle'' queries required for faithfully inferring the underlying prediction function. We present the first formal analysis of model extraction attacks through the lens of competitive analysis, establishing a foundational framework to evaluate their efficiency. Focusing on models based on additive decision trees (e.g., decision trees, gradient boosting, and random forests), we introduce novel reconstruction algorithms that achieve provably perfect fidelity while demonstrating strong anytime performance. Our framework provides theoretical bounds on the query complexity for extracting tree-based model, offering new insights into the security vulnerabilities of their deployment.
Authors: Benjamin J. Choi
Abstract: Electromyogenic (EMG) noise is a major contamination source in EEG data that can impede accurate analysis of brain-specific neural activity. Recent literature on EMG artifact removal has moved beyond traditional linear algorithms in favor of machine learning-based systems. However, existing deep learning-based filtration methods often have large compute footprints and prohibitively long training times. In this study, we present a new machine learning-based system for filtering EMG interference from EEG data using an autoencoder-targeted adversarial transformer (AT-AT). By leveraging the lightweight expressivity of an autoencoder to determine optimal time-series transformer application sites, our AT-AT architecture achieves a >90% model size reduction compared to published artifact removal models. The addition of adversarial training ensures that filtered signals adhere to the fundamental characteristics of EEG data. We trained AT-AT using published neural data from 67 subjects and found that the system was able to achieve comparable test performance to larger models; AT-AT posted a mean reconstructive correlation coefficient above 0.95 at an initial signal-to-noise ratio (SNR) of 2 dB and 0.70 at -7 dB SNR. Further research generalizing these results to broader sample sizes beyond these isolated test cases will be crucial; while outside the scope of this study, we also include results from a real-world deployment of AT-AT in the Appendix.
Authors: Benjamin J. Choi
Abstract: Brain-computer interfaces (BCIs) offer transformative potential, but decoding neural signals presents significant challenges. The core premise of this paper is built around demonstrating methods to elucidate the underlying low-dimensional geometric structure present in high-dimensional brainwave data in order to assist in downstream BCI-related neural classification tasks. We demonstrate two pipelines related to electroencephalography (EEG) signal processing: (1) a preliminary pipeline removing noise from individual EEG channels, and (2) a downstream manifold learning pipeline uncovering geometric structure across networks of EEG channels. We conduct preliminary validation using two EEG datasets and situate our demonstration in the context of the BCI-relevant imagined digit decoding problem. Our preliminary pipeline uses an attention-based EEG filtration network to extract clean signal from individual EEG channels. Our primary pipeline uses a fast Fourier transform, a Laplacian eigenmap, a discrete analog of Ricci flow via Ollivier's notion of Ricci curvature, and a graph convolutional network to perform dimensionality reduction on high-dimensional multi-channel EEG data in order to enable regularizable downstream classification. Our system achieves competitive performance with existing signal processing and classification benchmarks; we demonstrate a mean test correlation coefficient of >0.95 at 2 dB on semi-synthetic neural denoising and a downstream EEG-based classification accuracy of 0.97 on distinguishing digit- versus non-digit thoughts. Results are preliminary and our geometric machine learning pipeline should be validated by more extensive follow-up studies; generalizing these results to larger inter-subject sample sizes, different hardware systems, and broader use cases will be crucial.
Authors: Roussel Desmond Nzoyem, David A. W. Barton, Tom Deakin
Abstract: As foundational models reshape scientific discovery, a bottleneck persists in dynamical system reconstruction (DSR): the ability to learn across system hierarchies. Many meta-learning approaches have been applied successfully to single systems, but falter when confronted with sparse, loosely related datasets requiring multiple hierarchies to be learned. Mixture of Experts (MoE) offers a natural paradigm to address these challenges. Despite their potential, we demonstrate that naive MoEs are inadequate for the nuanced demands of hierarchical DSR, largely due to their gradient descent-based gating update mechanism which leads to slow updates and conflicted routing during training. To overcome this limitation, we introduce MixER: Mixture of Expert Reconstructors, a novel sparse top-1 MoE layer employing a custom gating update algorithm based on $K$-means and least squares. Extensive experiments validate MixER's capabilities, demonstrating efficient training and scalability to systems of up to ten parametric ordinary differential equations. However, our layer underperforms state-of-the-art meta-learners in high-data regimes, particularly when each expert is constrained to process only a fraction of a dataset composed of highly related data points. Further analysis with synthetic and neuroscientific time series suggests that the quality of the contextual representations generated by MixER is closely linked to the presence of hierarchical structure in the data.
Authors: Sanghoon Na, Haizhao Yang
Abstract: The curse of dimensionality in neural network optimization under the mean-field regime is studied. It is demonstrated that when a shallow neural network with a Lipschitz continuous activation function is trained using either empirical or population risk to approximate a target function that is $r$ times continuously differentiable on $[0,1]^d$, the population risk may not decay at a rate faster than $t^{-\frac{4r}{d-2r}}$, where $t$ is an analog of the total number of optimization iterations. This result highlights the presence of the curse of dimensionality in the optimization computation required to achieve a desired accuracy. Instead of analyzing parameter evolution directly, the training dynamics are examined through the evolution of the parameter distribution under the 2-Wasserstein gradient flow. Furthermore, it is established that the curse of dimensionality persists when a locally Lipschitz continuous activation function is employed, where the Lipschitz constant in $[-x,x]$ is bounded by $O(x^\delta)$ for any $x \in \mathbb{R}$. In this scenario, the population risk is shown to decay at a rate no faster than $t^{-\frac{(4+2\delta)r}{d-2r}}$. To the best of our knowledge, this work is the first to analyze the impact of function smoothness on the curse of dimensionality in neural network optimization theory.
Authors: Hanfei Yu, Xingqi Cui, Hong Zhang, Hao Wang, Hao Wang
Abstract: Large Language Models (LLMs) have gained immense success in revolutionizing various applications, including content generation, search and recommendation, and AI-assisted operation. To reduce high training costs, Mixture-of-Experts (MoE) architecture has become a popular backbone for modern LLMs. However, despite the benefits, serving MoE-based LLMs experience severe memory inefficiency due to sparsely activated experts. Recent studies propose to offload inactive experts from GPU memory to CPU memory to improve the serving efficiency of MoE models. However, they either incur high inference latency or high model memory footprints due to coarse-grained designs. To tame the latency-memory trade-off in MoE serving, we present fMoE, a fine-grained expert offloading system for MoE serving that achieves low inference latency with memory efficiency. We design fMoE to extract fine-grained expert selection patterns from MoE models and semantic hints from input prompts to efficiently guide expert prefetching, caching, and offloading decisions. fMoE is prototyped on top of HuggingFace Transformers and deployed on a six-GPU testbed. Experiments with open-source MoE models and real-world workloads show that fMoE reduces inference latency by 47% and improves expert hit rate by 36% over state-of-the-art solutions.
Authors: Huchen Yang, Chuanqi Chen, Jin-Long Wu
Abstract: Digital twins have been actively explored in many engineering applications, such as manufacturing and autonomous systems. However, model discrepancy is ubiquitous in most digital twin models and has significant impacts on the performance of using those models. In recent years, data-driven modeling techniques have been demonstrated promising in characterizing the model discrepancy in existing models, while the training data for the learning of model discrepancy is often obtained in an empirical way and an active approach of gathering informative data can potentially benefit the learning of model discrepancy. On the other hand, Bayesian experimental design (BED) provides a systematic approach to gathering the most informative data, but its performance is often negatively impacted by the model discrepancy. In this work, we build on sequential BED and propose an efficient approach to iteratively learn the model discrepancy based on the data from the BED. The performance of the proposed method is validated by a classical numerical example governed by a convection-diffusion equation, for which full BED is still feasible. The proposed method is then further studied in the same numerical example with a high-dimensional model discrepancy, which serves as a demonstration for the scenarios where full BED is not practical anymore. An ensemble-based approximation of information gain is further utilized to assess the data informativeness and to enhance learning model discrepancy. The results show that the proposed method is efficient and robust to the active learning of high-dimensional model discrepancy, using data suggested by the sequential BED. We also demonstrate that the proposed method is compatible with both classical numerical solvers and modern auto-differentiable solvers.
Authors: Chongyu Fan, Jinghan Jia, Yihua Zhang, Anil Ramakrishna, Mingyi Hong, Sijia Liu
Abstract: The LLM unlearning technique has recently been introduced to comply with data regulations and address the safety and ethical concerns of LLMs by removing the undesired data-model influence. However, state-of-the-art unlearning methods face a critical vulnerability: they are susceptible to ``relearning'' the removed information from a small number of forget data points, known as relearning attacks. In this paper, we systematically investigate how to make unlearned models robust against such attacks. For the first time, we establish a connection between robust unlearning and sharpness-aware minimization (SAM) through a unified robust optimization framework, in an analogy to adversarial training designed to defend against adversarial attacks. Our analysis for SAM reveals that smoothness optimization plays a pivotal role in mitigating relearning attacks. Thus, we further explore diverse smoothing strategies to enhance unlearning robustness. Extensive experiments on benchmark datasets, including WMDP and MUSE, demonstrate that SAM and other smoothness optimization approaches consistently improve the resistance of LLM unlearning to relearning attacks. Notably, smoothness-enhanced unlearning also helps defend against (input-level) jailbreaking attacks, broadening our proposal's impact in robustifying LLM unlearning. Codes are available at https://github.com/OPTML-Group/Unlearn-Smooth.
Authors: Reena Elangovan, Charbel Sakr, Anand Raghunathan, Brucek Khailany
Abstract: Post-training quantization (PTQ) is a promising approach to reducing the storage and computational requirements of large language models (LLMs) without additional training cost. Recent PTQ studies have primarily focused on quantizing only weights to sub-8-bits while maintaining activations at 8-bits or higher. Accurate sub-8-bit quantization for both weights and activations without relying on quantization-aware training remains a significant challenge. We propose a novel quantization method called block clustered quantization (BCQ) wherein each operand tensor is decomposed into blocks (a block is a group of contiguous scalars), blocks are clustered based on their statistics, and a dedicated optimal quantization codebook is designed for each cluster. As a specific embodiment of this approach, we propose a PTQ algorithm called Locally-Optimal BCQ (LO-BCQ) that iterates between the steps of block clustering and codebook design to greedily minimize the quantization mean squared error. When weight and activation scalars are encoded to W4A4 format (with 0.5-bits of overhead for storing scaling factors and codebook selectors), we advance the current state-of-the-art by demonstrating <1% loss in inference accuracy across several LLMs and downstream tasks.
Authors: Andreas Mueller
Abstract: Current research in time-series anomaly detection is using definitions that miss critical aspects of how anomaly detection is commonly used in practice. We list several areas that are of practical relevance and that we believe are either under-investigated or missing entirely from the current discourse. Based on an investigation of systems deployed in a cloud environment, we motivate the areas of streaming algorithms, human-in-the-loop scenarios, point processes, conditional anomalies and populations analysis of time series. This paper serves as a motivation and call for action, including opportunities for theoretical and applied research, as well as for building new dataset and benchmarks.
Authors: William Huey, Huaxiaoyue Wang, Anne Wu, Yoav Artzi, Sanjiban Choudhury
Abstract: We examine the problem of learning sequential tasks from a single visual demonstration. A key challenge arises when demonstrations are temporally misaligned due to variations in timing, differences in embodiment, or inconsistencies in execution. Existing approaches treat imitation as a distribution-matching problem, aligning individual frames between the agent and the demonstration. However, we show that such frame-level matching fails to enforce temporal ordering or ensure consistent progress. Our key insight is that matching should instead be defined at the level of sequences. We propose that perfect matching occurs when one sequence successfully covers all the subgoals in the same order as the other sequence. We present ORCA (ORdered Coverage Alignment), a dense per-timestep reward function that measures the probability of the agent covering demonstration frames in the correct order. On temporally misaligned demonstrations, we show that agents trained with the ORCA reward achieve $4.5$x improvement ($0.11 \rightarrow 0.50$ average normalized returns) for Meta-world tasks and $6.6$x improvement ($6.55 \rightarrow 43.3$ average returns) for Humanoid-v4 tasks compared to the best frame-level matching algorithms. We also provide empirical analysis showing that ORCA is robust to varying levels of temporal misalignment. Our code is available at https://github.com/portal-cornell/orca/
Authors: SriVarsha Mulakala, Umesh Vangapally, Benjamin Larkey, Aidan Henrichs, Corey Wojslaw
Abstract: Analyzing stocks and making higher accurate predictions on where the price is heading continues to become more and more challenging therefore, we designed a new financial algorithm that leverages social media sentiment analysis to enhance the prediction of key stock earnings and associated volatility. Our model integrates sentiment analysis and data retrieval techniques to extract critical information from social media, analyze company financials, and compare sentiments between Wall Street and the general public. This approach aims to provide investors with timely data to execute trades based on key events, rather than relying on long-term stock holding strategies. The stock market is characterized by rapid data flow and fluctuating community sentiments, which can significantly impact trading outcomes. Stock forecasting is complex given its stochastic dynamic. Standard traditional prediction methods often overlook key events and media engagement, focusing its practice into long-term investment options. Our research seeks to change the stochastic dynamic to a more predictable environment by examining the impact of media on stock volatility, understanding and identifying sentiment differences between Wall Street and retail investors, and evaluating the impact of various media networks in predicting earning reports.
Authors: Akash Kumar
Abstract: The success of deep networks is crucially attributed to their ability to capture latent features within a representation space. In this work, we investigate whether the underlying learned features of a model can be efficiently retrieved through feedback from an agent, such as a large language model (LLM), in the form of relative \textit{triplet comparisons}. These features may represent various constructs, including dictionaries in LLMs or components of a covariance matrix of Mahalanobis distances. We analyze the feedback complexity associated with learning a feature matrix in sparse settings. Our results establish tight bounds when the agent is permitted to construct activations and demonstrate strong upper bounds in sparse scenarios when the agent's feedback is limited to distributional information. We validate our theoretical findings through experiments on two distinct applications: feature recovery from Recursive Feature Machine-trained models and dictionary extraction from sparse autoencoders trained on Large Language Models.
Authors: Ali Al-Lawati, Jason Lucas, Zhiwei Zhang, Prasenjit Mitra, Suhang Wang
Abstract: In-context learning (ICL) effectively conditions large language models (LLMs) for molecular tasks, such as property prediction and molecule captioning, by embedding carefully selected demonstration examples into the input prompt. This approach avoids the computational overhead of extensive pertaining and fine-tuning. However, current prompt retrieval methods for molecular tasks have relied on molecule feature similarity, such as Morgan fingerprints, which do not adequately capture the global molecular and atom-binding relationships. As a result, these methods fail to represent the full complexity of molecular structures during inference. Moreover, small-to-medium-sized LLMs, which offer simpler deployment requirements in specialized systems, have remained largely unexplored in the molecular ICL literature. To address these gaps, we propose a self-supervised learning technique, GAMIC (Graph-Aligned Molecular In-Context learning, which aligns global molecular structures, represented by graph neural networks (GNNs), with textual captions (descriptions) while leveraging local feature similarity through Morgan fingerprints. In addition, we introduce a Maximum Marginal Relevance (MMR) based diversity heuristic during retrieval to optimize input prompt demonstration samples. Our experimental findings using diverse benchmark datasets show GAMIC outperforms simple Morgan-based ICL retrieval methods across all tasks by up to 45%.
Authors: Ruoyan Li, Dipti Ranjan Sahu, Guy Van den Broeck, Zhe Zeng
Abstract: While deep generative models~(DGMs) have demonstrated remarkable success in capturing complex data distributions, they consistently fail to learn constraints that encode domain knowledge and thus require constraint integration. Existing solutions to this challenge have primarily relied on heuristic methods and often ignore the underlying data distribution, harming the generative performance. In this work, we propose a probabilistically sound approach for enforcing the hard constraints into DGMs to generate constraint-compliant and realistic data. This is achieved by our proposed gradient estimators that allow the constrained distribution, the data distribution conditioned on constraints, to be differentiably learned. We carry out extensive experiments with various DGM model architectures over five image datasets and three scientific applications in which domain knowledge is governed by linear equality constraints. We validate that the standard DGMs almost surely generate data violating the constraints. Among all the constraint integration strategies, ours not only guarantees the satisfaction of constraints in generation but also archives superior generative performance than the other methods across every benchmark.
Authors: Xinyu Yang, Tianqi Chen, Beidi Chen
Abstract: Context-augmented generation (CAG) techniques, including RAG and ICL, require the efficient combination of multiple contexts to generate responses to user queries. Directly inputting these contexts as a sequence introduces a considerable computational burden by re-encoding the combined selection of contexts for every request. To address this, we explore the promising potential of parallel encoding to independently pre-compute and cache each context's KV states. This approach enables the direct loading of cached states during inference while accommodating more contexts through position reuse across contexts. However, due to misalignments in attention distribution, directly applying parallel encoding results in a significant performance drop. To enable effective and efficient CAG, we propose Adaptive Parallel Encoding ($\textbf{APE}$), which brings shared prefix, attention temperature, and scaling factor to align the distribution of parallel encoding with sequential encoding. Results on RAG and ICL tasks demonstrate that APE can preserve 98% and 93% sequential encoding performance using the same inputs while outperforming parallel encoding by 3.6% and 7.9%, respectively. It also scales to many-shot CAG, effectively encoding hundreds of contexts in parallel. Efficiency evaluation shows that APE can achieve an end-to-end 4.5$\times$ speedup by reducing 28$\times$ prefilling time for a 128K-length context.
Authors: Han Qi, Haochen Yang, Qiaosheng Zhang, Zhuoran Yang
Abstract: We study the problem of reinforcement learning from human feedback (RLHF), a critical problem in training large language models, from a theoretical perspective. Our main contribution is the design of novel sample-efficient RLHF algorithms based on information-directed sampling (IDS), an online decision-making principle inspired by information theory. Our algorithms maximize the sum of the value function and a mutual information term that encourages exploration of the unknown environment (which quantifies the information gained about the environment through observed human feedback data). To tackle the challenge of large state spaces and improve sample efficiency, we construct a simplified \emph{surrogate environment} and introduce a novel distance measure (named the \emph{$\ell_g$-distance}), enabling our IDS-based algorithm to achieve a Bayesian regret upper bound of order $O(H^{\frac{3}{2}}\sqrt{\log(K(\epsilon)) T})$, where $H$ is the episode length, $T$ is the number of episode and $K(\epsilon)$ is related to the covering number of the environment. Specializing to the tabular settings, this regret bound is of order $\tilde{O}(H^2\sqrt{SAT})$, where $S$ and $A$ are the numbers of states and actions. Finally, we propose an Approximate-IDS algorithm that is computationally more efficient while maintaining nearly the same sample efficiency. The design principle of this approximate algorithm is not only effective in RLHF settings but also applicable to the standard RL framework. Moreover, our work showcases the value of information theory in reinforcement learning and in the training of large language models.
Authors: Haoye Lu, Qifan Wu, Yaoliang Yu
Abstract: Recent diffusion-based generative models achieve remarkable results by training on massive datasets, yet this practice raises concerns about memorization and copyright infringement. A proposed remedy is to train exclusively on noisy data with potential copyright issues, ensuring the model never observes original content. However, through the lens of deconvolution theory, we show that although it is theoretically feasible to learn the data distribution from noisy samples, the practical challenge of collecting sufficient samples makes successful learning nearly unattainable. To overcome this limitation, we propose to pretrain the model with a small fraction of clean data to guide the deconvolution process. Combined with our Stochastic Forward--Backward Deconvolution (SFBD) method, we attain an FID of $6.31$ on CIFAR-10 with just $4\%$ clean images (and $3.58$ with $10\%$). Theoretically, we prove that SFBD guides the model to learn the true data distribution. The result also highlights the importance of pretraining on limited but clean data or the alternative from similar datasets. Empirical studies further support these findings and offer additional insights.
Authors: Prince Zizhuang Wang, Jinhao Liang, Shuyi Chen, Ferdinando Fioretto, Shixiang Zhu
Abstract: Decision-focused learning (DFL) integrates predictive models with downstream optimization, directly training machine learning models to minimize decision errors. While DFL has been shown to provide substantial advantages when compared to a counterpart that treats the predictive and prescriptive models separately, it has also been shown to struggle in high-dimensional and risk-sensitive settings, limiting its applicability in real-world settings. To address this limitation, this paper introduces decision-focused generative learning (Gen-DFL), a novel framework that leverages generative models to adaptively model uncertainty and improve decision quality. Instead of relying on fixed uncertainty sets, Gen-DFL learns a structured representation of the optimization parameters and samples from the tail regions of the learned distribution to enhance robustness against worst-case scenarios. This approach mitigates over-conservatism while capturing complex dependencies in the parameter space. The paper shows, theoretically, that Gen-DFL achieves improved worst-case performance bounds compared to traditional DFL. Empirically, it evaluates Gen-DFL on various scheduling and logistics problems, demonstrating its strong performance against existing DFL methods.
Authors: Simon Pepin Lehalleur, Jesse Hoogland, Matthew Farrugia-Roberts, Susan Wei, Alexander Gietelink Oldenziel, George Wang, Liam Carroll, Daniel Murfet
Abstract: In this position paper, we argue that understanding the relation between structure in the data distribution and structure in trained models is central to AI alignment. First, we discuss how two neural networks can have equivalent performance on the training set but compute their outputs in essentially different ways and thus generalise differently. For this reason, standard testing and evaluation are insufficient for obtaining assurances of safety for widely deployed generally intelligent systems. We argue that to progress beyond evaluation to a robust mathematical science of AI alignment, we need to develop statistical foundations for an understanding of the relation between structure in the data distribution, internal structure in models, and how these structures underlie generalisation.
Authors: Junqi He, Yifeng Wei, Daiguang Jin
Abstract: This article proposes a Mix Neural Network (MNN) based on CNN-FCNN for predicting magnetic loss of different materials. In traditional magnetic core loss models, empirical equations usually need to be regressed under the same external conditions. When the magnetic core material is different, it needs to be classified and discussed. If external factors increase, multiple models need to be proposed for classification and discussion, making the modeling process extremely cumbersome. And traditional empirical equations still has the problem of low accuracy, although various correction equations have been introduced later, the accuracy has always been unsatisfactory. By introducing machine learning and deep learning, it is possible to simultaneously solve prediction problems with low accuracy of empirical equations and complex conditions. Based on the MagNet database, through the training of the newly proposed MNN, it is found that a single model is sufficient to make predictions for at least four different materials under varying temperatures, frequencies, and waveforms, with accuracy far exceeding that of traditional models. At the same time, we also used three other machine learning and deep learning models (Random Forest, XGBoost, MLP-LSTM) for training, all of which had much higher accuracy than traditional models. On the basis of the predicted results, a hybrid model combining MNN and XGBoost was proposed, which predicted through weighting and found that the accuracy could continue to improve. This provides a solution for modeling magnetic core loss under different materials and operating modes.
Authors: Ya Zhou, Yujie Yang, Jianhuang Gan, Xiangjie Li, Jing Yuan, Wei Zhao
Abstract: Electrocardiogram (ECG) analysis is a fundamental tool for diagnosing cardiovascular conditions, yet anomaly detection in ECG signals remains challenging due to their inherent complexity and variability. We propose Multi-scale Masked Autoencoder for ECG anomaly detection (MMAE-ECG), a novel end-to-end framework that effectively captures both global and local dependencies in ECG data. Unlike state-of-the-art methods that rely on heartbeat segmentation or R-peak detection, MMAE-ECG eliminates the need for such pre-processing steps, enhancing its suitability for clinical deployment. MMAE-ECG partitions ECG signals into non-overlapping segments, with each segment assigned learnable positional embeddings. A novel multi-scale masking strategy and multi-scale attention mechanism, along with distinct positional embeddings, enable a lightweight Transformer encoder to effectively capture both local and global dependencies. The masked segments are then reconstructed using a single-layer Transformer block, with an aggregation strategy employed during inference to refine the outputs. Experimental results demonstrate that our method achieves performance comparable to state-of-the-art approaches while significantly reducing computational complexity-approximately 1/78 of the floating-point operations (FLOPs) required for inference. Ablation studies further validate the effectiveness of each component, highlighting the potential of multi-scale masked autoencoders for anomaly detection.
Authors: Qi Li
Abstract: Outlier detection tasks aim at discovering potential issues or opportunities and are widely used in cybersecurity, financial security, industrial inspection, etc. To date, thousands of outlier detection algorithms have been proposed. Clearly, in real-world scenarios, such a large number of algorithms is unnecessary. In other words, a large number of outlier detection algorithms are redundant. We believe the root cause of this redundancy lies in the current highly customized (i.e., non-generic) optimization strategies. Specifically, when researchers seek to improve the performance of existing outlier detection algorithms, they have to design separate optimized versions tailored to the principles of each algorithm, leading to an ever-growing number of outlier detection algorithms. To address this issue, in this paper, we introduce the explosion from physics into the outlier detection task and propose a generic optimization strategy based on feature explosion, called OSD (Optimization Strategy for outlier Detection algorithms). In the future, when improving the performance of existing outlier detection algorithms, it will be sufficient to invoke the OSD plugin without the need to design customized optimized versions for them. We compared the performances of 14 outlier detection algorithms on 24 datasets before and after invoking the OSD plugin. The experimental results show that the performances of all outlier detection algorithms are improved on almost all datasets. In terms of average accuracy, OSD make these outlier detection algorithms improve by 15% (AUC), 63.7% (AP).
Authors: Larkin Liu, Kashif Rasul, Yutong Chao, Jalal Etesami
Abstract: We present a novel framework for online learning in Stackelberg general-sum games, where two agents, the leader and follower, engage in sequential turn-based interactions. At the core of this approach is a learned diffeomorphism that maps the joint action space to a smooth Riemannian manifold, referred to as the Stackelberg manifold. This mapping, facilitated by neural normalizing flows, ensures the formation of tractable isoplanar subspaces, enabling efficient techniques for online learning. By assuming linearity between the agents' reward functions on the Stackelberg manifold, our construct allows the application of standard bandit algorithms. We then provide a rigorous theoretical basis for regret minimization on convex manifolds and establish finite-time bounds on simple regret for learning Stackelberg equilibria. This integration of manifold learning into game theory uncovers a previously unrecognized potential for neural normalizing flows as an effective tool for multi-agent learning. We present empirical results demonstrating the effectiveness of our approach compared to standard baselines, with applications spanning domains such as cybersecurity and economic supply chain optimization.
Authors: Zinan Lin, Tadas Baltrusaitis, Sergey Yekhanin
Abstract: Differentially private (DP) synthetic data, which closely resembles the original private data while maintaining strong privacy guarantees, has become a key tool for unlocking the value of private data without compromising privacy. Recently, Private Evolution (PE) has emerged as a promising method for generating DP synthetic data. Unlike other training-based approaches, PE only requires access to inference APIs from foundation models, enabling it to harness the power of state-of-the-art models. However, a suitable foundation model for a specific private data domain is not always available. In this paper, we discover that the PE framework is sufficiently general to allow inference APIs beyond foundation models. Specifically, we show that simulators -- such as computer graphics-based image synthesis tools -- can also serve as effective APIs within the PE framework. This insight greatly expands the applicability of PE, enabling the use of a wide variety of domain-specific simulators for DP data synthesis. We explore the potential of this approach, named Sim-PE, in the context of image synthesis. Across three diverse simulators, Sim-PE performs well, improving the downstream classification accuracy of PE by up to 3x and reducing the FID score by up to 80%. We also show that simulators and foundation models can be easily leveraged together within the PE framework to achieve further improvements. The code is open-sourced in the Private Evolution Python library: https://github.com/microsoft/DPSDA.
Authors: Hamed Poursiami, Ayana Moshruba, Maryam Parsa
Abstract: As machine learning models become integral to security-sensitive applications, concerns over data leakage from adversarial attacks continue to rise. Model Inversion (MI) attacks pose a significant privacy threat by enabling adversaries to reconstruct training data from model outputs. While MI attacks on Artificial Neural Networks (ANNs) have been widely studied, Spiking Neural Networks (SNNs) remain largely unexplored in this context. Due to their event-driven and discrete computations, SNNs introduce fundamental differences in information processing that may offer inherent resistance to such attacks. A critical yet underexplored aspect of this threat lies in black-box settings, where attackers operate through queries without direct access to model parameters or gradients-representing a more realistic adversarial scenario in deployed systems. This work presents the first study of black-box MI attacks on SNNs. We adapt a generative adversarial MI framework to the spiking domain by incorporating rate-based encoding for input transformation and decoding mechanisms for output interpretation. Our results show that SNNs exhibit significantly greater resistance to MI attacks than ANNs, as demonstrated by degraded reconstructions, increased instability in attack convergence, and overall reduced attack effectiveness across multiple evaluation metrics. Further analysis suggests that the discrete and temporally distributed nature of SNN decision boundaries disrupts surrogate modeling, limiting the attacker's ability to approximate the target model.
Authors: Bing Sun, Jun Sun, Wei Zhao
Abstract: Despite their advances and success, real-world deep neural networks are known to be vulnerable to adversarial attacks. Universal adversarial perturbation, an input-agnostic attack, poses a serious threat for them to be deployed in security-sensitive systems. In this case, a single universal adversarial perturbation deceives the model on a range of clean inputs without requiring input-specific optimization, which makes it particularly threatening. In this work, we observe that universal adversarial perturbations usually lead to abnormal entropy spectrum in hidden layers, which suggests that the prediction is dominated by a small number of ``feature'' in such cases (rather than democratically by many features). Inspired by this, we propose an efficient yet effective defense method for mitigating UAPs called \emph{Democratic Training} by performing entropy-based model enhancement to suppress the effect of the universal adversarial perturbations in a given model. \emph{Democratic Training} is evaluated with 7 neural networks trained on 5 benchmark datasets and 5 types of state-of-the-art universal adversarial attack methods. The results show that it effectively reduces the attack success rate, improves model robustness and preserves the model accuracy on clean samples.
Authors: Jingang Qu, David Holzm\"uller, Ga\"el Varoquaux, Marine Le Morvan
Abstract: The long-standing dominance of gradient-boosted decision trees on tabular data is currently challenged by tabular foundation models using In-Context Learning (ICL): setting the training data as context for the test data and predicting in a single forward pass without parameter updates. While the very recent TabPFNv2 foundation model (2025) excels on tables with up to 10K samples, its alternating column- and row-wise attentions make handling large training sets computationally prohibitive. So, can ICL be effectively scaled and deliver a benefit for larger tables? We introduce TabICL, a tabular foundation model for classification, pretrained on synthetic datasets with up to 60K samples and capable of handling 500K samples on affordable resources. This is enabled by a novel two-stage architecture: a column-then-row attention mechanism to build fixed-dimensional embeddings of rows, followed by a transformer for efficient ICL. Across 200 classification datasets from the TALENT benchmark, TabICL is on par with TabPFNv2 while being systematically faster (up to 10 times), and significantly outperforms all other approaches. On 56 datasets with over 10K samples, TabICL surpasses both TabPFNv2 and CatBoost, demonstrating the potential of ICL for large data.
Authors: Stefano Zampini, Jacob Christopher, Luca Oneto, Davide Anguita, Ferdinando Fioretto
Abstract: Stable diffusion models represent the state-of-the-art in data synthesis across diverse domains and hold transformative potential for applications in science and engineering, e.g., by facilitating the discovery of novel solutions and simulating systems that are computationally intractable to model explicitly. However, their current utility in these fields is severely limited by an inability to enforce strict adherence to physical laws and domain-specific constraints. Without this grounding, the deployment of such models in critical applications, ranging from material science to safety-critical systems, remains impractical. This paper addresses this fundamental limitation by proposing a novel approach to integrate stable diffusion models with constrained optimization frameworks, enabling them to generate outputs that satisfy stringent physical and functional requirements. We demonstrate the effectiveness of this approach through material science experiments requiring adherence to precise morphometric properties, inverse design problems involving the generation of stress-strain responses using video generation with a simulator in the loop, and safety settings where outputs must avoid copyright infringement.
Authors: Yangguang He, Wenhao Li, Minzhe Li, Juan Zhang, Xiangfeng Wang, Bo Jin
Abstract: State estimation remains a fundamental challenge across numerous domains, from autonomous driving, aircraft tracking to quantum system control. Although Bayesian filtering has been the cornerstone solution, its classical model-based paradigm faces two major limitations: it struggles with inaccurate state space model (SSM) and requires extensive prior knowledge of noise characteristics. We present TrackDiffuser, a generative framework addressing both challenges by reformulating Bayesian filtering as a conditional diffusion model. Our approach implicitly learns system dynamics from data to mitigate the effects of inaccurate SSM, while simultaneously circumventing the need for explicit measurement models and noise priors by establishing a direct relationship between measurements and states. Through an implicit predict-and-update mechanism, TrackDiffuser preserves the interpretability advantage of traditional model-based filtering methods. Extensive experiments demonstrate that our framework substantially outperforms both classical and contemporary hybrid methods, especially in challenging non-linear scenarios involving non-Gaussian noises. Notably, TrackDiffuser exhibits remarkable robustness to SSM inaccuracies, offering a practical solution for real-world state estimation problems where perfect models and prior knowledge are unavailable.
Authors: Diego Calanzone, Pierluca D'Oro, Pierre-Luc Bacon
Abstract: Recent advances in language models have enabled framing molecule generation as sequence modeling. However, existing approaches often rely on single-objective reinforcement learning, limiting their applicability to real-world drug design, where multiple competing properties must be optimized. Traditional multi-objective reinforcement learning (MORL) methods require costly retraining for each new objective combination, making rapid exploration of trade-offs impractical. To overcome these limitations, we introduce Mol-MoE, a mixture-of-experts (MoE) architecture that enables efficient test-time steering of molecule generation without retraining. Central to our approach is a preference-based router training objective that incentivizes the router to combine experts in a way that aligns with user-specified trade-offs. This provides improved flexibility in exploring the chemical property space at test time, facilitating rapid trade-off exploration. Benchmarking against state-of-the-art methods, we show that Mol-MoE achieves superior sample quality and steerability.
Authors: Shengyu Duan, Rishad Shafik, Alex Yakovlev
Abstract: The Tsetlin Machine (TM) is a novel alternative to deep neural networks (DNNs). Unlike DNNs, which rely on multi-path arithmetic operations, a TM learns propositional logic patterns from data literals using Tsetlin automata. This fundamental shift from arithmetic to logic underpinning makes TM suitable for empowering new applications with low-cost implementations. In TM, literals are often included by both positive and negative clauses within the same class, canceling out their impact on individual class definitions. This property can be exploited to develop compressed TM models, enabling energy-efficient and high-throughput inferences for machine learning (ML) applications. We introduce a training approach that incorporates excluded automata states to sparsify TM logic patterns in both positive and negative clauses. This exclusion is iterative, ensuring that highly class-correlated (and therefore significant) literals are retained in the compressed inference model, ETHEREAL, to maintain strong classification accuracy. Compared to standard TMs, ETHEREAL TM models can reduce model size by up to 87.54%, with only a minor accuracy compromise. We validate the impact of this compression on eight real-world Tiny machine learning (TinyML) datasets against standard TM, equivalent Random Forest (RF) and Binarized Neural Network (BNN) on the STM32F746G-DISCO platform. Our results show that ETHEREAL TM models achieve over an order of magnitude reduction in inference time (resulting in higher throughput) and energy consumption compared to BNNs, while maintaining a significantly smaller memory footprint compared to RFs.
Authors: Jacob Fein-Ashley
Abstract: We show that the standard discrete update rule of transformer layers can be naturally interpreted as a forward Euler discretization of a continuous dynamical system. Our Transformer Flow Approximation Theorem demonstrates that, under standard Lipschitz continuity assumptions, token representations converge uniformly to the unique solution of an ODE as the number of layers grows. Moreover, if the underlying mapping satisfies a one-sided Lipschitz condition with a negative constant, the resulting dynamics are contractive, causing perturbations to decay exponentially across layers. Beyond clarifying the empirical stability and expressivity of transformer models, these insights link transformer updates to a broader iterative reasoning framework, suggesting new avenues for accelerated convergence and architectural innovations inspired by dynamical systems theory.
Authors: Sholom Schechtman, Nicolas Schreuder
Abstract: We analyze the implicit bias of constant step stochastic subgradient descent (SGD). We consider the setting of binary classification with homogeneous neural networks - a large class of deep neural networks with ReLU-type activation functions such as MLPs and CNNs without biases. We interpret the dynamics of normalized SGD iterates as an Euler-like discretization of a conservative field flow that is naturally associated to the normalized classification margin. Owing to this interpretation, we show that normalized SGD iterates converge to the set of critical points of the normalized margin at late-stage training (i.e., assuming that the data is correctly classified with positive normalized margin). Up to our knowledge, this is the first extension of the analysis of Lyu and Li (2020) on the discrete dynamics of gradient descent to the nonsmooth and stochastic setting. Our main result applies to binary classification with exponential or logistic losses. We additionally discuss extensions to more general settings.
Authors: Keigo Nogami, Tamura Hiroto, Gouhei Tanaka
Abstract: With a growing data privacy concern, federated learning has emerged as a promising framework to train machine learning models without sharing locally distributed data. In federated learning, local model training by multiple clients and model integration by a server are repeated only through model parameter sharing. Most existing federated learning methods assume training deep learning models, which are often computationally demanding. To deal with this issue, we propose federated learning methods with reservoir state analysis to seek computational efficiency and data privacy protection simultaneously. Specifically, our method relies on Mahalanobis Distance of Reservoir States (MD-RS) method targeting time series anomaly detection, which learns a distribution of reservoir states for normal inputs and detects anomalies based on a deviation from the learned distribution. Iterative updating of statistical parameters in the MD-RS enables incremental federated learning (IncFed MD-RS). We evaluate the performance of IncFed MD-RS using benchmark datasets for time series anomaly detection. The results show that IncFed MD-RS outperforms other federated learning methods with deep learning and reservoir computing models particularly when clients' data are relatively short and heterogeneous. We demonstrate that IncFed MD-RS is robust against reduced sample data compared to other methods. We also show that the computational cost of IncFed MD-RS can be reduced by subsampling from the reservoir states without performance degradation. The proposed method is beneficial especially in anomaly detection applications where computational efficiency, algorithm simplicity, and low communication cost are required.
Authors: Shizhou Xu, Thomas Strohmer
Abstract: How can we effectively remove or "unlearn" undesirable information, such as specific features or individual data points, from a learning outcome while minimizing utility loss and ensuring rigorous guarantees? We introduce a mathematical framework based on information-theoretic regularization to address both feature and data point unlearning. For feature unlearning, we derive a unified solution that simultaneously optimizes diverse learning objectives, including entropy, conditional entropy, KL-divergence, and the energy of conditional probability. For data point unlearning, we first propose a novel definition that serves as a practical condition for unlearning via retraining, is easy to verify, and aligns with the principles of differential privacy from an inference perspective. Then, we provide provable guarantees for our framework on data point unlearning. By combining flexibility in learning objectives with simplicity in regularization design, our approach is highly adaptable and practical for a wide range of machine learning and AI applications.
Authors: Janghoon Yang
Abstract: With the evolution of large language models (LLMs), there is growing interest in leveraging LLMs for time series tasks. In this paper, we explore the characteristics of LLMs for time series forecasting by considering various existing and proposed prompting techniques. Forecasting for both short and long time series was evaluated. Our findings indicate that no single prompting method is universally applicable. It was also observed that simply providing proper context information related to the time series, without additional reasoning prompts, can achieve performance comparable to the best-performing prompt for each case. From this observation, it is expected that providing proper context information can be more crucial than a prompt for specific reasoning in time series forecasting. Several weaknesses in prompting for time series forecasting were also identified. First, LLMs often fail to follow the procedures described by the prompt. Second, when reasoning steps involve simple algebraic calculations with several operands, LLMs often fail to calculate accurately. Third, LLMs sometimes misunderstand the semantics of prompts, resulting in incomplete responses.
Authors: Janghoon Yang
Abstract: While large language models have rapidly evolved towards general artificial intelligence, their versatility in analyzing time series data remains limited. To address this limitation, we propose a novel normalization technique that considers the inherent nature of tokenization. The proposed Tokenization-Optimized Normalization (TOKON) simplifies time series data by representing each element with a single token, effectively reducing the number of tokens by 2 to 3 times. Additionally, we introduce a novel prompt for time series forecasting, termed Time Series Forecasting with Care (TFSC), to further enhance forecasting performance. Experimental results demonstrate that TOKON improves root mean square error (RMSE) for multi-step forecasting by approximately 7% to 18%, depending on the dataset and prompting method. Furthermore, TFSC, when used in conjunction with TOKON, shows additional improvements in forecasting accuracy for certain datasets
Authors: Junghwan Lee, Chen Xu, Yao Xie
Abstract: Conformal prediction for time series presents two key challenges: (1) leveraging sequential correlations in features and non-conformity scores and (2) handling multi-dimensional outcomes. We propose a novel conformal prediction method to address these two key challenges by integrating Transformer and Normalizing Flow. Specifically, the Transformer encodes the historical context of time series, and normalizing flow learns the transformation from the base distribution to the distribution of non-conformity scores conditioned on the encoded historical context. This enables the construction of prediction regions by transforming samples from the base distribution using the learned conditional flow. We ensure the marginal coverage by defining the prediction regions as sets in the transformed space that correspond to a predefined probability mass in the base distribution. The model is trained end-to-end by Flow Matching, avoiding the need for computationally intensive numerical solutions of ordinary differential equations. We demonstrate that our proposed method achieves smaller prediction regions compared to the baselines while satisfying the desired coverage through comprehensive experiments using simulated and real-world time series datasets.
Authors: Tanvir Islam
Abstract: Histogram-Based Outlier Score (HBOS) is a widely used outlier or anomaly detection method known for its computational efficiency and simplicity. However, its assumption of feature independence limits its ability to detect anomalies in datasets where interactions between features are critical. In this paper, we propose the Extended Histogram-Based Outlier Score (EHBOS), which enhances HBOS by incorporating two-dimensional histograms to capture dependencies between feature pairs. This extension allows EHBOS to identify contextual and dependency-driven anomalies that HBOS fails to detect. We evaluate EHBOS on 17 benchmark datasets, demonstrating its effectiveness and robustness across diverse anomaly detection scenarios. EHBOS outperforms HBOS on several datasets, particularly those where feature interactions are critical in defining the anomaly structure, achieving notable improvements in ROC AUC. These results highlight that EHBOS can be a valuable extension to HBOS, with the ability to model complex feature dependencies. EHBOS offers a powerful new tool for anomaly detection, particularly in datasets where contextual or relational anomalies play a significant role.
Authors: Naoki Saito, David Weber
Abstract: We propose a new method to extract discriminant and explainable features from a particular machine learning model, i.e., a combination of the scattering transform and the multiclass logistic regression. Although this model is well-known for its ability to learn various signal classes with high classification rate, it remains elusive to understand why it can generate such successful classification, mainly due to the nonlinearity of the scattering transform. In order to uncover the meaning of the scattering transform coefficients selected by the multiclass logistic regression (with the Lasso penalty), we adopt zeroth-order optimization algorithms to search an input pattern that maximizes the class probability of a class of interest given the learned model. In order to do so, it turns out that imposing sparsity and smoothness of input patterns is important. We demonstrate the effectiveness of our proposed method using a couple of synthetic time-series classification problems.
Authors: Mingguo He, Yuhe Guo, Yanping Zheng, Zhewei Wei, Stephan G\"unnemann, Xiaokui Xiao
Abstract: Link prediction for directed graphs is a crucial task with diverse real-world applications. Recent advances in embedding methods and Graph Neural Networks (GNNs) have shown promising improvements. However, these methods often lack a thorough analysis of embedding expressiveness and suffer from ineffective benchmarks for a fair evaluation. In this paper, we propose a unified framework to assess the expressiveness of existing methods, highlighting the impact of dual embeddings and decoder design on performance. To address limitations in current experimental setups, we introduce DirLinkBench, a robust new benchmark with comprehensive coverage and standardized evaluation. The results show that current methods struggle to achieve strong performance on the new benchmark, while DiGAE outperforms others overall. We further revisit DiGAE theoretically, showing its graph convolution aligns with GCN on an undirected bipartite graph. Inspired by these insights, we propose a novel spectral directed graph auto-encoder SDGAE that achieves SOTA results on DirLinkBench. Finally, we analyze key factors influencing directed link prediction and highlight open challenges.
Authors: Jayden Teoh, Wenjun Li, Pradeep Varakantham
Abstract: Unsupervised Environment Design (UED) formalizes the problem of autocurricula through interactive training between a teacher agent and a student agent. The teacher generates new training environments with high learning potential, curating an adaptive curriculum that strengthens the student's ability to handle unseen scenarios. Existing UED methods mainly rely on regret, a metric that measures the difference between the agent's optimal and actual performance, to guide curriculum design. Regret-driven methods generate curricula that progressively increase environment complexity for the student but overlook environment novelty -- a critical element for enhancing an agent's generalizability. Measuring environment novelty is especially challenging due to the underspecified nature of environment parameters in UED, and existing approaches face significant limitations. To address this, this paper introduces the Coverage-based Evaluation of Novelty In Environment (CENIE) framework. CENIE proposes a scalable, domain-agnostic, and curriculum-aware approach to quantifying environment novelty by leveraging the student's state-action space coverage from previous curriculum experiences. We then propose an implementation of CENIE that models this coverage and measures environment novelty using Gaussian Mixture Models. By integrating both regret and novelty as complementary objectives for curriculum design, CENIE facilitates effective exploration across the state-action space while progressively increasing curriculum complexity. Empirical evaluations demonstrate that augmenting existing regret-based UED algorithms with CENIE achieves state-of-the-art performance across multiple benchmarks, underscoring the effectiveness of novelty-driven autocurricula for robust generalization.
Authors: Nora Agah, Meiyi Li, Javad Mohammadi
Abstract: The increased integration of clean yet stochastic energy resources and the growing number of extreme weather events are narrowing the decision-making window of power grid operators. This time constraint is fueling a plethora of research on Machine Learning-, or ML-, based optimization proxies. While finding a fast solution is appealing, the inherent vulnerabilities of the learning-based methods are hindering their adoption. One of these vulnerabilities is data poisoning attacks, which adds perturbations to ML training data, leading to incorrect decisions. The impact of poisoning attacks on learning-based power system optimizers have not been thoroughly studied, which creates a critical vulnerability. In this paper, we examine the impact of data poisoning attacks on ML-based optimization proxies that are used to solve the DC Optimal Power Flow problem. Specifically, we compare the resilience of three different methods-a penalty-based method, a post-repair approach, and a direct mapping approach-against the adverse effects of poisoning attacks. We will use the optimality and feasibility of these proxies as performance metrics. The insights of this work will establish a foundation for enhancing the resilience of neural power system optimizers.
Authors: Xiao Li, Zekai Zhang, Xiang Li, Siyi Chen, Zhihui Zhu, Peng Wang, Qing Qu
Abstract: This work addresses the critical question of why and when diffusion models, despite being designed for generative tasks, can excel at learning high-quality representations in a self-supervised manner. To address this, we develop a mathematical framework based on a low-dimensional data model and posterior estimation, revealing a fundamental trade-off between generation and representation quality near the final stage of image generation. Our analysis explains the unimodal representation dynamics across noise scales, mainly driven by the interplay between data denoising and class specification. Building on these insights, we propose an ensemble method that aggregates features across noise levels, significantly improving both clean performance and robustness under label noise. Extensive experiments on both synthetic and real-world datasets validate our findings.
Authors: Xinrui Wang, Chuanxing Geng, Wenhai Wan, Shao-yuan Li, Songcan Chen
Abstract: Recent studies have verified that semi-supervised learning (SSL) is vulnerable to data poisoning backdoor attacks. Even a tiny fraction of contaminated training data is sufficient for adversaries to manipulate up to 90\% of the test outputs in existing SSL methods. Given the emerging threat of backdoor attacks designed for SSL, this work aims to protect SSL against such risks, marking it as one of the few known efforts in this area. Specifically, we begin by identifying that the spurious correlations between the backdoor triggers and the target class implanted by adversaries are the primary cause of manipulated model predictions during the test phase. To disrupt these correlations, we utilize three key techniques: Gaussian Filter, complementary learning and trigger mix-up, which collectively filter, obstruct and dilute the influence of backdoor attacks in both data pre-processing and feature learning. Experimental results demonstrate that our proposed method, Backdoor Invalidator (BI), significantly reduces the average attack success rate from 84.7\% to 1.8\% across different state-of-the-art backdoor attacks. It is also worth mentioning that BI does not sacrifice accuracy on clean data and is supported by a theoretical guarantee of its generalization capability.
Authors: Keren Fuentes, Mimee Xu, Irene Chen
Abstract: Access to diverse, high-quality datasets is crucial for machine learning model performance, yet data sharing remains limited by privacy concerns and competitive interests, particularly in regulated domains like healthcare. This dynamic especially disadvantages smaller organizations that lack resources to purchase data or negotiate favorable sharing agreements. We present SecureKL, a privacy-preserving framework that enables organizations to identify beneficial data partnerships without exposing sensitive information. Building on recent advances in dataset combination methods, we develop a secure multiparty computation protocol that maintains strong privacy guarantees while achieving >90\% correlation with plaintext evaluations. In experiments with real-world hospital data, SecureKL successfully identifies beneficial data partnerships that improve model performance for intensive care unit mortality prediction while preserving data privacy. Our framework provides a practical solution for organizations seeking to leverage collective data resources while maintaining privacy and competitive advantages. These results demonstrate the potential for privacy-preserving data collaboration to advance machine learning applications in high-stakes domains while promoting more equitable access to data resources.
Authors: Junbo Li, Zhangyang Wang, Qiang Liu
Abstract: Offline preference alignment for language models such as Direct Preference Optimization (DPO) is favored for its effectiveness and simplicity, eliminating the need for costly reinforcement learning. Various offline algorithms have been developed for different data settings, yet they lack a unified understanding. In this study, we introduce Pior-Informed Preference Alignment (PIPA), a unified, RL-free probabilistic framework that formulates language model preference alignment as a Maximum Likelihood Estimation (MLE) problem with prior constraints. This method effectively accommodates both paired and unpaired data, as well as answer and step-level annotations. We illustrate that DPO and KTO are special cases with different prior constraints within our framework. By integrating different types of prior information, we developed two variations of PIPA: PIPA-M and PIPA-N. Both algorithms demonstrate a $3\sim10\%$ performance enhancement on the GSM8K and MATH benchmarks across all configurations, achieving these gains without additional training or computational costs compared to existing algorithms.
Authors: Karthik Sivakoti
Abstract: Traditional automated crash analysis systems heavily rely on static statistical models and historical data, requiring significant manual interpretation and lacking real-time predictive capabilities. This research presents an innovative approach to traffic safety analysis through the integration of ensemble learning methods and multi-modal data fusion for real-time crash risk assessment and prediction. Our primary contribution lies in developing a hierarchical severity classification system that combines spatial-temporal crash patterns with environmental conditions, achieving significant improvements over traditional statistical approaches. The system demonstrates a Mean Average Precision (mAP) of 0.893, representing a 15% improvement over current state-of-the-art methods (baseline mAP: 0.776). We introduce a novel feature engineering technique that integrates crash location data with incident reports and weather conditions, achieving 92.4% accuracy in risk prediction and 89.7% precision in hotspot identification. Through extensive validation using 500,000 initial crash records filtered to 59,496 high-quality samples, our solution shows marked improvements in both prediction accuracy and computational efficiency. Key innovations include a robust data cleaning pipeline, adaptive feature generation, and a scalable real-time prediction system capable of handling peak loads of 1,000 concurrent requests while maintaining sub-100ms response times.
Authors: Danny Wang, Ruihong Qiu, Guangdong Bai, Zi Huang
Abstract: Despite graph neural networks' (GNNs) great success in modelling graph-structured data, out-of-distribution (OOD) test instances still pose a great challenge for current GNNs. One of the most effective techniques to detect OOD nodes is to expose the detector model with an additional OOD node-set, yet the extra OOD instances are often difficult to obtain in practice. Recent methods for image data address this problem using OOD data synthesis, typically relying on pre-trained generative models like Stable Diffusion. However, these approaches require vast amounts of additional data, as well as one-for-all pre-trained generative models, which are not available for graph data. Therefore, we propose the GOLD framework for graph OOD detection, an implicit adversarial learning pipeline with synthetic OOD exposure without pre-trained models. The implicit adversarial training process employs a novel alternating optimisation framework by training: (1) a latent generative model to regularly imitate the in-distribution (ID) embeddings from an evolving GNN, and (2) a GNN encoder and an OOD detector to accurately classify ID data while increasing the energy divergence between the ID embeddings and the generative model's synthetic embeddings. This novel approach implicitly transforms the synthetic embeddings into pseudo-OOD instances relative to the ID data, effectively simulating exposure to OOD scenarios without auxiliary data. Extensive OOD detection experiments are conducted on five benchmark graph datasets, verifying the superior performance of GOLD without using real OOD data compared with the state-of-the-art OOD exposure and non-exposure baselines.
Authors: Haochen Zhang, Junze Yin, Guanchu Wang, Zirui Liu, Tianyi Zhang, Anshumali Shrivastava, Lin Yang, Vladimir Braverman
Abstract: Low-rank optimization has emerged as a promising approach to enabling memory-efficient training of large language models (LLMs). Existing low-rank optimization methods typically project gradients onto a low-rank subspace, reducing the memory cost of storing optimizer states. A key challenge in these methods is identifying suitable subspaces to ensure an effective optimization trajectory. Most existing approaches select the dominant subspace to preserve gradient information, as this intuitively provides the best approximation. However, we find that in practice, the dominant subspace stops changing during pretraining, thereby constraining weight updates to similar subspaces. In this paper, we propose importance sampling subspace selection (I3S) for low-rank optimization, which theoretically offers a comparable convergence rate to the dominant subspace approach. Empirically, we demonstrate that I3S significantly outperforms previous methods in LLM pretraining tasks.
Authors: Wenfang Sun, Xinyuan Song, Pengxiang Li, Lu Yin, Yefeng Zheng, Shiwei Liu
Abstract: In this paper, we introduce the Curse of Depth, a concept that highlights, explains, and addresses the recent observation in modern Large Language Models(LLMs) where nearly half of the layers are less effective than expected. We first confirm the wide existence of this phenomenon across the most popular families of LLMs such as Llama, Mistral, DeepSeek, and Qwen. Our analysis, theoretically and empirically, identifies that the underlying reason for the ineffectiveness of deep layers in LLMs is the widespread usage of Pre-Layer Normalization (Pre-LN). While Pre-LN stabilizes the training of Transformer LLMs, its output variance exponentially grows with the model depth, which undesirably causes the derivative of the deep Transformer blocks to be an identity matrix, and therefore barely contributes to the training. To resolve this training pitfall, we propose LayerNorm Scaling, which scales the variance of output of the layer normalization inversely by the square root of its depth. This simple modification mitigates the output variance explosion of deeper Transformer layers, improving their contribution. Our experimental results, spanning model sizes from 130M to 1B, demonstrate that LayerNorm Scaling significantly enhances LLM pre-training performance compared to Pre-LN. Moreover, this improvement seamlessly carries over to supervised fine-tuning. All these gains can be attributed to the fact that LayerNorm Scaling enables deeper layers to contribute more effectively during training.
Authors: Rafa{\l} Karczewski, Markus Heinonen, Vikas Garg
Abstract: Diffusion models have emerged as a powerful class of generative models, capable of producing high-quality images by mapping noise to a data distribution. However, recent findings suggest that image likelihood does not align with perceptual quality: high-likelihood samples tend to be smooth, while lower-likelihood ones are more detailed. Controlling sample density is thus crucial for balancing realism and detail. In this paper, we analyze an existing technique, Prior Guidance, which scales the latent code to influence image detail. We introduce score alignment, a condition that explains why this method works and show that it can be tractably checked for any continuous normalizing flow model. We then propose Density Guidance, a principled modification of the generative ODE that enables exact log-density control during sampling. Finally, we extend Density Guidance to stochastic sampling, ensuring precise log-density control while allowing controlled variation in structure or fine details. Our experiments demonstrate that these techniques provide fine-grained control over image detail without compromising sample quality.
Authors: Tian-Shuang Wu, Shen-Huan Lyu, Ning Chen, Zhihao Qu, Baoliu Ye
Abstract: In recent years, as a compromise between privacy and performance, few-sample model compression has been widely adopted to deal with limited data resulting from privacy and security concerns. However, when the number of available samples is extremely limited, class imbalance becomes a common and tricky problem. Achieving an equal number of samples across all classes is often costly and impractical in real-world applications, and previous studies on few-sample model compression have mostly ignored this significant issue. Our experiments comprehensively demonstrate that class imbalance negatively affects the overall performance of few-sample model compression methods. To address this problem, we propose a novel and adaptive framework named OOD-Enhanced Few-Sample Model Compression (OE-FSMC). This framework integrates easily accessible out-of-distribution (OOD) data into both the compression and fine-tuning processes, effectively rebalancing the training distribution. We also incorporate a joint distillation loss and a regularization term to reduce the risk of the model overfitting to the OOD data. Extensive experiments on multiple benchmark datasets show that our framework can be seamlessly incorporated into existing few-sample model compression methods, effectively mitigating the accuracy degradation caused by class imbalance.
Authors: Yunhui Liu, Zhen Tao, Xiang Zhao, Jianhua Zhao, Tao Zheng, Tieke He
Abstract: Multiplex graphs, with multiple edge types (graph views) among common nodes, provide richer structural semantics and better modeling capabilities. Multiplex Graph Neural Networks (MGNNs), typically comprising view-specific GNNs and a multi-view integration layer, have achieved advanced performance in various downstream tasks. However, their reliance on neighborhood aggregation poses challenges for deployment in latency-sensitive applications. Motivated by recent GNN-to-MLP knowledge distillation frameworks, we propose Multiplex Graph-Free Neural Networks (MGFNN and MGFNN+) to combine MGNNs' superior performance and MLPs' efficient inference via knowledge distillation. MGFNN directly trains student MLPs with node features as input and soft labels from teacher MGNNs as targets. MGFNN+ further employs a low-rank approximation-based reparameterization to learn node-wise coefficients, enabling adaptive knowledge ensemble from each view-specific GNN. This node-wise multi-view ensemble distillation strategy allows student MLPs to learn more informative multiplex semantic knowledge for different nodes. Experiments show that MGFNNs achieve average accuracy improvements of about 10% over vanilla MLPs and perform comparably or even better to teacher MGNNs (accurate); MGFNNs achieve a 35.40$\times$-89.14$\times$ speedup in inference over MGNNs (efficient); MGFNN+ adaptively assigns different coefficients for multi-view ensemble distillation regarding different nodes (interpretable).
Authors: Yunhui Liu, Huaisong Zhang, Xinyi Gao, Liuye Guo, Zhen Tao, Tieke He
Abstract: Link Prediction (LP) is a crucial problem in graph-structured data. Graph Neural Networks (GNNs) have gained prominence in LP, with Graph AutoEncoders (GAEs) being a notable representation. However, our empirical findings reveal that GAEs' LP performance suffers heavily from the long-tailed node degree distribution, i.e., low-degree nodes tend to exhibit inferior LP performance compared to high-degree nodes. \emph{What causes this degree-related bias, and how can it be mitigated?} In this study, we demonstrate that the norm of node embeddings learned by GAEs exhibits variation among nodes with different degrees, underscoring its central significance in influencing the final performance of LP. Specifically, embeddings with larger norms tend to guide the decoder towards predicting higher scores for positive links and lower scores for negative links, thereby contributing to superior performance. This observation motivates us to improve GAEs' LP performance on low-degree nodes by increasing their embedding norms, which can be implemented simply yet effectively by introducing additional self-loops into the training objective for low-degree nodes. This norm augmentation strategy can be seamlessly integrated into existing GAE methods with light computational cost. Extensive experiments on various datasets and GAE methods show the superior performance of norm-augmented GAEs.
Authors: Abdelwahed Khamis, Sara Khalifa
Abstract: Real-world sensing challenges such as sensor failures, communication issues, and power constraints lead to data intermittency. An issue that is known to undermine the traditional classification task that assumes a continuous data stream. Previous works addressed this issue by designing bespoke solutions (i.e. task-specific and/or modality-specific imputation). These approaches, while effective for their intended purposes, had limitations in their applicability across different tasks and sensor modalities. This raises an important question: Can we build a task-agnostic imputation pipeline that is transferable to new sensors without requiring additional training? In this work, we formalise the concept of zero-shot imputation and propose a novel approach that enables the adaptation of pre-trained models to handle data intermittency. This framework, named NeuralPrefix, is a generative neural component that precedes a task model during inference, filling in gaps caused by data intermittency. NeuralPrefix is built as a continuous dynamical system, where its internal state can be estimated at any point in time by solving an Ordinary Differential Equation (ODE). This approach allows for a more versatile and adaptable imputation method, overcoming the limitations of task-specific and modality-specific solutions. We conduct a comprehensive evaluation of NeuralPrefix on multiple sensory datasets, demonstrating its effectiveness across various domains. When tested on intermittent data with a high 50% missing data rate, NeuralPreifx accurately recovers all the missing samples, achieving SSIM score between 0.93-0.96. Zero-shot evaluations show that NeuralPrefix generalises well to unseen datasets, even when the measurements come from a different modality.
Authors: Aymene Berriche, Mehdi Zakaria Adjal, Riyadh Baghdadi
Abstract: Backpropagation (BP) has long been the predominant method for training neural networks due to its effectiveness. However, numerous alternative approaches, broadly categorized under feedback alignment, have been proposed, many of which are motivated by the search for biologically plausible learning mechanisms. Despite their theoretical appeal, these methods have consistently underperformed compared to BP, leading to a decline in research interest. In this work, we revisit the role of such methods and explore how they can be integrated into standard neural network training pipelines. Specifically, we propose fine-tuning BP-pre-trained models using Sign-Symmetry learning rules and demonstrate that this approach not only maintains performance parity with BP but also enhances robustness. Through extensive experiments across multiple tasks and benchmarks, we establish the validity of our approach. Our findings introduce a novel perspective on neural network training and open new research directions for leveraging biologically inspired learning rules in deep learning.
Authors: Ahmed Abdelaziz, Ahmed Fathi, Ahmed Fares
Abstract: EEG-based neural networks, pivotal in medical diagnosis and brain-computer interfaces, face significant intellectual property (IP) risks due to their reliance on sensitive neurophysiological data and resource-intensive development. Current watermarking methods, particularly those using abstract trigger sets, lack robust authentication and fail to address the unique challenges of EEG models. This paper introduces a cryptographic wonder filter-based watermarking framework tailored for EEG-based neural networks. Leveraging collision-resistant hashing and public-key encryption, the wonder filter embeds the watermark during training, ensuring minimal distortion ($\leq 5\%$ drop in EEG task accuracy) and high reliability (100\% watermark detection). The framework is rigorously evaluated against adversarial attacks, including fine-tuning, transfer learning, and neuron pruning. Results demonstrate persistent watermark retention, with classification accuracy for watermarked states remaining above 90\% even after aggressive pruning, while primary task performance degrades faster, deterring removal attempts. Piracy resistance is validated by the inability to embed secondary watermarks without severe accuracy loss ( $>10\%$ in EEGNet and CCNN models). Cryptographic hashing ensures authentication, reducing brute-force attack success probabilities. Evaluated on the DEAP dataset across models (CCNN, EEGNet, TSception), the method achieves $>99.4\%$ null-embedding accuracy, effectively eliminating false positives. By integrating wonder filters with EEG-specific adaptations, this work bridges a critical gap in IP protection for neurophysiological models, offering a secure, tamper-proof solution for healthcare and biometric applications. The framework's robustness against adversarial modifications underscores its potential to safeguard sensitive EEG models while maintaining diagnostic utility.
Authors: Tenglong Liu, Jianxiong Li, Yinan Zheng, Haoyi Niu, Yixing Lan, Xin Xu, Xianyuan Zhan
Abstract: Humans excel at reusing prior knowledge to address new challenges and developing skills while solving problems. This paradigm becomes increasingly popular in the development of autonomous agents, as it develops systems that can self-evolve in response to new challenges like human beings. However, previous methods suffer from limited training efficiency when expanding new skills and fail to fully leverage prior knowledge to facilitate new task learning. In this paper, we propose Parametric Skill Expansion and Composition (PSEC), a new framework designed to iteratively evolve the agents' capabilities and efficiently address new challenges by maintaining a manageable skill library. This library can progressively integrate skill primitives as plug-and-play Low-Rank Adaptation (LoRA) modules in parameter-efficient finetuning, facilitating efficient and flexible skill expansion. This structure also enables the direct skill compositions in parameter space by merging LoRA modules that encode different skills, leveraging shared information across skills to effectively program new skills. Based on this, we propose a context-aware module to dynamically activate different skills to collaboratively handle new tasks. Empowering diverse applications including multi-objective composition, dynamics shift, and continual policy shift, the results on D4RL, DSRL benchmarks, and the DeepMind Control Suite show that PSEC exhibits superior capacity to leverage prior knowledge to efficiently tackle new challenges, as well as expand its skill libraries to evolve the capabilities. Project website: https://ltlhuuu.github.io/PSEC/.
Authors: Stanislav R. Kirpichenko, Lev V. Utkin, Andrei V. Konstantinov, Natalya M. Verbova
Abstract: Concept-based learning enhances prediction accuracy and interpretability by leveraging high-level, human-understandable concepts. However, existing CBL frameworks do not address survival analysis tasks, which involve predicting event times in the presence of censored data -- a common scenario in fields like medicine and reliability analysis. To bridge this gap, we propose two novel models: SurvCBM (Survival Concept-based Bottleneck Model) and SurvRCM (Survival Regularized Concept-based Model), which integrate concept-based learning with survival analysis to handle censored event time data. The models employ the Cox proportional hazards model and the Beran estimator. SurvCBM is based on the architecture of the well-known concept bottleneck model, offering interpretable predictions through concept-based explanations. SurvRCM uses concepts as regularization to enhance accuracy. Both models are trained end-to-end and provide interpretable predictions in terms of concepts. Two interpretability approaches are proposed: one leveraging the linear relationship in the Cox model and another using an instance-based explanation framework with the Beran estimator. Numerical experiments demonstrate that SurvCBM outperforms SurvRCM and traditional survival models, underscoring the importance and advantages of incorporating concept information. The code for the proposed algorithms is publicly available.
Authors: Sharmita Dey
Abstract: Recent advances in large-scale machine learning have produced high-capacity foundation models capable of adapting to a broad array of downstream tasks. While such models hold great promise for robotics, the prevailing paradigm still portrays robots as single, autonomous decision-makers, performing tasks like manipulation and navigation, with limited human involvement. However, a large class of real-world robotic systems, including wearable robotics (e.g., prostheses, orthoses, exoskeletons), teleoperation, and neural interfaces, are semiautonomous, and require ongoing interactive coordination with human partners, challenging single-agent assumptions. In this position paper, we argue that robot foundation models must evolve to an interactive multi-agent perspective in order to handle the complexities of real-time human-robot co-adaptation. We propose a generalizable, neuroscience-inspired architecture encompassing four modules: (1) a multimodal sensing module informed by sensorimotor integration principles, (2) an ad-hoc teamwork model reminiscent of joint-action frameworks in cognitive science, (3) a predictive world belief model grounded in internal model theories of motor control, and (4) a memory/feedback mechanism that echoes concepts of Hebbian and reinforcement-based plasticity. Although illustrated through the lens of cyborg systems, where wearable devices and human physiology are inseparably intertwined, the proposed framework is broadly applicable to robots operating in semi-autonomous or interactive contexts. By moving beyond single-agent designs, our position emphasizes how foundation models in robotics can achieve a more robust, personalized, and anticipatory level of performance.
Authors: Saaketh Narayan, Abhay Gupta, Mansheej Paul, Davis Blalock
Abstract: Large Language Model training with 8-bit floating point (FP8) formats promises significant efficiency improvements, but reduced numerical precision makes training challenging. It is currently possible to train in FP8 only if one is willing to tune various hyperparameters, reduce model scale, or accept the overhead of computing dynamic scale factors. We demonstrate simple, scalable FP8 training that requires no dynamic scaling factors or special hyperparameters, even at large model sizes. Our method, $\mu$nit Scaling ($\mu$S), also enables simple hyperparameter transfer across model widths, matched numerics across training and inference, and other desirable properties. $\mu$nit Scaling is straightforward to implement, consisting of a set of minimal interventions based on a first-principles analysis of common transformer operations. We validate our method by training models from 1B to 13B parameters, performing all hidden linear layer computations in FP8. We achieve quality equal to higher precision baselines while also training up to 33% faster.
Authors: Nofit Segal, Aviv Netanyahu, Kevin P. Greenman, Pulkit Agrawal, Rafael Gomez-Bombarelli
Abstract: Discovery of high-performance materials and molecules requires identifying extremes with property values that fall outside the known distribution. Therefore, the ability to extrapolate to out-of-distribution (OOD) property values is critical for both solid-state materials and molecular design. Our objective is to train predictor models that extrapolate zero-shot to higher ranges than in the training data, given the chemical compositions of solids or molecular graphs and their property values. We propose using a transductive approach to OOD property prediction, achieving improvements in prediction accuracy. In particular, the True Positive Rate (TPR) of OOD classification of materials and molecules improved by 3x and 2.5x, respectively, and precision improved by 2x and 1.5x compared to non-transductive baselines. Our method leverages analogical input-target relations in the training and test sets, enabling generalization beyond the training target support, and can be applied to any other material and molecular tasks.
Authors: Haolin Liu, Chen-Yu Wei, Julian Zimmert
Abstract: Recent work by Foster et al. (2021, 2022, 2023) and Xu and Zeevi (2023) developed the framework of decision estimation coefficient (DEC) that characterizes the complexity of general online decision making problems and provides a general algorithm design principle. These works, however, either focus on the pure stochastic regime where the world remains fixed over time, or the pure adversarial regime where the world arbitrarily changes over time. For the hybrid regime where the dynamics of the world is fixed while the reward arbitrarily changes, they only give pessimistic bounds on the decision complexity. In this work, we propose a general extension of DEC that more precisely characterizes this case. Besides applications in special cases, our framework leads to a flexible algorithm design where the learner learns over subsets of the hypothesis set, trading estimation complexity with decision complexity, which could be of independent interest. Our work covers model-based learning and model-free learning in the hybrid regime, with a newly proposed extension of the bilinear classes (Du et al., 2021) to the adversarial-reward case. We also recover some existing model-free learning results in the pure stochastic regime.
Authors: Jusheng Zhang, Yijia Fan, Kaitong Cai, Keze Wang
Abstract: Although Kolmogorov-Arnold based interpretable networks (KAN) have strong theoretical expressiveness, they face significant parameter explosion and high-frequency feature capture challenges in high-dimensional tasks. To address this issue, we propose the Kolmogorov-Arnold-Fourier Network (KAF), which effectively integrates trainable Random Fourier Features (RFF) and a novel hybrid GELU-Fourier activation mechanism to balance parameter efficiency and spectral representation capabilities. Our key technical contributions include: (1) merging KAN's dual-matrix structure through matrix association properties to substantially reduce parameters; (2) introducing learnable RFF initialization strategies to eliminate spectral distortion in high-dimensional approximation tasks; (3) implementing an adaptive hybrid activation function that progressively enhances frequency representation during the training process. Comprehensive experiments demonstrate the superiority of our KAF across various domains including vision, NLP, audio processing, and differential equation-solving tasks, effectively combining theoretical interpretability with practical utility and computational efficiency.
Authors: Elisa Negrini, Yuxuan Liu, Liu Yang, Stanley J. Osher, Hayden Schaeffer
Abstract: Neural networks are one tool for approximating non-linear differential equations used in scientific computing tasks such as surrogate modeling, real-time predictions, and optimal control. PDE foundation models utilize neural networks to train approximations to multiple differential equations simultaneously and are thus a general purpose solver that can be adapted to downstream tasks. Current PDE foundation models focus on either learning general solution operators and/or the governing system of equations, and thus only handle numerical or symbolic modalities. However, real-world applications may require more flexible data modalities, e.g. text analysis or descriptive outputs. To address this gap, we propose a novel multimodal deep learning approach that leverages a transformer-based architecture to approximate solution operators for a wide variety of ODEs and PDEs. Our method integrates numerical inputs, such as equation parameters and initial conditions, with text descriptions of physical processes or system dynamics. This enables our model to handle settings where symbolic representations may be incomplete or unavailable. In addition to providing accurate numerical predictions, our approach generates interpretable scientific text descriptions, offering deeper insights into the underlying dynamics and solution properties. The numerical experiments show that our model provides accurate solutions for in-distribution data (with average relative error less than 3.3%) and out-of-distribution data (average relative error less than 7.8%) together with precise text descriptions (with correct descriptions generated 100% of times). In certain tests, the model is also shown to be capable of extrapolating solutions in time.
Authors: Ziqi Chen, Bo Peng, Tianhua Zhai, Daniel Adu-Ampratwum, Xia Ning
Abstract: Drug development is a critical but notoriously resource- and time-consuming process. In this manuscript, we develop a novel generative artificial intelligence (genAI) method DiffSMol to facilitate drug development. DiffSmol generates 3D binding molecules based on the shapes of known ligands. DiffSMol encapsulates geometric details of ligand shapes within pre-trained, expressive shape embeddings and then generates new binding molecules through a diffusion model. DiffSMol further modifies the generated 3D structures iteratively via shape guidance to better resemble the ligand shapes. It also tailors the generated molecules toward optimal binding affinities under the guidance of protein pockets. Here, we show that DiffSMol outperforms the state-of-the-art methods on benchmark datasets. When generating binding molecules resembling ligand shapes, DiffSMol with shape guidance achieves a success rate 61.4%, substantially outperforming the best baseline (11.2%), meanwhile producing molecules with novel molecular graph structures. DiffSMol with pocket guidance also outperforms the best baseline in binding affinities by 13.2%, and even by 17.7% when combined with shape guidance. Case studies for two critical drug targets demonstrate very favorable physicochemical and pharmacokinetic properties of the generated molecules, thus, the potential of DiffSMol in developing promising drug candidates.
Authors: Willa Potosnak, Cristian Challu, Mononito Goswami, Kin G. Olivares, Micha{\l} Wili\'nski, Nina \.Zukowska, Artur Dubrawski
Abstract: Large pre-trained time series foundation models (TSFMs) have demonstrated promising zero-shot performance across a wide range of domains. However, a question remains: Do TSFMs succeed solely by memorizing training patterns, or do they possess the ability to reason? While reasoning is a topic of great interest in the study of Large Language Models (LLMs), it is undefined and largely unexplored in the context of TSFMs. In this work, inspired by language modeling literature, we formally define compositional reasoning in forecasting and distinguish it from in-distribution generalization. We evaluate the reasoning and generalization capabilities of 23 popular deep learning forecasting models on multiple synthetic and real-world datasets. Additionally, through controlled studies, we systematically examine which design choices in TSFMs contribute to improved reasoning abilities. Our study yields key insights into the impact of TSFM architecture design on compositional reasoning and generalization. We find that patch-based Transformers have the best reasoning performance, closely followed by residualized MLP-based architectures, which are 97\% less computationally complex in terms of FLOPs and 86\% smaller in terms of the number of trainable parameters. Interestingly, in some zero-shot out-of-distribution scenarios, these models can outperform moving average and exponential smoothing statistical baselines trained on in-distribution data. Only a few design choices, such as the tokenization method, had a significant (negative) impact on Transformer model performance.
Authors: Lev Stambler, Seyed Sajjad Nezhadi, Matthew Coudron
Abstract: We develop an algorithm which, given a trained transformer model $\mathcal{M}$ as input, as well as a string of tokens $s$ of length $n_{fix}$ and an integer $n_{free}$, can generate a mathematical proof that $\mathcal{M}$ is ``overwhelmed'' by $s$, in time and space $\widetilde{O}(n_{fix}^2 + n_{free}^3)$. We say that $\mathcal{M}$ is ``overwhelmed'' by $s$ when the output of the model evaluated on this string plus any additional string $t$, $\mathcal{M}(s + t)$, is completely insensitive to the value of the string $t$ whenever length($t$) $\leq n_{free}$. Along the way, we prove a particularly strong worst-case form of ``over-squashing'', which we use to bound the model's behavior. Our technique uses computer-aided proofs to establish this type of operationally relevant guarantee about transformer models. We empirically test our algorithm on a single layer transformer complete with an attention head, layer-norm, MLP/ReLU layers, and RoPE positional encoding. We believe that this work is a stepping stone towards the difficult task of obtaining useful guarantees for trained transformer models.
Authors: Louis Bethune, David Grangier, Dan Busbridge, Eleonora Gualdoni, Marco Cuturi, Pierre Ablin
Abstract: A widespread strategy to obtain a language model that performs well on a target domain is to finetune a pretrained model to perform unsupervised next-token prediction on data from that target domain. Finetuning presents two challenges: (i) if the amount of target data is limited, as in most practical applications, the model will quickly overfit, and (ii) the model will drift away from the original model, forgetting the pretraining data and the generic knowledge that comes with it. We aim to derive scaling laws that quantify these two phenomena for various target domains, amounts of available target data, and model scales. We measure the efficiency of injecting pretraining data into the finetuning data mixture to avoid forgetting and mitigate overfitting. A key practical takeaway from our study is that injecting as little as 1% of pretraining data in the finetuning data mixture prevents the model from forgetting the pretraining set.
Authors: Yesom Park, Imseong Park, Jooyoung Hahn, Myungjoo Kang
Abstract: In this paper, we propose the neural shortest path (NSP), a vector-valued implicit neural representation (INR) that approximates a distance function and its gradient. The key feature of NSP is to learn the exact shortest path (ESP), which directs an arbitrary point to its nearest point on the target surface. The NSP is decomposed into its magnitude and direction, and a variable splitting method is used that each decomposed component approximates a distance function and its gradient, respectively. Unlike to existing methods of learning the distance function itself, the NSP ensures the simultaneous recovery of the distance function and its gradient. We mathematically prove that the decomposed representation of NSP guarantees the convergence of the magnitude of NSP in the $H^1$ norm. Furthermore, we devise a novel loss function that enforces the property of ESP, demonstrating that its global minimum is the ESP. We evaluate the performance of the NSP through comprehensive experiments on diverse datasets, validating its capacity to reconstruct high-quality surfaces with the robustness to noise and data sparsity. The numerical results show substantial improvements over state-of-the-art methods, highlighting the importance of learning the ESP, the product of distance function and its gradient, for representing a wide variety of complex surfaces.
Authors: Qingyue Zhao, Kaixuan Ji, Heyang Zhao, Tong Zhang, Quanquan Gu
Abstract: KL-regularized policy optimization has become a workhorse in learning-based decision making, while its theoretical understanding is still very limited. Although recent progress has been made towards settling the sample complexity of KL-regularized contextual bandits, existing sample complexity bounds are either $\tilde{O}(\epsilon^{-2})$ under single-policy concentrability or $\tilde{O}(\epsilon^{-1})$ under all-policy concentrability. In this paper, we propose the \emph{first} algorithm with $\tilde{O}(\epsilon^{-1})$ sample complexity under single-policy concentrability for offline contextual bandits. Our algorithm is designed for general function approximation and based on the principle of \emph{pessimism in the face of uncertainty}. The core of our proof leverages the strong convexity of the KL regularization, and the conditional non-negativity of the gap between the true reward and its pessimistic estimator to refine a mean-value-type risk upper bound to its extreme. This in turn leads to a novel covariance-based analysis, effectively bypassing the need for uniform control over the discrepancy between any two functions in the function class. The near-optimality of our algorithm is demonstrated by an $\tilde{\Omega}(\epsilon^{-1})$ lower bound. Furthermore, we extend our algorithm to contextual dueling bandits and achieve a similar nearly optimal sample complexity.
Authors: Jiajun Fan, Shuaike Shen, Chaoran Cheng, Yuxin Chen, Chumeng Liang, Ge Liu
Abstract: Recent advancements in reinforcement learning (RL) have achieved great success in fine-tuning diffusion-based generative models. However, fine-tuning continuous flow-based generative models to align with arbitrary user-defined reward functions remains challenging, particularly due to issues such as policy collapse from overoptimization and the prohibitively high computational cost of likelihoods in continuous-time flows. In this paper, we propose an easy-to-use and theoretically sound RL fine-tuning method, which we term Online Reward-Weighted Conditional Flow Matching with Wasserstein-2 Regularization (ORW-CFM-W2). Our method integrates RL into the flow matching framework to fine-tune generative models with arbitrary reward functions, without relying on gradients of rewards or filtered datasets. By introducing an online reward-weighting mechanism, our approach guides the model to prioritize high-reward regions in the data manifold. To prevent policy collapse and maintain diversity, we incorporate Wasserstein-2 (W2) distance regularization into our method and derive a tractable upper bound for it in flow matching, effectively balancing exploration and exploitation of policy optimization. We provide theoretical analyses to demonstrate the convergence properties and induced data distributions of our method, establishing connections with traditional RL algorithms featuring Kullback-Leibler (KL) regularization and offering a more comprehensive understanding of the underlying mechanisms and learning behavior of our approach. Extensive experiments on tasks including target image generation, image compression, and text-image alignment demonstrate the effectiveness of our method, where our method achieves optimal policy convergence while allowing controllable trade-offs between reward maximization and diversity preservation.
Authors: Xiangcheng Zhang, Yige Hong, Weina Wang
Abstract: Heterogeneity poses a fundamental challenge for many real-world large-scale decision-making problems but remains largely understudied. In this paper, we study the fully heterogeneous setting of a prominent class of such problems, known as weakly-coupled Markov decision processes (WCMDPs). Each WCMDP consists of $N$ arms (or subproblems), which have distinct model parameters in the fully heterogeneous setting, leading to the curse of dimensionality when $N$ is large. We show that, under mild assumptions, a natural adaptation of the ID policy, although originally proposed for a homogeneous special case of WCMDPs, in fact achieves an $O(1/\sqrt{N})$ optimality gap in long-run average reward per arm for fully heterogeneous WCMDPs as $N$ becomes large. This is the first asymptotic optimality result for fully heterogeneous average-reward WCMDPs. Our techniques highlight the construction of a novel projection-based Lyapunov function, which witnesses the convergence of rewards and costs to an optimal region in the presence of heterogeneity.
Authors: Daksh Mittal, Yuanzhe Ma, Shalmali Joshi, Hongseok Namkoong
Abstract: Ground truth labels/outcomes are critical for advancing scientific and engineering applications, e.g., evaluating the treatment effect of an intervention or performance of a predictive model. Since randomly sampling inputs for labeling can be prohibitively expensive, we introduce an adaptive labeling framework where measurement effort can be reallocated in batches. We formulate this problem as a Markov decision process where posterior beliefs evolve over time as batches of labels are collected (state transition), and batches (actions) are chosen to minimize uncertainty at the end of data collection. We design a computational framework that is agnostic to different uncertainty quantification approaches including those based on deep learning, and allows a diverse array of policy gradient approaches by relying on continuous policy parameterizations. On real and synthetic datasets, we demonstrate even a one-step lookahead policy can substantially outperform common adaptive labeling heuristics, highlighting the virtue of planning. On the methodological side, we note that standard REINFORCE-style policy gradient estimators can suffer high variance since they rely only on zeroth order information. We propose a direct backpropagation-based approach, Smoothed-Autodiff, based on a carefully smoothed version of the original non-differentiable MDP. Our method enjoys low variance at the price of introducing bias, and we theoretically and empirically show that this trade-off can be favorable.
Authors: Cheuk Kit Lee, Paul Jeha, Jes Frellsen, Pietro Lio, Michael Samuel Albergo, Francisco Vargas
Abstract: Discrete diffusion models are a class of generative models that produce samples from an approximated data distribution within a discrete state space. Often, there is a need to target specific regions of the data distribution. Current guidance methods aim to sample from a distribution with mass proportional to $p_0(x_0) p(\zeta|x_0)^\alpha$ but fail to achieve this in practice. We introduce a Sequential Monte Carlo algorithm that generates unbiasedly from this target distribution, utilising the learnt unconditional and guided process. We validate our approach on low-dimensional distributions, controlled images and text generations. For text generation, our method provides strong control while maintaining low perplexity compared to guidance-based approaches.
Authors: Runlong Yu, Chonghao Qiu, Robert Ladwig, Paul Hanson, Yiqun Xie, Xiaowei Jia
Abstract: Physics-guided machine learning (PGML) has become a prevalent approach in studying scientific systems due to its ability to integrate scientific theories for enhancing machine learning (ML) models. However, most PGML approaches are tailored to isolated and relatively simple tasks, which limits their applicability to complex systems involving multiple interacting processes and numerous influencing features. In this paper, we propose a \textit{\textbf{P}hysics-\textbf{G}uided \textbf{F}oundation \textbf{M}odel (\textbf{PGFM})} that combines pre-trained ML models and physics-based models and leverages their complementary strengths to improve the modeling of multiple coupled processes. To effectively conduct pre-training, we construct a simulated environmental system that encompasses a wide range of influencing features and various simulated variables generated by physics-based models. The model is pre-trained in this system to adaptively select important feature interactions guided by multi-task objectives. We then fine-tune the model for each specific task using true observations, while maintaining consistency with established physical theories, such as the principles of mass and energy conservation. We demonstrate the effectiveness of this methodology in modeling water temperature and dissolved oxygen dynamics in real-world lakes. The proposed PGFM is also broadly applicable to a range of scientific fields where physics-based models are being used.
Authors: Pascale Gourdeau, Tosca Lechner, Ruth Urner
Abstract: We study the problem of computable multiclass learnability within the Probably Approximately Correct (PAC) learning framework of Valiant (1984). In the recently introduced computable PAC (CPAC) learning framework of Agarwal et al. (2020), both learners and the functions they output are required to be computable. We focus on the case of finite label space and start by proposing a computable version of the Natarajan dimension and showing that it characterizes CPAC learnability in this setting. We further generalize this result by establishing a meta-characterization of CPAC learnability for a certain family of dimensions: computable distinguishers. Distinguishers were defined by Ben-David et al. (1992) as a certain family of embeddings of the label space, with each embedding giving rise to a dimension. It was shown that the finiteness of each such dimension characterizes multiclass PAC learnability for finite label space in the non-computable setting. We show that the corresponding computable dimensions for distinguishers characterize CPAC learning. We conclude our analysis by proving that the DS dimension, which characterizes PAC learnability for infinite label space, cannot be expressed as a distinguisher (even in the case of finite label space).
Authors: Robert Akinie, Nana Kankam Brym Gyimah, Mansi Bhavsar, John Kelly
Abstract: The rapid advancement of machine learning (ML) and on-device computing has revolutionized various industries, including transportation, through the development of Connected and Autonomous Vehicles (CAVs) and Intelligent Transportation Systems (ITS). These technologies improve traffic management and vehicle safety, but also introduce significant security and privacy concerns, such as cyberattacks and data breaches. Traditional Intrusion Detection Systems (IDS) are increasingly inadequate in detecting modern threats, leading to the adoption of ML-based IDS solutions. Federated Learning (FL) has emerged as a promising method for enabling the decentralized training of IDS models on distributed edge devices without sharing sensitive data. However, deploying FL-based IDS in CAV networks poses unique challenges, including limited computational and memory resources on edge devices, competing demands from critical applications such as navigation and safety systems, and the need to scale across diverse hardware and connectivity conditions. To address these issues, we propose a hybrid server-edge FL framework that offloads pre-training to a central server while enabling lightweight fine-tuning on edge devices. This approach reduces memory usage by up to 42%, decreases training times by up to 75%, and achieves competitive IDS accuracy of up to 99.2%. Scalability analyses further demonstrates minimal performance degradation as the number of clients increase, highlighting the framework's feasibility for CAV networks and other IoT applications.
Authors: Yueyan Li, Caixia Yuan, Xiaojie Wang
Abstract: The study of mechanistic interpretability aims to reverse-engineer a model to explain its behaviors. While recent studies have focused on the static mechanism of a certain behavior, the training dynamics inside a model remain to be explored. In this work, we develop an interpretable method for fine-tuning and reveal the mechanism behind learning. We first propose the concept of node redundancy as an extension of intrinsic dimension and explain the idea behind circuit discovery from a fresh view. Based on the theory, we propose circuit-tuning, a two-stage algorithm that iteratively performs circuit discovery to mask out irrelevant edges and updates the remaining parameters responsible for a specific task. Experiments show that our method not only improves performance on a wide range of tasks but is also scalable while preserving general capabilities. We visualize and analyze the circuits before, during, and after fine-tuning, providing new insights into the self-organization mechanism of a neural network in the learning process.
Authors: Dongyuan Li, Satoshi Kosugi, Ying Zhang, Manabu Okumura, Feng Xia, Renhe Jiang
Abstract: Dynamic graph clustering aims to detect and track time-varying clusters in dynamic graphs, revealing the evolutionary mechanisms of complex real-world dynamic systems. Matrix factorization-based methods are promising approaches for this task; however, these methods often struggle with scalability and can be time-consuming when applied to large-scale dynamic graphs. Moreover, they tend to lack robustness and are vulnerable to real-world noisy data. To address these issues, we make three key contributions. First, to improve scalability, we propose temporal separated matrix factorization, where a single matrix is divided into multiple smaller matrices for independent factorization, resulting in faster computation. Second, to improve robustness, we introduce bi-clustering regularization, which jointly optimizes graph embedding and clustering, thereby filtering out noisy features from the graph embeddings. Third, to further enhance effectiveness and efficiency, we propose selective embedding updating, where we update only the embeddings of dynamic nodes while the embeddings of static nodes are fixed among different timestamps. Experimental results on six synthetic and five real-world benchmarks demonstrate the scalability, robustness and effectiveness of our proposed method. Source code is available at https://github.com/Clearloveyuan/DyG-MF.
Authors: Pawel Renc, Michal K. Grzeszczyk, Nassim Oufattole, Deirdre Goode, Yugang Jia, Szymon Bieganski, Matthew B. A. McDermott, Jaroslaw Was, Anthony E. Samir, Jonathan W. Cunningham, David W. Bates, Arkadiusz Sitek
Abstract: We developed the Enhanced Transformer for Health Outcome Simulation (ETHOS), an AI model that tokenizes patient health timelines (PHTs) from EHRs. ETHOS predicts future PHTs using transformer-based architectures. The Adaptive Risk Estimation System (ARES) employs ETHOS to compute dynamic and personalized risk probabilities for clinician-defined critical events. ARES incorporates a personalized explainability module that identifies key clinical factors influencing risk estimates for individual patients. ARES was evaluated on the MIMIC-IV v2.2 dataset in emergency department (ED) settings, benchmarking its performance against traditional early warning systems and machine learning models. We processed 299,721 unique patients from MIMIC-IV into 285,622 PHTs, with 60% including hospital admissions. The dataset contained over 357 million tokens. ETHOS outperformed benchmark models in predicting hospital admissions, ICU admissions, and prolonged hospital stays, achieving superior AUC scores. ETHOS-based risk estimates demonstrated robustness across demographic subgroups with strong model reliability, confirmed via calibration curves. The personalized explainability module provides insights into patient-specific factors contributing to risk. ARES, powered by ETHOS, advances predictive healthcare AI by providing dynamic, real-time, and personalized risk estimation with patient-specific explainability to enhance clinician trust. Its adaptability and superior accuracy position it as a transformative tool for clinical decision-making, potentially improving patient outcomes and resource allocation in emergency and inpatient settings. We release the full code at github.com/ipolharvard/ethos-ares to facilitate future research.
Authors: Dai Shi, Kuan Yan, Lequan Lin, Yue Zeng, Ting Zhang, Dmytro Matsypura, Mark C. Gillies, Ling Zhu, Junbin Gao
Abstract: Understanding disease progression at the molecular pathway level usually requires capturing both structural dependencies between pathways and the temporal dynamics of disease evolution. In this work, we solve the former challenge by developing a biologically informed graph-forming method to efficiently construct pathway graphs for subjects from our newly curated JR5558 mouse transcriptomics dataset. We then develop Graph-level Pseudotime Analysis (GPA) to infer graph-level trajectories that reveal how disease progresses at the population level, rather than in individual subjects. Based on the trajectories estimated by GPA, we identify the most sensitive pathways that drive disease stage transitions. In addition, we measure changes in pathway features using neural stochastic differential equations (SDEs), which enables us to formally define and compute pathway stability and disease bifurcation points (points of no return), two fundamental problems in disease progression research. We further extend our theory to the case when pathways can interact with each other, enabling a more comprehensive and multi-faceted characterization of disease phenotypes. The comprehensive experimental results demonstrate the effectiveness of our framework in reconstructing the dynamics of the pathway, identifying critical transitions, and providing novel insights into the mechanistic understanding of disease evolution.
Authors: Rucha Bhalchandra Joshi, Sagar Prakash Barad, Nidhi Tiwari, Subhankar Mishra
Abstract: Graph Neural Networks (GNNs) have emerged as powerful tools for learning representations of graph-structured data. In addition to real-valued GNNs, quaternion GNNs also perform well on tasks on graph-structured data. With the aim of reducing the energy footprint, we reduce the model size while maintaining accuracy comparable to that of the original-sized GNNs. This paper introduces Quaternion Message Passing Neural Networks (QMPNNs), a framework that leverages quaternion space to compute node representations. Our approach offers a generalizable method for incorporating quaternion representations into GNN architectures at one-fourth of the original parameter count. Furthermore, we present a novel perspective on Graph Lottery Tickets, redefining their applicability within the context of GNNs and QMPNNs. We specifically aim to find the initialization lottery from the subnetwork of the GNNs that can achieve comparable performance to the original GNN upon training. Thereby reducing the trainable model parameters even further. To validate the effectiveness of our proposed QMPNN framework and LTH for both GNNs and QMPNNs, we evaluate their performance on real-world datasets across three fundamental graph-based tasks: node classification, link prediction, and graph classification.
Authors: Annie Feng, Nishanth Kumar, Tomas Lozano-Perez, Leslie Pack-Kaelbling
Abstract: Efficient exploration is critical for learning relational models in large-scale environments with complex, long-horizon tasks. Random exploration methods often collect redundant or irrelevant data, limiting their ability to learn accurate relational models of the environment. Goal-literal babbling (GLIB) improves upon random exploration by setting and planning to novel goals, but its reliance on random actions and random novel goal selection limits its scalability to larger domains. In this work, we identify the principles underlying efficient exploration in relational domains: (1) operator initialization with demonstrations that cover the distinct lifted effects necessary for planning and (2) refining preconditions to collect maximally informative transitions by selecting informative goal-action pairs and executing plans to them. To demonstrate these principles, we introduce Baking-Large, a challenging domain with extensive state-action spaces and long-horizon tasks. We evaluate methods using oracle-driven demonstrations for operator initialization and precondition-targeting guidance to efficiently gather critical transitions. Experiments show that both the oracle demonstrations and precondition-targeting oracle guidance significantly improve sample efficiency and generalization, paving the way for future methods to use these principles to efficiently learn accurate relational models in complex domains.
Authors: Kareem Hegazy, Michael W. Mahoney, N. Benjamin Erichson
Abstract: Transformers have recently shown strong performance in time-series forecasting, but their all-to-all attention mechanism overlooks the (temporal) causal and often (temporally) local nature of data. We introduce Powerformer, a novel Transformer variant that replaces noncausal attention weights with causal weights that are reweighted according to a smooth heavy-tailed decay. This simple yet effective modification endows the model with an inductive bias favoring temporally local dependencies, while still allowing sufficient flexibility to learn the unique correlation structure of each dataset. Our empirical results demonstrate that Powerformer not only achieves state-of-the-art accuracy on public time-series benchmarks, but also that it offers improved interpretability of attention patterns. Our analyses show that the model's locality bias is amplified during training, demonstrating an interplay between time-series data and power-law-based attention. These findings highlight the importance of domain-specific modifications to the Transformer architecture for time-series forecasting, and they establish Powerformer as a strong, efficient, and principled baseline for future research and real-world applications.
Authors: Yihang Gao, Michael K. Ng, Vincent Y. F. Tan
Abstract: Kolmogorov--Arnold networks (KANs) have demonstrated their potential as an alternative to multi-layer perceptions (MLPs) in various domains, especially for science-related tasks. However, transfer learning of KANs remains a relatively unexplored area. In this paper, inspired by Tucker decomposition of tensors and evidence on the low tensor-rank structure in KAN parameter updates, we develop low tensor-rank adaptation (LoTRA) for fine-tuning KANs. We study the expressiveness of LoTRA based on Tucker decomposition approximations. Furthermore, we provide a theoretical analysis to select the learning rates for each LoTRA component to enable efficient training. Our analysis also shows that using identical learning rates across all components leads to inefficient training, highlighting the need for an adaptive learning rate strategy. Beyond theoretical insights, we explore the application of LoTRA for efficiently solving various partial differential equations (PDEs) by fine-tuning KANs. Additionally, we propose Slim KANs that incorporate the inherent low-tensor-rank properties of KAN parameter tensors to reduce model size while maintaining superior performance. Experimental results validate the efficacy of the proposed learning rate selection strategy and demonstrate the effectiveness of LoTRA for transfer learning of KANs in solving PDEs. Further evaluations on Slim KANs for function representation and image classification tasks highlight the expressiveness of LoTRA and the potential for parameter reduction through low tensor-rank decomposition.
Authors: Jack Spalding-Jamieson, Eliot Wong Robson, Da Wei Zheng
Abstract: For very large values of $k$, we consider methods for fast $k$-means clustering of massive datasets with $10^7\sim10^9$ points in high-dimensions ($d\geq100$). All current practical methods for this problem have runtimes at least $\Omega(k^2)$. We find that initialization routines are not a bottleneck for this case. Instead, it is critical to improve the speed of Lloyd's local-search algorithm, particularly the step that reassigns points to their closest center. Attempting to improve this step naturally leads us to leverage approximate nearest-neighbor search methods, although this alone is not enough to be practical. Instead, we propose a family of problems we call "Seeded Approximate Nearest-Neighbor Search", for which we propose "Seeded Search-Graph" methods as a solution.
Authors: Panqi Chen, Lei Cheng, Jianlong Li, Weichang Li, Weiqing Liu, Jiang Bian, Shikai Fang
Abstract: Tensor decomposition is a fundamental tool for analyzing multi-dimensional data by learning low-rank factors to represent high-order interactions. While recent works on temporal tensor decomposition have made significant progress by incorporating continuous timestamps in latent factors, they still struggle with general tensor data with continuous indexes not only in the temporal mode but also in other modes, such as spatial coordinates in climate data. Additionally, the problem of determining the tensor rank remains largely unexplored in temporal tensor models. To address these limitations, we propose \underline{G}eneralized temporal tensor decomposition with \underline{R}ank-r\underline{E}vealing laten\underline{T}-ODE (GRET). Our approach encodes continuous spatial indexes as learnable Fourier features and employs neural ODEs in latent space to learn the temporal trajectories of factors. To automatically reveal the rank of temporal tensors, we introduce a rank-revealing Gaussian-Gamma prior over the factor trajectories. We develop an efficient variational inference scheme with an analytical evidence lower bound, enabling sampling-free optimization. Through extensive experiments on both synthetic and real-world datasets, we demonstrate that GRET not only reveals the underlying ranks of temporal tensors but also significantly outperforms existing methods in prediction performance and robustness against noise.
Authors: Yifang Chen, Xiaoyu Li, Yingyu Liang, Zhenmei Shi, Zhao Song
Abstract: We investigate the fundamental limits of transformer-based foundation models, extending our analysis to include Visual Autoregressive (VAR) transformers. VAR represents a big step toward generating images using a novel, scalable, coarse-to-fine ``next-scale prediction'' framework. These models set a new quality bar, outperforming all previous methods, including Diffusion Transformers, while having state-of-the-art performance for image synthesis tasks. Our primary contributions establish that, for single-head VAR transformers with a single self-attention layer and single interpolation layer, the VAR Transformer is universal. From the statistical perspective, we prove that such simple VAR transformers are universal approximators for any image-to-image Lipschitz functions. Furthermore, we demonstrate that flow-based autoregressive transformers inherit similar approximation capabilities. Our results provide important design principles for effective and computationally efficient VAR Transformer strategies that can be used to extend their utility to more sophisticated VAR models in image generation and other related areas.
Authors: Sanket Jantre, Tianle Wang, Gilchan Park, Kriti Chopra, Nicholas Jeon, Xiaoning Qian, Nathan M. Urban, Byung-Jun Yoon
Abstract: Identification of protein-protein interactions (PPIs) helps derive cellular mechanistic understanding, particularly in the context of complex conditions such as neurodegenerative disorders, metabolic syndromes, and cancer. Large Language Models (LLMs) have demonstrated remarkable potential in predicting protein structures and interactions via automated mining of vast biomedical literature; yet their inherent uncertainty remains a key challenge for deriving reproducible findings, critical for biomedical applications. In this study, we present an uncertainty-aware adaptation of LLMs for PPI analysis, leveraging fine-tuned LLaMA-3 and BioMedGPT models. To enhance prediction reliability, we integrate LoRA ensembles and Bayesian LoRA models for uncertainty quantification (UQ), ensuring confidence-calibrated insights into protein behavior. Our approach achieves competitive performance in PPI identification across diverse disease contexts while addressing model uncertainty, thereby enhancing trustworthiness and reproducibility in computational biology. These findings underscore the potential of uncertainty-aware LLM adaptation for advancing precision medicine and biomedical research.
Authors: Guanglong Sun, Hongwei Yan, Liyuan Wang, Qian Li, Bo Lei, Yi Zhong
Abstract: Knowledge distillation (KD) is a powerful strategy for training deep neural networks (DNNs). Although it was originally proposed to train a more compact ``student'' model from a large ``teacher'' model, many recent efforts have focused on adapting it to promote generalization of the model itself, such as online KD and self KD. % as an effective way Here, we propose an accessible and compatible strategy named Spaced KD to improve the effectiveness of both online KD and self KD, in which the student model distills knowledge from a teacher model trained with a space interval ahead. This strategy is inspired by a prominent theory named \emph{spacing effect} in biological learning and memory, positing that appropriate intervals between learning trials can significantly enhance learning performance. With both theoretical and empirical analyses, we demonstrate that the benefits of the proposed Spaced KD stem from convergence to a flatter loss landscape during stochastic gradient descent (SGD). We perform extensive experiments to validate the effectiveness of Spaced KD in improving the learning performance of DNNs (e.g., the performance gain is up to 2.31\% and 3.34\% on Tiny-ImageNet over online KD and self KD, respectively).
Authors: Yeho Gwon, Sehyun Hwang, Hoyoung Kim, Jungseul Ok, Suha Kwak
Abstract: This paper introduces a cost-efficient active learning (AL) framework for classification, featuring a novel query design called candidate set query. Unlike traditional AL queries requiring the oracle to examine all possible classes, our method narrows down the set of candidate classes likely to include the ground-truth class, significantly reducing the search space and labeling cost. Moreover, we leverage conformal prediction to dynamically generate small yet reliable candidate sets, adapting to model enhancement over successive AL rounds. To this end, we introduce an acquisition function designed to prioritize data points that offer high information gain at lower cost. Empirical evaluations on CIFAR-10, CIFAR-100, and ImageNet64x64 demonstrate the effectiveness and scalability of our framework. Notably, it reduces labeling cost by 42% on ImageNet64x64.
Authors: Aojun Lu, Junchao Ke, Chunhui Ding, Jiahao Fan, Yanan Sun
Abstract: Deep neural networks have demonstrated remarkable success in machine learning; however, they remain fundamentally ill-suited for Continual Learning (CL). Recent research has increasingly focused on achieving CL without the need for rehearsal. Among these, parameter isolation-based methods have proven particularly effective in enhancing CL by optimizing model weights for each incremental task. Despite their success, they fall short in optimizing architectures tailored to distinct incremental tasks. To address this limitation, updating a group of models with different architectures offers a promising alternative to the traditional CL paradigm that relies on a single unified model. Building on this insight, this study introduces a novel Population-based Continual Learning (PCL) framework. PCL extends CL to the architectural level by maintaining and evolving a population of neural network architectures, which are continually refined for the current task through NAS. Importantly, the well-evolved population for the current incremental task is naturally inherited by the subsequent one, thereby facilitating forward transfer, a crucial objective in CL. Throughout the CL process, the population evolves, yielding task-specific architectures that collectively form a robust CL system. Experimental results demonstrate that PCL outperforms state-of-the-art rehearsal-free CL methods that employs a unified model, highlighting its potential as a new paradigm for CL.
Authors: Zeman Li, Yuan Deng, Peilin Zhong, Meisam Razaviyayn, Vahab Mirrokni
Abstract: Modern machine learning models are trained on diverse datasets and tasks to improve generalization. A key challenge in multitask learning is determining the optimal data mixing and sampling strategy across different data sources. Prior research in this multi-task learning setting has primarily focused on mitigating gradient conflicts between tasks. However, we observe that many real-world multitask learning scenarios-such as multilingual training and multi-domain learning in large foundation models-exhibit predominantly positive task interactions with minimal or no gradient conflict. Building on this insight, we introduce PiKE (Positive gradient interaction-based K-task weights Estimator), an adaptive data mixing algorithm that dynamically adjusts task contributions throughout training. PiKE optimizes task sampling to minimize overall loss, effectively leveraging positive gradient interactions with almost no additional computational overhead. We establish theoretical convergence guarantees for PiKE and demonstrate its superiority over static and non-adaptive mixing strategies. Additionally, we extend PiKE to promote fair learning across tasks, ensuring balanced progress and preventing task underrepresentation. Empirical evaluations on large-scale language model pretraining show that PiKE consistently outperforms existing heuristic and static mixing strategies, leading to faster convergence and improved downstream task performance.
Authors: Yaohua Zang, Phaedon-Stelios Koutsourelakis
Abstract: Solving parametric partial differential equations (PDEs) and associated PDE-based, inverse problems is a central task in engineering and physics, yet existing neural operator methods struggle with high-dimensional, discontinuous inputs and require large amounts of {\em labeled} training data. We propose the Deep Generative Neural Operator (DGNO), a physics-aware framework that addresses these challenges by leveraging a deep, generative, probabilistic model in combination with a set of lower-dimensional, latent variables that simultaneously encode PDE-inputs and PDE-outputs. This formulation can make use of unlabeled data and significantly improves inverse problem-solving, particularly for discontinuous or discrete-valued input functions. DGNO enforces physics constraints without labeled data by incorporating as virtual observables, weak-form residuals based on compactly supported radial basis functions (CSRBFs). These relax regularity constraints and eliminate higher-order derivatives from the objective function. We also introduce MultiONet, a novel neural operator architecture, which is a more expressive generalization of the popular DeepONet that significantly enhances the approximating power of the proposed model. These innovations make DGNO particularly effective for challenging forward and inverse, PDE-based problems, such as those involving multi-phase media. Numerical experiments demonstrate that DGNO achieves higher accuracy across multiple benchmarks while exhibiting robustness to noise and strong generalization to out-of-distribution cases. Its adaptability, and the ability to handle sparse, noisy data while providing probabilistic estimates, make DGNO a powerful tool for scientific and engineering applications.
Authors: Changxi Zhu, Mehdi Dastani, Shihan Wang
Abstract: In decentralized multi-agent deep reinforcement learning (MADRL), communication can help agents to gain a better understanding of the environment to better coordinate their behaviors. Nevertheless, communication may involve uncertainty, which potentially introduces variance to the learning of decentralized agents. In this paper, we focus on a specific decentralized MADRL setting with communication and conduct a theoretical analysis to study the variance that is caused by communication in policy gradients. We propose modular techniques to reduce the variance in policy gradients during training. We adopt our modular techniques into two existing algorithms for decentralized MADRL with communication and evaluate them on multiple tasks in the StarCraft Multi-Agent Challenge and Traffic Junction domains. The results show that decentralized MADRL communication methods extended with our proposed techniques not only achieve high-performing agents but also reduce variance in policy gradients during training.
Authors: Lingkun Luo, Shiqiang Hu, Liming Chen
Abstract: In domain adaptation (DA), the effectiveness of deep learning-based models is often constrained by batch learning strategies that fail to fully apprehend the global statistical and geometric characteristics of data distributions. Addressing this gap, we introduce 'Global Awareness Enhanced Domain Adaptation' (GAN-DA), a novel approach that transcends traditional batch-based limitations. GAN-DA integrates a unique predefined feature representation (PFR) to facilitate the alignment of cross-domain distributions, thereby achieving a comprehensive global statistical awareness. This representation is innovatively expanded to encompass orthogonal and common feature aspects, which enhances the unification of global manifold structures and refines decision boundaries for more effective DA. Our extensive experiments, encompassing 27 diverse cross-domain image classification tasks, demonstrate GAN-DA's remarkable superiority, outperforming 24 established DA methods by a significant margin. Furthermore, our in-depth analyses shed light on the decision-making processes, revealing insights into the adaptability and efficiency of GAN-DA. This approach not only addresses the limitations of existing DA methodologies but also sets a new benchmark in the realm of domain adaptation, offering broad implications for future research and applications in this field.
Authors: Zhaoying Wang, Yingdan Shi, Xiang Liu, Can Chen, Jun Wen, Ren Wang
Abstract: Drug-side effect research is vital for understanding adverse reactions arising in complex multi-drug therapies. However, the scarcity of higher-order datasets that capture the combinatorial effects of multiple drugs severely limits progress in this field. Existing resources such as TWOSIDES primarily focus on pairwise interactions. To fill this critical gap, we introduce HODDI, the first Higher-Order Drug-Drug Interaction Dataset, constructed from U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) records spanning the past decade, to advance computational pharmacovigilance. HODDI contains 109,744 records involving 2,506 unique drugs and 4,569 unique side effects, specifically curated to capture multi-drug interactions and their collective impact on adverse effects. Comprehensive statistical analyses demonstrate HODDI's extensive coverage and robust analytical metrics, making it a valuable resource for studying higher-order drug relationships. Evaluating HODDI with multiple models, we found that simple Multi-Layer Perceptron (MLP) can outperform graph models, while hypergraph models demonstrate superior performance in capturing complex multi-drug interactions, further validating HODDI's effectiveness. Our findings highlight the inherent value of higher-order information in drug-side effect prediction and position HODDI as a benchmark dataset for advancing research in pharmacovigilance, drug safety, and personalized medicine. The dataset and codes are available at https://github.com/TIML-Group/HODDI.
Authors: Zhixun Li, Dingshuo Chen, Tong Zhao, Daixin Wang, Hongrui Liu, Zhiqiang Zhang, Jun Zhou, Jeffrey Xu Yu
Abstract: Graph Neural Networks (GNNs) have achieved great success in dealing with non-Euclidean graph-structured data and have been widely deployed in many real-world applications. However, their effectiveness is often jeopardized under class-imbalanced training sets. Most existing studies have analyzed class-imbalanced node classification from a supervised learning perspective, but they do not fully utilize the large number of unlabeled nodes in semi-supervised scenarios. We claim that the supervised signal is just the tip of the iceberg and a large number of unlabeled nodes have not yet been effectively utilized. In this work, we propose IceBerg, a debiased self-training framework to address the class-imbalanced and few-shot challenges for GNNs at the same time. Specifically, to figure out the Matthew effect and label distribution shift in self-training, we propose Double Balancing, which can largely improve the performance of existing baselines with just a few lines of code as a simple plug-and-play module. Secondly, to enhance the long-range propagation capability of GNNs, we disentangle the propagation and transformation operations of GNNs. Therefore, the weak supervision signals can propagate more effectively to address the few-shot issue. In summary, we find that leveraging unlabeled nodes can significantly enhance the performance of GNNs in class-imbalanced and few-shot scenarios, and even small, surgical modifications can lead to substantial performance improvements. Systematic experiments on benchmark datasets show that our method can deliver considerable performance gain over existing class-imbalanced node classification baselines. Additionally, due to IceBerg's outstanding ability to leverage unsupervised signals, it also achieves state-of-the-art results in few-shot node classification scenarios. The code of IceBerg is available at: https://github.com/ZhixunLEE/IceBerg.
Authors: Gennadiy Averkov, Christopher Hojny, Maximilian Merkert
Abstract: To confirm that the expressive power of ReLU neural networks grows with their depth, the function $F_n = \max \{0,x_1,\ldots,x_n\}$ has been considered in the literature. A conjecture by Hertrich, Basu, Di Summa, and Skutella [NeurIPS 2021] states that any ReLU network that exactly represents $F_n$ has at least $\lceil\log_2 (n+1)\rceil$ hidden layers. The conjecture has recently been confirmed for networks with integer weights by Haase, Hertrich, and Loho [ICLR 2023]. We follow up on this line of research and show that, within ReLU networks whose weights are decimal fractions, $F_n$ can only be represented by networks with at least $\lceil\log_3 (n+1)\rceil$ hidden layers. Moreover, if all weights are $N$-ary fractions, then $F_n$ can only be represented by networks with at least $\Omega( \frac{\ln n}{\ln \ln N})$ layers. These results are a partial confirmation of the above conjecture for rational ReLU networks, and provide the first non-constant lower bound on the depth of practically relevant ReLU networks.
Authors: Yunchu Han, Zhaojun Nan, Sheng Zhou, Zhisheng Niu
Abstract: The rapid development of deep neural networks (DNNs) is inherently accompanied by the problem of high computational costs. To tackle this challenge, dynamic voltage frequency scaling (DVFS) is emerging as a promising technology for balancing the latency and energy consumption of DNN inference by adjusting the computing frequency of processors. However, most existing models of DNN inference time are based on the CPU-DVFS technique, and directly applying the CPU-DVFS model to DNN inference on GPUs will lead to significant errors in optimizing latency and energy consumption. In this paper, we propose a DVFS-aware latency model to precisely characterize DNN inference time on GPUs. We first formulate the DNN inference time based on extensive experiment results for different devices and analyze the impact of fitting parameters. Then by dividing DNNs into multiple blocks and obtaining the actual inference time, the proposed model is further verified. Finally, we compare our proposed model with the CPU-DVFS model in two specific cases. Evaluation results demonstrate that local inference optimization with our proposed model achieves a reduction of no less than 66% and 69% in inference time and energy consumption respectively. In addition, cooperative inference with our proposed model can improve the partition policy and reduce the energy consumption compared to the CPU-DVFS model.
Authors: Ofir Schlisselberg, Ran Darshan
Abstract: In traditional machine learning, models are defined by a set of parameters, which are optimized to perform specific tasks. In neural networks, these parameters correspond to the synaptic weights. However, in reality, it is often infeasible to control or update all weights. This challenge is not limited to artificial networks but extends to biological networks, such as the brain, where the extent of distributed synaptic weight modification during learning remains unclear. Motivated by these insights, we theoretically investigate how different allocations of a fixed number of learnable weights influence the capacity of neural networks. Using a teacher-student setup, we introduce a benchmark to quantify the expressivity associated with each allocation. We establish conditions under which allocations have maximal or minimal expressive power in linear recurrent neural networks and linear multi-layer feedforward networks. For suboptimal allocations, we propose heuristic principles to estimate their expressivity. These principles extend to shallow ReLU networks as well. Finally, we validate our theoretical findings with empirical experiments. Our results emphasize the critical role of strategically distributing learnable weights across the network, showing that a more widespread allocation generally enhances the network's expressive power.
Authors: Maty\'a\v{s} Lorenc
Abstract: In this paper, we experiment with novelty-based variants of OpenAI-ES, the NS-ES and NSR-ES algorithms, and evaluate their effectiveness in training complex, transformer-based architectures designed for the problem of reinforcement learning such as Decision Transformers. We also test if we can accelerate the novelty-based training of these larger models by seeding the training by a pretrained models. By this, we build on our previous work, where we tested the ability of evolution strategies - specifically the aforementioned OpenAI-ES - to train the Decision Transformer architecture. The results were mixed. NS-ES showed progress, but it would clearly need many more iterations for it to yield interesting results. NSR-ES, on the other hand, proved quite capable of being straightforwardly used on larger models, since its performance appears as similar between the feed-forward model and Decision Transformer, as it was for the OpenAI-ES in our previous work.
Authors: Zhaoxian Wu, Quan Xian, Tayfun Gokmen, Omobayode Fagbohungbe, Tianyi Chen
Abstract: As the economic and environmental costs of training and deploying large vision or language models increase dramatically, analog in-memory computing (AIMC) emerges as a promising energy-efficient solution. However, the training perspective, especially its training dynamic, is underexplored. In AIMC hardware, the trainable weights are represented by the conductance of resistive elements and updated using consecutive electrical pulses. Among all the physical properties of resistive elements, the response to the pulses directly affects the training dynamics. This paper first provides a theoretical foundation for gradient-based training on AIMC hardware and studies the impact of response functions. We demonstrate that noisy update and asymmetric response functions negatively impact Analog SGD by imposing an implicit penalty term on the objective. To overcome the issue, Tiki-Taka, a residual learning algorithm, converges exactly to a critical point by optimizing a main array and a residual array bilevelly. The conclusion is supported by simulations validating our theoretical insights.
Authors: Alice Bizeul, Thomas Sutter, Alain Ryser, Bernhard Sch\"olkopf, Julius von K\"ugelgen, Julia E. Vogt
Abstract: Predicting masked from visible parts of an image is a powerful self-supervised approach for visual representation learning. However, the common practice of masking random patches of pixels exhibits certain failure modes, which can prevent learning meaningful high-level features, as required for downstream tasks. We propose an alternative masking strategy that operates on a suitable transformation of the data rather than on the raw pixels. Specifically, we perform principal component analysis and then randomly mask a subset of components, which accounts for a fixed ratio of the data variance. The learning task then amounts to reconstructing the masked components from the visible ones. Compared to local patches of pixels, the principal components of images carry more global information. We thus posit that predicting masked from visible components involves more high-level features, allowing our masking strategy to extract more useful representations. This is corroborated by our empirical findings which demonstrate improved image classification performance for component over pixel masking. Our method thus constitutes a simple and robust data-driven alternative to traditional masked image modeling approaches.
Authors: Qi Wang, Tianfei Zhou, Ye Yuan, Rui Mao
Abstract: Continual Graph Learning (CGL), which aims to accommodate new tasks over evolving graph data without forgetting prior knowledge, is garnering significant research interest. Mainstream solutions adopt the memory replay-based idea, ie, caching representative data from earlier tasks for retraining the graph model. However, this strategy struggles with scalability issues for constantly evolving graphs and raises concerns regarding data privacy. Inspired by recent advancements in the prompt-based learning paradigm, this paper introduces a novel prompt-driven continual graph learning (PROMPTCGL) framework, which learns a separate prompt for each incoming task and maintains the underlying graph neural network model fixed. In this way, PROMPTCGL naturally avoids catastrophic forgetting of knowledge from previous tasks. More specifically, we propose hierarchical prompting to instruct the model from both feature- and topology-level to fully address the variability of task graphs in dynamic continual learning. Additionally, we develop a personalized prompt generator to generate tailored prompts for each graph node while minimizing the number of prompts needed, leading to constant memory consumption regardless of the graph scale. Extensive experiments on four benchmarks show that PROMPTCGL achieves superior performance against existing CGL approaches while significantly reducing memory consumption. Our code is available at https://github.com/QiWang98/PromptCGL.
Authors: Emanuel Sommer, Jakob Robnik, Giorgi Nozadze, Uros Seljak, David R\"ugamer
Abstract: Despite recent advances, sampling-based inference for Bayesian Neural Networks (BNNs) remains a significant challenge in probabilistic deep learning. While sampling-based approaches do not require a variational distribution assumption, current state-of-the-art samplers still struggle to navigate the complex and highly multimodal posteriors of BNNs. As a consequence, sampling still requires considerably longer inference times than non-Bayesian methods even for small neural networks, despite recent advances in making software implementations more efficient. Besides the difficulty of finding high-probability regions, the time until samplers provide sufficient exploration of these areas remains unpredictable. To tackle these challenges, we introduce an ensembling approach that leverages strategies from optimization and a recently proposed sampler called Microcanonical Langevin Monte Carlo (MCLMC) for efficient, robust and predictable sampling performance. Compared to approaches based on the state-of-the-art No-U-Turn Sampler, our approach delivers substantial speedups up to an order of magnitude, while maintaining or improving predictive performance and uncertainty quantification across diverse tasks and data modalities. The suggested Microcanonical Langevin Ensembles and modifications to MCLMC additionally enhance the method's predictability in resource requirements, facilitating easier parallelization. All in all, the proposed method offers a promising direction for practical, scalable inference for BNNs.
Authors: Riccardo Cadei, Ilker Demirel, Piersilvio De Bartolomeis, Lukas Lindorfer, Sylvia Cremer, Cordelia Schmid, Francesco Locatello
Abstract: A plethora of real-world scientific investigations is waiting to scale with the support of trustworthy predictive models that can reduce the need for costly data annotations. We focus on causal inferences on a target experiment with unlabeled factual outcomes, retrieved by a predictive model fine-tuned on a labeled similar experiment. First, we show that factual outcome estimation via Empirical Risk Minimization (ERM) may fail to yield valid causal inferences on the target population, even in a randomized controlled experiment and infinite training samples. Then, we propose to leverage the observed experimental settings during training to empower generalization to downstream interventional investigations, ``Causal Lifting'' the predictive model. We propose Deconfounded Empirical Risk Minimization (DERM), a new simple learning procedure minimizing the risk over a fictitious target population, preventing potential confounding effects. We validate our method on both synthetic and real-world scientific data. Notably, for the first time, we zero-shot generalize causal inferences on ISTAnt dataset (without annotation) by causal lifting a predictive model on our experiment variant.
Authors: Won-Jun Jang, Hyeon-Seo Park, Si-Hyeon Lee
Abstract: Federated ensemble distillation addresses client heterogeneity by generating pseudo-labels for an unlabeled server dataset based on client predictions and training the server model using the pseudo-labeled dataset. The unlabeled server dataset can either be pre-existing or generated through a data-free approach. The effectiveness of this approach critically depends on the method of assigning weights to client predictions when creating pseudo-labels, especially in highly heterogeneous settings. Inspired by theoretical results from GANs, we propose a provably near-optimal weighting method that leverages client discriminators trained with a server-distributed generator and local datasets. Our experiments on various image classification tasks demonstrate that the proposed method significantly outperforms baselines. Furthermore, we show that the additional communication cost, client-side privacy leakage, and client-side computational overhead introduced by our method are negligible, both in scenarios with and without a pre-existing server dataset.
Authors: Yawei Li, David R\"ugamer, Bernd Bischl, Mina Rezaei
Abstract: Fine-tuned large language models (LLMs) often exhibit overconfidence, particularly when trained on small datasets, resulting in poor calibration and inaccurate uncertainty estimates. Evidential Deep Learning (EDL), an uncertainty-aware approach, enables uncertainty estimation in a single forward pass, making it a promising method for calibrating fine-tuned LLMs. However, despite its computational efficiency, EDL is prone to overfitting, as its training objective can result in overly concentrated probability distributions. To mitigate this, we propose regularizing EDL by incorporating an information bottleneck (IB). Our approach IB-EDL suppresses spurious information in the evidence generated by the model and encourages truly predictive information to influence both the predictions and uncertainty estimates. Extensive experiments across various fine-tuned LLMs and tasks demonstrate that IB-EDL outperforms both existing EDL and non-EDL approaches. By improving the trustworthiness of LLMs, IB-EDL facilitates their broader adoption in domains requiring high levels of confidence calibration. Code is available at https://github.com/sandylaker/ib-edl.
Authors: Finn Rietz, Oleg Smirnov, Sara Karimi, Lele Cao
Abstract: Prompting has emerged as the dominant paradigm for adapting large, pre-trained transformer-based models to downstream tasks. The Prompting Decision Transformer (PDT) enables large-scale, multi-task offline reinforcement learning pre-training by leveraging stochastic trajectory prompts to identify the target task. However, these prompts are sampled uniformly from expert demonstrations, overlooking a critical limitation: Not all prompts are equally informative for differentiating between tasks. To address this, we propose an inference time bandit-based prompt-tuning framework that explores and optimizes trajectory prompt selection to enhance task performance. Our experiments indicate not only clear performance gains due to bandit-based prompt-tuning, but also better sample complexity, scalability, and prompt space exploration compared to prompt-tuning baselines.
Authors: Shogo Iwazaki, Shion Takeno
Abstract: We study the Gaussian process (GP) bandit problem, whose goal is to minimize regret under an unknown reward function lying in some reproducing kernel Hilbert space (RKHS). The maximum posterior variance analysis is vital in analyzing near-optimal GP bandit algorithms such as maximum variance reduction (MVR) and phased elimination (PE). Therefore, we first show the new upper bound of the maximum posterior variance, which improves the dependence of the noise variance parameters of the GP. By leveraging this result, we refine the MVR and PE to obtain (i) a nearly optimal regret upper bound in the noiseless setting and (ii) regret upper bounds that are optimal with respect to the RKHS norm of the reward function. Furthermore, as another application of our proposed bound, we analyze the GP bandit under the time-varying noise variance setting, which is the kernelized extension of the linear bandit with heteroscedastic noise. For this problem, we show that MVR and PE-based algorithms achieve noise variance-dependent regret upper bounds, which matches our regret lower bound.
Authors: Gauri Pradhan, Joonas J\"alk\"o, Marlon Tobaben, Antti Honkela
Abstract: Membership Inference Attacks (MIAs) have emerged as a valuable framework for evaluating privacy leakage by machine learning models. Score-based MIAs are distinguished, in particular, by their ability to exploit the confidence scores that the model generates for particular inputs. Existing score-based MIAs implicitly assume that the adversary has access to the target model's hyperparameters, which can be used to train the shadow models for the attack. In this work, we demonstrate that the knowledge of target hyperparameters is not a prerequisite for MIA in the transfer learning setting. Based on this, we propose a novel approach to select the hyperparameters for training the shadow models for MIA when the attacker has no prior knowledge about them by matching the output distributions of target and shadow models. We demonstrate that using the new approach yields hyperparameters that lead to an attack near indistinguishable in performance from an attack that uses target hyperparameters to train the shadow models. Furthermore, we study the empirical privacy risk of unaccounted use of training data for hyperparameter optimization (HPO) in differentially private (DP) transfer learning. We find no statistically significant evidence that performing HPO using training data would increase vulnerability to MIA.
Authors: Vasileios Tsouvalas, Tanir Ozcelebi, Nirvana Meratnia
Abstract: Federated Learning (FL) traditionally assumes homogeneous client tasks; however, in real-world scenarios, clients often specialize in diverse tasks, introducing task heterogeneity. To address this challenge, Many-Task FL (MaT-FL) has emerged, enabling clients to collaborate effectively despite task diversity. Existing MaT-FL approaches rely on client grouping or personalized layers, requiring the server to manage individual models and failing to account for clients handling multiple tasks. We propose MaTU, a MaT-FL approach that enables joint learning of task vectors across clients, eliminating the need for clustering or client-specific weight storage at the server. Our method introduces a novel aggregation mechanism that determines task similarity based on the direction of clients task vectors and constructs a unified task vector encapsulating all tasks. To address task-specific requirements, we augment the unified task vector with lightweight modulators that facilitate knowledge transfer among related tasks while disentangling dissimilar ones. Evaluated across 30 datasets, MaTU achieves superior performance over state-of-the-art MaT-FL approaches, with results comparable to per-task fine-tuning, while delivering significant communication savings.
Authors: Filip Ekstr\"om Kelvinius, Zheng Zhao, Fredrik Lindsten
Abstract: A recent line of research has exploited pre-trained generative diffusion models as priors for solving Bayesian inverse problems. We contribute to this research direction by designing a sequential Monte Carlo method for linear-Gaussian inverse problems which builds on ``decoupled diffusion", where the generative process is designed such that larger updates to the sample are possible. The method is asymptotically exact and we demonstrate the effectiveness of our Decoupled Diffusion Sequential Monte Carlo (DDSMC) algorithm on both synthetic data and image reconstruction tasks. Further, we demonstrate how the approach can be extended to discrete data.
Authors: Yiru Jiao, Sander van Cranenburgh, Simeon Calvert, Hans van Lint
Abstract: Informative representations enhance model performance and generalisability in downstream tasks. However, learning self-supervised representations for spatially characterised time series, like traffic interactions, poses challenges as it requires maintaining fine-grained similarity relations in the latent space. In this study, we incorporate two structure-preserving regularisers for the contrastive learning of spatial time series: one regulariser preserves the topology of similarities between instances, and the other preserves the graph geometry of similarities across spatial and temporal dimensions. To balance contrastive learning and structure preservation, we propose a dynamic mechanism that adaptively weighs the trade-off and stabilises training. We conduct experiments on multivariate time series classification, as well as macroscopic and microscopic traffic prediction. For all three tasks, our approach preserves the structures of similarity relations more effectively and improves state-of-the-art task performances. The proposed approach can be applied to an arbitrary encoder and is particularly beneficial for time series with spatial or geographical features. Furthermore, this study suggests that higher similarity structure preservation indicates more informative and useful representations. This may help to understand the contribution of representation learning in pattern recognition with neural networks. Our code is made openly accessible with all resulting data at https://github.com/yiru-jiao/spclt.
Authors: Shang Liu, Hanzhao Wang, Zhongyao Ma, Xiaocheng Li
Abstract: Human-annotated preference data play an important role in aligning large language models (LLMs). In this paper, we investigate the questions of assessing the performance of human annotators and incentivizing them to provide high-quality annotations. The quality assessment of language/text annotation faces two challenges: (i) the intrinsic heterogeneity among annotators, which prevents the classic methods that assume the underlying existence of a true label; and (ii) the unclear relationship between the annotation quality and the performance of downstream tasks, which excludes the possibility of inferring the annotators' behavior based on the model performance trained from the annotation data. Then we formulate a principal-agent model to characterize the behaviors of and the interactions between the company and the human annotators. The model rationalizes a practical mechanism of a bonus scheme to incentivize annotators which benefits both parties and it underscores the importance of the joint presence of an assessment system and a proper contract scheme. From a technical perspective, our analysis extends the existing literature on the principal-agent model by considering a continuous action space for the agent. We show the gap between the first-best and the second-best solutions (under the continuous action space) is of $\Theta(1/\sqrt{n \log n})$ for the binary contracts and $\Theta(1/n)$ for the linear contracts, where $n$ is the number of samples used for performance assessment; this contrasts with the known result of $\exp(-\Theta(n))$ for the binary contracts when the action space is discrete. Throughout the paper, we use real preference annotation data to accompany our discussions.
Authors: Peng Wu, Haoxuan Li, Chunyuan Zheng, Yan Zeng, Jiawei Chen, Yang Liu, Ruocheng Guo, Kun Zhang
Abstract: Counterfactual inference aims to estimate the counterfactual outcome at the individual level given knowledge of an observed treatment and the factual outcome, with broad applications in fields such as epidemiology, econometrics, and management science. Previous methods rely on a known structural causal model (SCM) or assume the homogeneity of the exogenous variable and strict monotonicity between the outcome and exogenous variable. In this paper, we propose a principled approach for identifying and estimating the counterfactual outcome. We first introduce a simple and intuitive rank preservation assumption to identify the counterfactual outcome without relying on a known structural causal model. Building on this, we propose a novel ideal loss for theoretically unbiased learning of the counterfactual outcome and further develop a kernel-based estimator for its empirical estimation. Our theoretical analysis shows that the rank preservation assumption is not stronger than the homogeneity and strict monotonicity assumptions, and shows that the proposed ideal loss is convex, and the proposed estimator is unbiased. Extensive semi-synthetic and real-world experiments are conducted to demonstrate the effectiveness of the proposed method.
Authors: Haofei Lu, Yifei Shen, Dongsheng Li, Junliang Xing, Dongqi Han
Abstract: Diffusion models have shown great promise in decision-making, also known as diffusion planning. However, the slow inference speeds limit their potential for broader real-world applications. Here, we introduce Habi, a general framework that transforms powerful but slow diffusion planning models into fast decision-making models, which mimics the cognitive process in the brain that costly goal-directed behavior gradually transitions to efficient habitual behavior with repetitive practice. Even using a laptop CPU, the habitized model can achieve an average 800+ Hz decision-making frequency (faster than previous diffusion planners by orders of magnitude) on standard offline reinforcement learning benchmarks D4RL, while maintaining comparable or even higher performance compared to its corresponding diffusion planner. Our work proposes a fresh perspective of leveraging powerful diffusion models for real-world decision-making tasks. We also provide robust evaluations and analysis, offering insights from both biological and engineering perspectives for efficient and effective decision-making.
Authors: Alessio benavoli, Alessandro facchini, Marco Zaffalon
Abstract: The off-switch problem is a critical challenge in AI control: if an AI system resists being switched off, it poses a significant risk. In this paper, we model the off-switch problem as a signalling game, where a human decision-maker communicates its preferences about some underlying decision problem to an AI agent, which then selects actions to maximise the human's utility. We assume that the human is a bounded rational agent and explore various bounded rationality mechanisms. Using real machine learning models, we reprove prior results and demonstrate that a necessary condition for an AI system to refrain from disabling its off-switch is its uncertainty about the human's utility. We also analyse how message costs influence optimal strategies and extend the analysis to scenarios involving incomparability.
Authors: Baobing Zhang, Paul Sullivan, Benjie Tang, Ghulam Nabi, Mustafa Suphi Erden
Abstract: In laparoscopy surgical training and evaluation, real-time detection of surgical actions with interpretable outputs is crucial for automated and real-time instructional feedback and skill development. Such capability would enable development of machine guided training systems. This paper presents a rapid deployment approach utilizing automated machine learning methods, based on surgical action data collected from both experienced and trainee surgeons. The proposed approach effectively tackles the challenge of highly imbalanced class distributions, ensuring robust predictions across varying skill levels of surgeons. Additionally, our method partially incorporates model transparency, addressing the reliability requirements in medical applications. Compared to deep learning approaches, traditional machine learning models not only facilitate efficient rapid deployment but also offer significant advantages in interpretability. Through experiments, this study demonstrates the potential of this approach to provide quick, reliable and effective real-time detection in surgical training environments
Authors: Qian Chen, Xingjian Dong, Kui Hu, Kangkang Chen, Zhike Peng, Guang Meng
Abstract: Neural networks (NNs), with their powerful nonlinear mapping and end-to-end capabilities, are widely applied in mechanical intelligent fault diagnosis (IFD). However, as typical black-box models, they pose challenges in understanding their decision basis and logic, limiting their deployment in high-reliability scenarios. Hence, various methods have been proposed to enhance the interpretability of IFD. Among these, post-hoc approaches can provide explanations without changing model architecture, preserving its flexibility and scalability. However, existing post-hoc methods often suffer from limitations in explanation forms. They either require preprocessing that disrupts the end-to-end nature or overlook fault mechanisms, leading to suboptimal explanations. To address these issues, we derived the cyclic-spectral (CS) transform and proposed the CS-SHAP by extending Shapley additive explanations (SHAP) to the CS domain. CS-SHAP can evaluate contributions from both carrier and modulation frequencies, aligning more closely with fault mechanisms and delivering clearer and more accurate explanations. Three datasets are utilized to validate the superior interpretability of CS-SHAP, ensuring its correctness, reproducibility, and practical performance. With open-source code and outstanding interpretability, CS-SHAP has the potential to be widely adopted and become the post-hoc interpretability benchmark in IFD, even in other classification tasks. The code is available on https://github.com/ChenQian0618/CS-SHAP.
Authors: Anna Tegon, Thorir Mar Ingolfsson, Xiaying Wang, Luca Benini, Yawei Li
Abstract: Accurate and efficient electroencephalography (EEG) analysis is essential for detecting seizures and artifacts in long-term monitoring, with applications spanning hospital diagnostics to wearable health devices. Robust EEG analytics have the potential to greatly improve patient care. However, traditional deep learning models, especially Transformer-based architectures, are hindered by their quadratic time and memory complexity, making them less suitable for resource-constrained environments. To address these challenges, we present FEMBA (Foundational EEG Mamba + Bidirectional Architecture), a novel self-supervised framework that establishes new efficiency benchmarks for EEG analysis through bidirectional state-space modeling. Unlike Transformer-based models, which incur quadratic time and memory complexity, FEMBA scales linearly with sequence length, enabling more scalable and efficient processing of extended EEG recordings. Trained on over 21,000 hours of unlabeled EEG and fine-tuned on three downstream tasks, FEMBA achieves competitive performance in comparison with transformer models, with significantly lower computational cost. Specifically, it reaches 81.82% balanced accuracy (0.8921 AUROC) on TUAB and 0.949 AUROC on TUAR, while a tiny 7.8M-parameter variant demonstrates viability for resource-constrained devices. These results pave the way for scalable, general-purpose EEG analytics in both clinical and highlight FEMBA as a promising candidate for wearable applications.
Authors: Elisabetta Cornacchia, Dan Mikulincer, Elchanan Mossel
Abstract: The problem of learning single index and multi index models has gained significant interest as a fundamental task in high-dimensional statistics. Many recent works have analysed gradient-based methods, particularly in the setting of isotropic data distributions, often in the context of neural network training. Such studies have uncovered precise characterisations of algorithmic sample complexity in terms of certain analytic properties of the target function, such as the leap, information, and generative exponents. These properties establish a quantitative separation between low and high complexity learning tasks. In this work, we show that high complexity cases are rare. Specifically, we prove that introducing a small random perturbation to the data distribution--via a random shift in the first moment--renders any Gaussian single index model as easy to learn as a linear function. We further extend this result to a class of multi index models, namely sparse Boolean functions, also known as Juntas.
Authors: Kaixuan Huang, Jiacheng Guo, Zihao Li, Xiang Ji, Jiawei Ge, Wenzhe Li, Yingqing Guo, Tianle Cai, Hui Yuan, Runzhe Wang, Yue Wu, Ming Yin, Shange Tang, Yangsibo Huang, Chi Jin, Xinyun Chen, Chiyuan Zhang, Mengdi Wang
Abstract: Large language models have demonstrated impressive performance on challenging mathematical reasoning tasks, which has triggered the discussion of whether the performance is achieved by true reasoning capability or memorization. To investigate this question, prior work has constructed mathematical benchmarks when questions undergo simple perturbations -- modifications that still preserve the underlying reasoning patterns of the solutions. However, no work has explored hard perturbations, which fundamentally change the nature of the problem so that the original solution steps do not apply. To bridge the gap, we construct MATH-P-Simple and MATH-P-Hard via simple perturbation and hard perturbation, respectively. Each consists of 279 perturbed math problems derived from level-5 (hardest) problems in the MATH dataset (Hendrycksmath et. al., 2021). We observe significant performance drops on MATH-P-Hard across various models, including o1-mini (-16.49%) and gemini-2.0-flash-thinking (-12.9%). We also raise concerns about a novel form of memorization where models blindly apply learned problem-solving skills without assessing their applicability to modified contexts. This issue is amplified when using original problems for in-context learning. We call for research efforts to address this challenge, which is critical for developing more robust and reliable reasoning models.
Authors: Victor Boone, Bruno Gaujal
Abstract: In average reward Markov decision processes, state-of-the-art algorithms for regret minimization follow a well-established framework: They are model-based, optimistic and episodic. First, they maintain a confidence region from which optimistic policies are computed using a well-known subroutine called Extended Value Iteration (EVI). Second, these policies are used over time windows called episodes, each ended by the Doubling Trick (DT) rule or a variant thereof. In this work, without modifying EVI, we show that there is a significant advantage in replacing (DT) by another simple rule, that we call the Vanishing Multiplicative (VM) rule. When managing episodes with (VM), the algorithm's regret is, both in theory and in practice, as good if not better than with (DT), while the one-shot behavior is greatly improved. More specifically, the management of bad episodes (when sub-optimal policies are being used) is much better under (VM) than (DT) by making the regret of exploration logarithmic rather than linear. These results are made possible by a new in-depth understanding of the contrasting behaviors of confidence regions during good and bad episodes.
Authors: Shenghong He
Abstract: Model-based offline reinforcement learning (MORL) aims to learn a policy by exploiting a dynamics model derived from an existing dataset. Applying conservative quantification to the dynamics model, most existing works on MORL generate trajectories that approximate the real data distribution to facilitate policy learning by using current information (e.g., the state and action at time step $t$). However, these works neglect the impact of historical information on environmental dynamics, leading to the generation of unreliable trajectories that may not align with the real data distribution. In this paper, we propose a new MORL algorithm \textbf{R}eliability-guaranteed \textbf{T}ransformer (RT), which can eliminate unreliable trajectories by calculating the cumulative reliability of the generated trajectory (i.e., using a weighted variational distance away from the real data). Moreover, by sampling candidate actions with high rewards, RT can efficiently generate high-return trajectories from the existing offline data. We theoretically prove the performance guarantees of RT in policy learning, and empirically demonstrate its effectiveness against state-of-the-art model-based methods on several benchmark tasks.
Authors: Soobin Um, Beomsu Kim, Jong Chul Ye
Abstract: Minority samples are underrepresented instances located in low-density regions of a data manifold, and are valuable in many generative AI applications, such as data augmentation, creative content generation, etc. Unfortunately, existing diffusion-based minority generators often rely on computationally expensive guidance dedicated for minority generation. To address this, here we present a simple yet powerful guidance-free approach called Boost-and-Skip for generating minority samples using diffusion models. The key advantage of our framework requires only two minimal changes to standard generative processes: (i) variance-boosted initialization and (ii) timestep skipping. We highlight that these seemingly-trivial modifications are supported by solid theoretical and empirical evidence, thereby effectively promoting emergence of underrepresented minority features. Our comprehensive experiments demonstrate that Boost-and-Skip greatly enhances the capability of generating minority samples, even rivaling guidance-based state-of-the-art approaches while requiring significantly fewer computations.
Authors: Thinh Nguyen, Cuong N. Nguyen, Quang Pham, Binh T. Nguyen, Savitha Ramasamy, Xiaoli Li, Cuong V. Nguyen
Abstract: In continual learning, understanding the properties of task sequences and their relationships to model performance is important for developing advanced algorithms with better accuracy. However, efforts in this direction remain underdeveloped despite encouraging progress in methodology development. In this work, we investigate the impacts of sequence transferability on continual learning and propose two novel measures that capture the total transferability of a task sequence, either in the forward or backward direction. Based on the empirical properties of these measures, we then develop a new method for the task order selection problem in continual learning. Our method can be shown to offer a better performance than the conventional strategy of random task selection.
Authors: Annie Marsden, Elad Hazan
Abstract: Previously, methods for learning marginally stable linear dynamical systems either required the transition matrix to be symmetric or incurred regret bounds that scale polynomially with the system's hidden dimension. In this work, we introduce a novel method that overcomes this trade-off, achieving dimension-free regret despite the presence of asymmetric matrices and marginal stability. Our method combines spectral filtering with linear predictors and employs Chebyshev polynomials in the complex plane to construct a novel spectral filtering basis. This construction guarantees sublinear regret in an online learning framework, without relying on any statistical or generative assumptions. Specifically, we prove that as long as the transition matrix has eigenvalues with complex component bounded by $1/\mathrm{poly} \log T$, then our method achieves regret $\tilde{O}(T^{9/10})$ when compared to the best linear dynamical predictor in hindsight.
Authors: Marika Swanberg, Ryan McKenna, Edo Roth, Albert Cheu, Peter Kairouz
Abstract: Differentially private (DP) synthetic data is a versatile tool for enabling the analysis of private data. Recent advancements in large language models (LLMs) have inspired a number of algorithm techniques for improving DP synthetic data generation. One family of approaches uses DP finetuning on the foundation model weights; however, the model weights for state-of-the-art models may not be public. In this work we propose two DP synthetic tabular data algorithms that only require API access to the foundation model. We adapt the Private Evolution algorithm (Lin et al., 2023; Xie et al., 2024) -- which was designed for image and text data -- to the tabular data domain. In our extension of Private Evolution, we define a query workload-based distance measure, which may be of independent interest. We propose a family of algorithms that use one-shot API access to LLMs, rather than adaptive queries to the LLM. Our findings reveal that API-access to powerful LLMs does not always improve the quality of DP synthetic data compared to established baselines that operate without such access. We provide insights into the underlying reasons and propose improvements to LLMs that could make them more effective for this application.
Authors: Francisco N. F. Q. Simoes, Itai Feigenbaum, Mehdi Dastani, Thijs van Ommen
Abstract: Causal knowledge can be used to support decision-making problems. This has been recognized in the causal bandits literature, where a causal (multi-armed) bandit is characterized by a causal graphical model and a target variable. The arms are then interventions on the causal model, and rewards are samples of the target variable. Causal bandits were originally studied with a focus on hard interventions. We focus instead on cases where the arms are conditional interventions, which more accurately model many real-world decision-making problems by allowing the value of the intervened variable to be chosen based on the observed values of other variables. This paper presents a graphical characterization of the minimal set of nodes guaranteed to contain the optimal conditional intervention, which maximizes the expected reward. We then propose an efficient algorithm with a time complexity of $O(|V| + |E|)$ to identify this minimal set of nodes. We prove that the graphical characterization and the proposed algorithm are correct. Finally, we empirically demonstrate that our algorithm significantly prunes the search space and substantially accelerates convergence rates when integrated into standard multi-armed bandit algorithms.
Authors: Aur\'elien Renault, Alexis Bondu, Antoine Cornu\'ejols, Vincent Lemaire
Abstract: Early Classification of Time Series (ECTS) has been recognized as an important problem in many areas where decisions have to be taken as soon as possible, before the full data availability, while time pressure increases. Numerous ECTS approaches have been proposed, based on different triggering functions, each taking into account various pieces of information related to the incoming time series and/or the output of a classifier. Although their performances have been empirically compared in the literature, no studies have been carried out on the optimality of these triggering functions that involve ``man-tailored'' decision rules. Based on the same information, could there be better triggering functions? This paper presents one way to investigate this question by showing first how to translate ECTS problems into Reinforcement Learning (RL) ones, where the very same information is used in the state space. A thorough comparison of the performance obtained by ``handmade'' approaches and their ``RL-based'' counterparts has been carried out. A second question investigated in this paper is whether a different combination of information, defining the state space in RL systems, can achieve even better performance. Experiments show that the system we describe, called \textsc{Alert}, significantly outperforms its state-of-the-art competitors on a large number of datasets.
Authors: Ron Shapira Weber, Oren Freifeld
Abstract: In time-series analysis, nonlinear temporal misalignment remains a pivotal challenge that forestalls even simple averaging. Since its introduction, the Diffeomorphic Temporal Alignment Net (DTAN), which we first introduced (Weber et al., 2019) and further developed in (Weber & Freifeld, 2023), has proven itself as an effective solution for this problem (these conference papers are earlier partial versions of the current manuscript). DTAN predicts and applies diffeomorphic transformations in an input-dependent manner, thus facilitating the joint alignment (JA) and averaging of time-series ensembles in an unsupervised or a weakly-supervised manner. The inherent challenges of the weakly/unsupervised setting, particularly the risk of trivial solutions through excessive signal distortion, are mitigated using either one of two distinct strategies: 1) a regularization term for warps; 2) using the Inverse Consistency Averaging Error (ICAE). The latter is a novel, regularization-free approach which also facilitates the JA of variable-length signals. We also further extend our framework to incorporate multi-task learning (MT-DTAN), enabling simultaneous time-series alignment and classification. Additionally, we conduct a comprehensive evaluation of different backbone architectures, demonstrating their efficacy in time-series alignment tasks. Finally, we showcase the utility of our approach in enabling Principal Component Analysis (PCA) for misaligned time-series data. Extensive experiments across 128 UCR datasets validate the superiority of our approach over contemporary averaging methods, including both traditional and learning-based approaches, marking a significant advancement in the field of time-series analysis.
Authors: Nikita P. Kalinin, Jalaj Upadhyay, Christoph H. Lampert
Abstract: We propose Joint Moment Estimation (JME), a method for continually and privately estimating both the first and second moments of data with reduced noise compared to naive approaches. JME uses the matrix mechanism and a joint sensitivity analysis to allow the second moment estimation with no additional privacy cost, thereby improving accuracy while maintaining privacy. We demonstrate JME's effectiveness in two applications: estimating the running mean and covariance matrix for Gaussian density estimation, and model training with DP-Adam on CIFAR-10.
Authors: Sarthak Mittal, Niels Leif Bracher, Guillaume Lajoie, Priyank Jaini, Marcus Brubaker
Abstract: Bayesian inference provides a natural way of incorporating prior beliefs and assigning a probability measure to the space of hypotheses. Current solutions rely on iterative routines like Markov Chain Monte Carlo (MCMC) sampling and Variational Inference (VI), which need to be re-run whenever new observations are available. Amortization, through conditional estimation, is a viable strategy to alleviate such difficulties and has been the guiding principle behind simulation-based inference, neural processes and in-context methods using pre-trained models. In this work, we conduct a thorough comparative analysis of amortized in-context Bayesian posterior estimation methods from the lens of different optimization objectives and architectural choices. Such methods train an amortized estimator to perform posterior parameter inference by conditioning on a set of data examples passed as context to a sequence model such as a transformer. In contrast to language models, we leverage permutation invariant architectures as the true posterior is invariant to the ordering of context examples. Our empirical study includes generalization to out-of-distribution tasks, cases where the assumed underlying model is misspecified, and transfer from simulated to real problems. Subsequently, it highlights the superiority of the reverse KL estimator for predictive problems, especially when combined with the transformer architecture and normalizing flows.
Authors: Zhiqiang Zhong, Simon Sataa-Yu Larsen, Haoyu Guo, Tao Tang, Kuangyu Zhou, Davide Mottin
Abstract: Recent advancements in AI for biological research focus on integrating molecular data with natural language to accelerate drug discovery. However, the scarcity of high-quality annotations limits progress in this area. This paper introduces LA$^3$, a Language-based Automatic Annotation Augmentation framework that leverages large language models to augment existing datasets, thereby improving AI training. We demonstrate the effectiveness of LA$^3$ by creating an enhanced dataset, LaChEBI-20, where we systematically rewrite the annotations of molecules from an established dataset. These rewritten annotations preserve essential molecular information while providing more varied sentence structures and vocabulary. Using LaChEBI-20, we train LaMolT5 based on a benchmark architecture to learn the mapping between molecular representations and augmented annotations. Experimental results on text-based *de novo* molecule generation and molecule captioning demonstrate that LaMolT5 outperforms state-of-the-art models. Notably, incorporating LA$^3$ leads to improvements of up to 301% over the benchmark architecture. Furthermore, we validate the effectiveness of LA$^3$ notable applications in *image*, *text* and *graph* tasks, affirming its versatility and utility.
Authors: Seokjin Go, Divya Mahajan
Abstract: Mixture-of-Experts (MoE) model architecture has emerged as a promising solution for scaling transformer models efficiently, offering sparse activation that reduces computational costs while increasing model capacity. However, as MoE models scale, they need to be distributed across GPU devices, thus face critical performance bottlenecks due to their large memory footprint. Expert parallelism distributes experts across GPUs, however, faces key challenges including an unbalanced token routing and expert activation, resulting in communication tail latency and processing inefficiencies. While existing solutions address some of these issues, they fail to resolve the dual challenges of load imbalance and communication skew. The imbalance in token processing load across experts causes uneven processing times on different GPUs, while communication skew between GPUs leads to unbalanced inter-GPU data transfers. These factors degrade the performance of MoE models by increasing tail latency and reducing overall throughput. To address these limitations, we propose an Integer Linear Programming (ILP) formulation to optimize expert placement by jointly considering token load, communication, and computation costs. We exploit the property that there is a token routing dependency across layers, where tokens routed to a specific expert in one layer are likely to be routed to a limited set of experts in the subsequent layer. Our solution, MoETuner, offers an optimal expert-to-GPU assignment that minimizes inter-GPU token routing costs and balances token processing across devices, thereby reducing tail latency and end-to-end execution time. Experimental results demonstrate 9.3% and 17.5% of end-to-end speedups for single-node and multi-node inference respectively, showcasing the potential of our ILP-based optimization for offering expert parallel solutions for next-generation MoEs.
Authors: Petar Bevanda, Max Beier, Armin Lederer, Alexandre Capone, Stefan Sosnowski, Sandra Hirche
Abstract: Credible forecasting and representation learning of dynamical systems are of ever-increasing importance for reliable decision-making. To that end, we propose a family of Gaussian processes (GP) for dynamical systems with linear time-invariant responses, which are nonlinear only in initial conditions. This linearity allows us to tractably quantify forecasting and representational uncertainty, simultaneously alleviating the challenge of computing the distribution of trajectories from a GP-based dynamical system and enabling a new probabilistic treatment of learning Koopman operator representations. Using a trajectory-based equivariance -- which we refer to as \textit{Koopman equivariance} -- we obtain a GP model with enhanced generalization capabilities. To allow for large-scale regression, we equip our framework with variational inference based on suitable inducing points. Experiments demonstrate on-par and often better forecasting performance compared to kernel-based methods for learning dynamical systems.
Authors: E. Mehmet K{\i}ral, Nur\c{s}en Ayd{\i}n, \c{S}. \.Ilker Birbil
Abstract: There is a growing need for investigating how machine learning models operate. With this work, we aim to understand trained machine learning models by questioning their data preferences. We propose a mathematical framework that allows us to probe trained models and identify their preferred samples in various scenarios including prediction-risky, parameter-sensitive, or model-contrastive samples. To showcase our framework, we pose these queries to a range of models trained on a range of classification and regression tasks, and receive answers in the form of generated data.
Authors: Xingrun Xing, Zheng Liu, Shitao Xiao, Boyan Gao, Yiming Liang, Wanpeng Zhang, Haokun Lin, Guoqi Li, Jiajun Zhang
Abstract: Modern large language models (LLMs) driven by scaling laws, achieve intelligence emergency in large model sizes. Recently, the increasing concerns about cloud costs, latency, and privacy make it an urgent requirement to develop compact edge language models. Distinguished from direct pretraining that bounded by the scaling law, this work proposes the pruning-aware pretraining, focusing on retaining performance of much larger optimized models. It features following characteristics: 1) Data-scalable: we introduce minimal parameter groups in LLM and continuously optimize structural pruning, extending post-training pruning methods like LLM-Pruner and SparseGPT into the pretraining phase. 2) Architecture-agnostic: the LLM architecture is auto-designed using saliency-driven pruning, which is the first time to exceed SoTA human-designed LLMs in modern pretraining. We reveal that it achieves top-quality edge language models, termed EfficientLLM, by scaling up LLM compression and extending its boundary. EfficientLLM significantly outperforms SoTA baselines with $100M \sim 1B$ parameters, such as MobileLLM, SmolLM, Qwen2.5-0.5B, OLMo-1B, Llama3.2-1B in common sense benchmarks. As the first attempt, EfficientLLM bridges the performance gap between traditional LLM compression and direct pretraining methods, and we will fully open source at https://github.com/Xingrun-Xing2/EfficientLLM.
Authors: Michael Arbel, David Salinas, Frank Hutter
Abstract: Recent foundational models for tabular data, such as TabPFN, have demonstrated remarkable effectiveness in adapting to new tasks through in-context learning. However, these models overlook a crucial equivariance property: the arbitrary ordering of target dimensions should not influence model predictions. In this study, we identify this oversight as a source of incompressible error, termed the equivariance gap, which introduces instability in predictions. To mitigate these issues, we propose a novel model designed to preserve equivariance across output dimensions. Our experimental results indicate that our proposed model not only addresses these pitfalls effectively but also achieves competitive benchmark performance.
Authors: Jiajun He, Yuanqi Du, Francisco Vargas, Dinghuai Zhang, Shreyas Padhy, RuiKang OuYang, Carla Gomes, Jos\'e Miguel Hern\'andez-Lobato
Abstract: We consider the sampling problem, where the aim is to draw samples from a distribution whose density is known only up to a normalization constant. Recent breakthroughs in generative modeling to approximate a high-dimensional data distribution have sparked significant interest in developing neural network-based methods for this challenging problem. However, neural samplers typically incur heavy computational overhead due to simulating trajectories during training. This motivates the pursuit of simulation-free training procedures of neural samplers. In this work, we propose an elegant modification to previous methods, which allows simulation-free training with the help of a time-dependent normalizing flow. However, it ultimately suffers from severe mode collapse. On closer inspection, we find that nearly all successful neural samplers rely on Langevin preconditioning to avoid mode collapsing. We systematically analyze several popular methods with various objective functions and demonstrate that, in the absence of Langevin preconditioning, most of them fail to adequately cover even a simple target. Finally, we draw attention to a strong baseline by combining the state-of-the-art MCMC method, Parallel Tempering (PT), with an additional generative model to shed light on future explorations of neural samplers.
Authors: Amin Adibi, Xu Cao, Zongliang Ji, Jivat Neet Kaur, Winston Chen, Elizabeth Healey, Brighton Nuwagira, Wenqian Ye, Geoffrey Woollard, Maxwell A Xu, Hejie Cui, Johnny Xi, Trenton Chang, Vasiliki Bikia, Nicole Zhang, Ayush Noori, Yuan Xia, Md. Belal Hossain, Hanna A. Frank, Alina Peluso, Yuan Pu, Shannon Zejiang Shen, John Wu, Adibvafa Fallahpour, Sazan Mahbub, Ross Duncan, Yuwei Zhang, Yurui Cao, Zuheng Xu, Michael Craig, Rahul G. Krishnan, Rahmatollah Beheshti, James M. Rehg, Mohammad Ehsanul Karim, Megan Coffee, Leo Anthony Celi, Jason Alan Fries, Mohsen Sadatsafavi, Dennis Shung, Shannon McWeeney, Jessica Dafflon, Sarah Jabbour
Abstract: The fourth Machine Learning for Health (ML4H) symposium was held in person on December 15th and 16th, 2024, in the traditional, ancestral, and unceded territories of the Musqueam, Squamish, and Tsleil-Waututh Nations in Vancouver, British Columbia, Canada. The symposium included research roundtable sessions to foster discussions between participants and senior researchers on timely and relevant topics for the ML4H community. The organization of the research roundtables at the conference involved 13 senior and 27 junior chairs across 13 tables. Each roundtable session included an invited senior chair (with substantial experience in the field), junior chairs (responsible for facilitating the discussion), and attendees from diverse backgrounds with an interest in the session's topic.
Authors: Geraldin Nanfack, Eugene Belilovsky
Abstract: Deep learning models frequently exploit spurious features in training data to achieve low training error, often resulting in poor generalization when faced with shifted testing distributions. To address this issue, various methods from imbalanced learning, representation learning, and classifier recalibration have been proposed to enhance the robustness of deep neural networks against spurious correlations. In this paper, we observe that models trained with empirical risk minimization tend to generalize well for examples from the majority groups while memorizing instances from minority groups. Building on recent findings that show memorization can be localized to a limited number of neurons, we apply example-tied dropout as a method we term FairDropout, aimed at redirecting this memorization to specific neurons that we subsequently drop out during inference. We empirically evaluate FairDropout using the subpopulation benchmark suite encompassing vision, language, and healthcare tasks, demonstrating that it significantly reduces reliance on spurious correlations, and outperforms state-of-the-art methods.
Authors: Amirhossein Dadashzadeh Taromi, Sina Heydari, Mohsen Hooshmand, Majid Ramezani
Abstract: Recommender systems play a crucial role in modern life, including information retrieval, the pharmaceutical industry, retail, and entertainment. The entertainment sector, in particular, attracts significant attention and generates substantial profits. This work proposes a new method for predicting unknown user-movie ratings to enhance customer satisfaction. To achieve this, we utilize the MovieLens 100K dataset. Our approach introduces an attention-based autoencoder to create meaningful representations and the XGBoost method for rating predictions. The results demonstrate that our proposal outperforms most of the existing state-of-the-art methods. Availability: github.com/ComputationIASBS/RecommSys
Authors: Mogens Henrik From, Jacob Nielsen, Lukas Galke, Peter Schneider-Kamp
Abstract: Training large neural network models requires extensive computational resources, often distributed across several nodes and accelerators. Recent findings suggest that it may be sufficient to only exchange the fast moving components of the gradients, while accumulating momentum locally (Decoupled Momentum, or DeMo). However, when considering larger models that do not fit on a single accelerate, the exchange of gradient information and the integration of DeMo needs to be reconsidered. Here, we propose employing a hybrid strategy, FlexDeMo, whereby nodes fully synchronize locally between different GPUs and inter-node communication is improved through only using the fast-moving components. This effectively combines previous hybrid sharding strategies with the advantages of decoupled momentum. Our experimental results show that FlexDeMo is on par with AdamW in terms of validation loss, demonstrating its viability.
Authors: Daouda Sow, Herbert Woisetschl\"ager, Saikiran Bulusu, Shiqiang Wang, Hans-Arno Jacobsen, Yingbin Liang
Abstract: Pretraining large language models (LLMs) on vast and heterogeneous datasets is crucial for achieving state-of-the-art performance across diverse downstream tasks. However, current training paradigms treat all samples equally, overlooking the importance or relevance of individual samples throughout the training process. Existing reweighting strategies, which primarily focus on group-level data importance, fail to leverage fine-grained instance-level information and do not adapt dynamically to individual sample importance as training progresses. In this paper, we introduce novel algorithms for dynamic, instance-level data reweighting aimed at improving both the efficiency and effectiveness of LLM pretraining. Our methods adjust the weight of each training sample based on its loss value in an online fashion, allowing the model to dynamically focus on more informative or important samples at the current training stage. In particular, our framework allows us to systematically devise reweighting strategies deprioritizing redundant or uninformative data, which we find tend to work best. Furthermore, we develop a new theoretical framework for analyzing the impact of loss-based reweighting on the convergence of gradient-based optimization, providing the first formal characterization of how these strategies affect convergence bounds. We empirically validate our approach across a spectrum of tasks, from pretraining 7B and 1.4B parameter LLMs to smaller-scale language models and linear regression problems, demonstrating that our loss-based reweighting approach can lead to faster convergence and significantly improved performance.
Authors: Thomas Zeng, Shuibai Zhang, Shutong Wu, Christian Classen, Daewon Chae, Ethan Ewer, Minjae Lee, Heeju Kim, Wonjun Kang, Jackson Kunde, Ying Fan, Jungtaek Kim, Hyung Il Koo, Kannan Ramchandran, Dimitris Papailiopoulos, Kangwook Lee
Abstract: Process Reward Models (PRMs) have proven effective at enhancing mathematical reasoning for Large Language Models (LLMs) by leveraging increased inference-time computation. However, they are predominantly trained on mathematical data and their generalizability to non-mathematical domains has not been rigorously studied. In response, this work first shows that current PRMs have poor performance in other domains. To address this limitation, we introduce VersaPRM, a multi-domain PRM trained on synthetic reasoning data generated using our novel data generation and annotation method. VersaPRM achieves consistent performance gains across diverse domains. For instance, in the MMLU-Pro category of Law, VersaPRM via weighted majority voting, achieves a 7.9% performance gain over the majority voting baseline -- surpassing Qwen2.5-Math-PRM's gain of 1.3%. We further contribute to the community by open-sourcing all data, code and models for VersaPRM.
Authors: Igor Ivanov, Dmitrii Volkov
Abstract: Recent work showed that small changes in benchmark questions can reduce LLMs' reasoning and recall. We explore two such changes: pairing questions and adding more answer options, on three benchmarks: WMDP-bio, GPQA, and MMLU variants. We find that for more capable models, these predictably reduce performance, essentially heightening the performance ceiling of a benchmark and unsaturating it again. We suggest this approach can resurrect old benchmarks.
Authors: Sauro Succi
Abstract: We highlight a formal and substantial analogy between Machine Learning (ML) algorithms and discrete dynamical systems (DDS) in relaxation form. The analogy offers a transparent interpretation of the weights in terms of physical information-propagation processes and identifies the model function of the forward ML step with the local attractor of the corresponding discrete dynamics. Besides improving the explainability of current ML applications, this analogy may also facilitate the development of a new class ML algorithms with a reduced number of weights.
Authors: Meyer Scetbon, Chao Ma, Wenbo Gong, Edward Meeds
Abstract: Training large language models (LLMs) typically relies on adaptive optimizers like Adam (Kingma & Ba, 2015) which store additional state information to accelerate convergence but incur significant memory overhead. Recent efforts, such as SWAN (Ma et al., 2024) address this by eliminating the need for optimizer states while achieving performance comparable to Adam via a multi-step preprocessing procedure applied to instantaneous gradients. Motivated by the success of SWAN, we introduce a novel framework for designing stateless optimizers that normalizes stochastic gradients according to multiple norms. To achieve this, we propose a simple alternating scheme to enforce the normalization of gradients w.r.t these norms. We show that our procedure can produce, up to an arbitrary precision, a fixed-point of the problem, and that SWAN is a particular instance of our approach with carefully chosen norms, providing a deeper understanding of its design. However, SWAN's computationally expensive whitening/orthogonalization step limit its practicality for large LMs. Using our principled perspective, we develop of a more efficient, scalable, and practical stateless optimizer. Our algorithm relaxes the properties of SWAN, significantly reducing its computational cost while retaining its memory efficiency, making it applicable to training large-scale models. Experiments on pre-training LLaMA models with up to 1 billion parameters demonstrate a 3X speedup over Adam with significantly reduced memory requirements, outperforming other memory-efficient baselines.
Authors: Alex Vitvitskyi, Jo\~ao G. M. Ara\'ujo, Marc Lackenby, Petar Veli\v{c}kovi\'c
Abstract: As implied by the plethora of literature on graph rewiring, the choice of computational graph employed by a neural network can make a significant impact on its downstream performance. Certain effects related to the computational graph, such as under-reaching and over-squashing, may even render the model incapable of learning certain functions. Most of these effects have only been thoroughly studied in the domain of undirected graphs; however, recent years have seen a significant rise in interest in feedforward computational graphs: directed graphs without any back edges. In this paper, we study the desirable properties of a feedforward computational graph, discovering two important complementary measures: fidelity and mixing time, and evaluating a few popular choices of graphs through the lens of these measures. Our study is backed by both theoretical analyses of the metrics' asymptotic behaviour for various graphs, as well as correlating these metrics to the performance of trained neural network models using the corresponding graphs.
Authors: Niccol\`o Ajroldi, Antonio Orvieto, Jonas Geiping
Abstract: Averaging checkpoints along the training trajectory is a simple yet powerful approach to improve the generalization performance of Machine Learning models and reduce training time. Motivated by these potential gains, and in an effort to fairly and thoroughly benchmark this technique, we present an extensive evaluation of averaging techniques in modern Deep Learning, which we perform using AlgoPerf \citep{dahl_benchmarking_2023}, a large-scale benchmark for optimization algorithms. We investigate whether weight averaging can reduce training time, improve generalization, and replace learning rate decay, as suggested by recent literature. Our evaluation across seven architectures and datasets reveals that averaging significantly accelerates training and yields considerable efficiency gains, at the price of a minimal implementation and memory cost, while mildly improving generalization across all considered workloads. Finally, we explore the relationship between averaging and learning rate annealing and show how to optimally combine the two to achieve the best performances.
Authors: Kiwhan Song, Boyuan Chen, Max Simchowitz, Yilun Du, Russ Tedrake, Vincent Sitzmann
Abstract: Classifier-free guidance (CFG) is a key technique for improving conditional generation in diffusion models, enabling more accurate control while enhancing sample quality. It is natural to extend this technique to video diffusion, which generates video conditioned on a variable number of context frames, collectively referred to as history. However, we find two key challenges to guiding with variable-length history: architectures that only support fixed-size conditioning, and the empirical observation that CFG-style history dropout performs poorly. To address this, we propose the Diffusion Forcing Transformer (DFoT), a video diffusion architecture and theoretically grounded training objective that jointly enable conditioning on a flexible number of history frames. We then introduce History Guidance, a family of guidance methods uniquely enabled by DFoT. We show that its simplest form, vanilla history guidance, already significantly improves video generation quality and temporal consistency. A more advanced method, history guidance across time and frequency further enhances motion dynamics, enables compositional generalization to out-of-distribution history, and can stably roll out extremely long videos. Website: https://boyuan.space/history-guidance
Authors: Jaeyeon Kim, Kulin Shah, Vasilis Kontonis, Sham Kakade, Sitan Chen
Abstract: In recent years, masked diffusion models (MDMs) have emerged as a promising alternative approach for generative modeling over discrete domains. Compared to autoregressive models (ARMs), MDMs trade off complexity at training time with flexibility at inference time. At training time, they must learn to solve an exponentially large number of infilling problems, but at inference time, they can decode tokens in essentially arbitrary order. In this work, we closely examine these two competing effects. On the training front, we theoretically and empirically demonstrate that MDMs indeed train on computationally intractable subproblems compared to their autoregressive counterparts. On the inference front, we show that a suitable strategy for adaptively choosing the token decoding order significantly enhances the capabilities of MDMs, allowing them to sidestep hard subproblems. On logic puzzles like Sudoku, we show that adaptive inference can boost solving accuracy in pretrained MDMs from $<7$% to $\approx 90$%, even outperforming ARMs with $7\times$ as many parameters and that were explicitly trained via teacher forcing to learn the right order of decoding.
Authors: Giacomo Lastrucci, Artur M. Schweidtmann
Abstract: Ensuring neural networks adhere to domain-specific constraints is crucial for addressing safety and ethical concerns while also enhancing prediction accuracy. Despite the nonlinear nature of most real-world tasks, existing methods are predominantly limited to affine or convex constraints. We introduce ENFORCE, a neural network architecture that guarantees predictions to satisfy nonlinear constraints exactly. ENFORCE is trained with standard unconstrained gradient-based optimizers (e.g., Adam) and leverages autodifferentiation and local neural projections to enforce any $\mathcal{C}^1$ constraint to arbitrary tolerance $\epsilon$. We build an adaptive-depth neural projection (AdaNP) module that dynamically adjusts its complexity to suit the specific problem and the required tolerance levels. ENFORCE guarantees satisfaction of equality constraints that are nonlinear in both inputs and outputs of the neural network with minimal (and adjustable) computational cost.
Authors: Geyu Liang, Senne Michielssen, Salar Fattahi
Abstract: The trade-off between accuracy and interpretability has long been a challenge in machine learning (ML). This tension is particularly significant for emerging interpretable-by-design methods, which aim to redesign ML algorithms for trustworthy interpretability but often sacrifice accuracy in the process. In this paper, we address this gap by investigating the impact of deviations in concept representations-an essential component of interpretable models-on prediction performance and propose a novel framework to mitigate these effects. The framework builds on the principle of optimizing concept embeddings under constraints that preserve interpretability. Using a generative model as a test-bed, we rigorously prove that our algorithm achieves zero loss while progressively enhancing the interpretability of the resulting model. Additionally, we evaluate the practical performance of our proposed framework in generating explainable predictions for image classification tasks across various benchmarks. Compared to existing explainable methods, our approach not only improves prediction accuracy while preserving model interpretability across various large-scale benchmarks but also achieves this with significantly lower computational cost.
Authors: Brandon Trabucco, Gunnar Sigurdsson, Robinson Piramuthu, Ruslan Salakhutdinov
Abstract: The predominant approach for training web navigation agents gathers human demonstrations for a set of popular websites and hand-written tasks, but it is becoming clear that human data are an inefficient resource. We develop a pipeline to facilitate Internet-scale training for agents without laborious human annotations. In the first stage, an LLM generates tasks for 150k diverse websites. In the next stage, LLM agents complete tasks and produce trajectories. In the final stage, an LLM reviews the trajectories and judges their success. Language models are competitive with human annotators, detecting and filtering out harmful content with an accuracy of 97%, generating feasible tasks with an 89% rate, and judging successful trajectories with an 82.6% accuracy. Scaling the pipeline, agents based on Llama 3.1 70B solve 16.7% of tasks for 150k sites. Training on the data generated by our pipeline is competitive with training on human demonstrations. In data-limited settings derived from Mind2Web and WebLINX, we improve Step Accuracy by up to +89.5% and +122.1% respectively for agents trained on mixtures of data from our pipeline, and human data. When training agents with all available human data from these benchmarks, agents fail to generalize to diverse real sites, and adding our data improves their generalization by +149.0% for WebLINX and +156.3% for Mind2Web. Code will be available at: data-for-agents.github.io.
Authors: Tianlang Chen, Charilaos Kanatsoulis, Jure Leskovec
Abstract: Predictive tasks on relational databases are critical in real-world applications spanning e-commerce, healthcare, and social media. To address these tasks effectively, Relational Deep Learning (RDL) encodes relational data as graphs, enabling Graph Neural Networks (GNNs) to exploit relational structures for improved predictions. However, existing heterogeneous GNNs often overlook the intrinsic structural properties of relational databases, leading to modeling inefficiencies. Here we introduce RelGNN, a novel GNN framework specifically designed to capture the unique characteristics of relational databases. At the core of our approach is the introduction of atomic routes, which are sequences of nodes forming high-order tripartite structures. Building upon these atomic routes, RelGNN designs new composite message passing mechanisms between heterogeneous nodes, allowing direct single-hop interactions between them. This approach avoids redundant aggregations and mitigates information entanglement, ultimately leading to more efficient and accurate predictive modeling. RelGNN is evaluated on 30 diverse real-world tasks from RelBench (Fey et al., 2024), and consistently achieves state-of-the-art accuracy with up to 25% improvement.
Authors: Mike Heddes, Adel Javanmard, Kyriakos Axiotis, Gang Fu, MohammadHossein Bateni, Vahab Mirrokni
Abstract: Transformer networks have achieved remarkable success across diverse domains, leveraging a variety of architectural innovations, including residual connections. However, traditional residual connections, which simply sum the outputs of previous layers, can dilute crucial information. This work introduces DeepCrossAttention (DCA), an approach that enhances residual learning in transformers. DCA employs learnable, input-dependent weights to dynamically combine layer outputs, enabling the model to selectively focus on the most relevant information in any of the previous layers. Furthermore, DCA incorporates depth-wise cross-attention, allowing for richer interactions between layers at different depths. Our language modeling experiments show that DCA achieves improved perplexity for a given training time. Moreover, DCA obtains the same model quality up to 3x faster while adding a negligible number of parameters. Theoretical analysis confirms that DCA provides an improved trade-off between accuracy and model size when the ratio of collective layer ranks to the ambient dimension falls below a critical threshold.
Authors: Pranav Nair, Puranjay Datta, Jeff Dean, Prateek Jain, Aditya Kusupati
Abstract: Quantizing model weights is critical for reducing the communication and inference costs of large models. However, quantizing models -- especially to low precisions like int4 or int2 -- requires a trade-off in model quality; int2, in particular, is known to severely degrade model quality. Consequently, practitioners are often forced to maintain multiple models with different quantization levels or serve a single model that best satisfies the quality-latency trade-off. On the other hand, integer data types, such as int8, inherently possess a nested (Matryoshka) structure where smaller bit-width integers, like int4 or int2, are nested within the most significant bits. This paper proposes Matryoshka Quantization (MatQuant), a novel multi-scale quantization technique that addresses the challenge of needing multiple quantized models. It allows training and maintaining just one model, which can then be served at different precision levels. Furthermore, due to the co-training and co-distillation regularization provided by MatQuant, the int2 precision models extracted by MatQuant can be up to $10\%$ more accurate than standard int2 quantization (using techniques like QAT or OmniQuant). This represents significant progress in model quantization, demonstrated by the fact that, with the same recipe, an int2 FFN-quantized Gemma-2 9B model is more accurate than an int8 FFN-quantized Gemma-2 2B model.
Authors: Xueru Wen, Jie Lou, Xinyu Lu, Junjie Yang, Yanjiang Liu, Yaojie Lu, Debing Zhang, XingYu
Abstract: As AI capabilities increasingly surpass human proficiency in complex tasks, current alignment techniques including SFT and RLHF face fundamental challenges in ensuring reliable oversight. These methods rely on direct human assessment and become untenable when AI outputs exceed human cognitive thresholds. In response to this challenge, we explore two hypotheses: (1) critique of critique can be easier than critique itself, extending the widely-accepted observation that verification is easier than generation to the critique domain, as critique itself is a specialized form of generation; (2) this difficulty relationship is recursively held, suggesting that when direct evaluation is infeasible, performing high-order critiques (e.g., critique of critique of critique) offers a more tractable supervision pathway. To examine these hypotheses, we perform Human-Human, Human-AI, and AI-AI experiments across multiple tasks. Our results demonstrate encouraging evidence supporting these hypotheses and suggest that recursive self-critiquing is a promising direction for scalable oversight.
Authors: Johnny Chan, Yuming Li
Abstract: This research-in-progress paper presents a new project management framework that utilises GenAI technology. The framework is designed to address the common challenge of uniform team compositions in academic and research project teams, particularly in universities and research institutions. It does so by integrating sociologically identified patterns of successful team member personalities and roles, using GenAI agents to fill gaps in team dynamics. This approach adds an additional layer of analysis to conventional project management processes by evaluating team members' personalities and roles and employing GenAI agents, fine-tuned on personality datasets, to fill specific team roles. Our initial experiments have shown improvements in the model's ability to understand and process personality traits, suggesting the potential effectiveness of GenAI teammates in real-world project settings. This paper aims to explore the practical application of AI in enhancing team diversity and project management
Authors: Furkan Karada\c{s}, Bahaeddin Eravc{\i}, Ahmet Murat \"Ozbayo\u{g}lu
Abstract: In an era where financial markets are heavily influenced by many static and dynamic factors, it has become increasingly critical to carefully integrate diverse data sources with machine learning for accurate stock price prediction. This paper explores a multimodal machine learning approach for stock price prediction by combining data from diverse sources, including traditional financial metrics, tweets, and news articles. We capture real-time market dynamics and investor mood through sentiment analysis on these textual data using both ChatGPT-4o and FinBERT models. We look at how these integrated data streams augment predictions made with a standard Long Short-Term Memory (LSTM model) to illustrate the extent of performance gains. Our study's results indicate that incorporating the mentioned data sources considerably increases the forecast effectiveness of the reference model by up to 5%. We also provide insights into the individual and combined predictive capacities of these modalities, highlighting the substantial impact of incorporating sentiment analysis from tweets and news articles. This research offers a systematic and effective framework for applying multimodal data analytics techniques in financial time series forecasting that provides a new view for investors to leverage data for decision-making.
Authors: Zhijian Duan, Yusen Huo, Tianyu Wang, Zhilin Zhang, Yeshu Li, Chuan Yu, Jian Xu, Bo Zheng, Xiaotie Deng
Abstract: In online advertising, advertisers commonly utilize auto-bidding services to bid for impression opportunities. A typical objective of the auto-bidder is to optimize the advertiser's cumulative value of winning impressions within specified budget constraints. However, such a problem is challenging due to the complex bidding environment faced by diverse advertisers. To address this challenge, we introduce ABPlanner, a few-shot adaptable budget planner designed to improve budget-constrained auto-bidding. ABPlanner is based on a hierarchical bidding framework that decomposes the bidding process into shorter, manageable stages. Within this framework, ABPlanner allocates the budget across all stages, allowing a low-level auto-bidder to bids based on the budget allocation plan. The adaptability of ABPlanner is achieved through a sequential decision-making approach, inspired by in-context reinforcement learning. For each advertiser, ABPlanner adjusts the budget allocation plan episode by episode, using data from previous episodes as prompt for current decisions. This enables ABPlanner to quickly adapt to different advertisers with few-shot data, providing a sample-efficient solution. Extensive simulation experiments and real-world A/B testing validate the effectiveness of ABPlanner, demonstrating its capability to enhance the cumulative value achieved by auto-bidders.
Authors: Jing Sun, Tiexing Wang, Eric Verschuur, Ivan Vasconcelos
Abstract: In recent years, deep learning (DL) has emerged as a promising alternative approach for various seismic processing tasks, including primary estimation (or multiple elimination), a crucial step for accurate subsurface imaging. In geophysics, DL methods are commonly based on supervised learning from large amounts of high-quality labelled data. Instead of relying on traditional supervised learning, in the context of free-surface multiple elimination, we propose a method in which the DL model learns to effectively parameterize the free-surface multiple-free wavefield from the full wavefield by incorporating the underlying physics into the loss computation. This, in turn, yields high-quality estimates without ever being shown any ground truth data. Currently, the network reparameterization is performed independently for each dataset. We demonstrate its effectiveness through tests on both synthetic and field data. We employ industry-standard Surface-Related Multiple Elimination (SRME) using, respectively, global least-squares adaptive subtraction and local least-squares adaptive subtraction as benchmarks. The comparison shows that the proposed method outperforms the benchmarks in estimation accuracy, achieving the most complete primary estimation and the least multiple energy leakage, but at the cost of a higher computational burden.
Authors: Jing Sun, Lu Li, Liang Zhang
Abstract: Downward continuation is a critical task in potential field processing, including gravity and magnetic fields, which aims to transfer data from one observation surface to another that is closer to the source of the field. Its effectiveness directly impacts the success of detecting and highlighting subsurface anomalous sources. We treat downward continuation as an inverse problem that relies on solving a forward problem defined by the formula for upward continuation, and we propose a new physics-trained deep neural network (DNN)-based solution for this task. We hard-code the upward continuation process into the DNN's learning framework, where the DNN itself learns to act as the inverse problem solver and can perform downward continuation without ever being shown any ground truth data. We test the proposed method on both synthetic magnetic data and real-world magnetic data from West Antarctica. The preliminary results demonstrate its effectiveness through comparison with selected benchmarks, opening future avenues for the combined use of DNNs and established geophysical theories to address broader potential field inverse problems, such as density and geometry modelling.
Authors: Reza Taherdangkoo, Mostafa Mollaali, Matthias Ehrhardt, Thomas Nagel, Lyesse Laloui, Alessio Ferrari, Christoph Butscher
Abstract: The hydro-mechanical behavior of clay-sulfate rocks, especially their swelling properties, poses significant challenges in geotechnical engineering. This study presents a hybrid constrained machine learning (ML) model developed using the categorical boosting algorithm (CatBoost) tuned with a Bayesian optimization algorithm to predict and analyze the swelling behavior of these complex geological materials. Initially, a coupled hydro-mechanical model based on the Richards' equation coupled to a deformation process with linear kinematics implemented within the finite element framework OpenGeoSys was used to simulate the observed ground heave in Staufen, Germany, caused by water inflow into the clay-sulfate bearing Triassic Grabfeld Formation. A systematic parametric analysis using Gaussian distributions of key parameters, including Young's modulus, Poisson's ratio, maximum swelling pressure, permeability, and air entry pressure, was performed to construct a synthetic database. The ML model takes time, spatial coordinates, and these parameter values as inputs, while water saturation, porosity, and vertical displacement are outputs. In addition, penalty terms were incorporated into the CatBoost objective function to enforce physically meaningful predictions. Results show that the hybrid approach effectively captures the nonlinear and dynamic interactions that govern hydro-mechanical processes. The study demonstrates the ability of the model to predict the swelling behavior of clay-sulfate rocks, providing a robust tool for risk assessment and management in affected regions. The results highlight the potential of ML-driven models to address complex geotechnical challenges.
Authors: Nadav Timor, Jonathan Mamou, Daniel Korat, Moshe Berchansky, Oren Pereg, Gaurav Jain, Roy Schwartz, Moshe Wasserblat, David Harel
Abstract: Accelerating the inference of large language models (LLMs) is a critical challenge in generative AI. Speculative decoding (SD) methods offer substantial efficiency gains by generating multiple tokens using a single target forward pass. However, existing SD approaches require the drafter and target models to share the same vocabulary, thus limiting the pool of possible drafters, often necessitating the training of a drafter from scratch. We present three new SD methods that remove this shared-vocabulary constraint. All three methods preserve the target distribution (i.e., they are lossless) and work with off-the-shelf models without requiring additional training or modifications. Empirically, on summarization, programming, and long-context tasks, our algorithms achieve significant speedups over standard autoregressive decoding. By enabling any off-the-shelf model to serve as drafter and requiring no retraining, this work substantially broadens the applicability of the SD framework in practice.
Authors: Jonah Botvinick-Greenhouse
Abstract: We propose a novel approach for performing dynamical system identification, based upon the comparison of simulated and observed physical invariant measures. While standard methods adopt a Lagrangian perspective by directly treating time-trajectories as inference data, we take on an Eulerian perspective and instead seek models fitting the observed global time-invariant statistics. With this change in perspective, we gain robustness against pervasive challenges in system identification including noise, chaos, and slow sampling. In the first half of this paper, we pose the system identification task as a partial differential equation (PDE) constrained optimization problem, in which synthetic stationary solutions of the Fokker-Planck equation, obtained as fixed points of a finite-volume discretization, are compared to physical invariant measures extracted from observed trajectory data. In the latter half of the paper, we improve upon this approach in two crucial directions. First, we develop a Galerkin-inspired modification to the finite-volume surrogate model, based on data-adaptive unstructured meshes and Monte-Carlo integration, enabling the approach to efficiently scale to high-dimensional problems. Second, we leverage Takens' seminal time-delay embedding theory to introduce a critical data-dependent coordinate transformation which can guarantee unique system identifiability from the invariant measure alone. This contribution resolves a major challenge of system identification through invariant measures, as systems exhibiting distinct transient behaviors may still share the same time-invariant statistics in their state-coordinates. Throughout, we present comprehensive numerical tests which highlight the effectiveness of our approach on a variety of challenging system identification tasks.
Authors: Yago Romano Martinez, Brady Carter, Abhijeet Solanki, Wesam Al Amiri, Syed Rafay Hasan, Terry N. Guo
Abstract: Autonomous vehicles (AVs) rely heavily on cameras and artificial intelligence (AI) to make safe and accurate driving decisions. However, since AI is the core enabling technology, this raises serious cyber threats that hinder the large-scale adoption of AVs. Therefore, it becomes crucial to analyze the resilience of AV security systems against sophisticated attacks that manipulate camera inputs, deceiving AI models. In this paper, we develop camera-camouflaged adversarial attacks targeting traffic sign recognition (TSR) in AVs. Specifically, if the attack is initiated by modifying the texture of a stop sign to fool the AV's object detection system, thereby affecting the AV actuators. The attack's effectiveness is tested using the CARLA AV simulator and the results show that such an attack can delay the auto-braking response to the stop sign, resulting in potential safety issues. We conduct extensive experiments under various conditions, confirming that our new attack is effective and robust. Additionally, we address the attack by presenting mitigation strategies. The proposed attack and defense methods are applicable to other end-to-end trained autonomous cyber-physical systems.
Authors: Shicheng Zhou, Zizhou Zhang, Rong Zhang, Yuchen Yin, Chia Hong Chang, Qinyan Shen
Abstract: This paper analyses the investment returns of three stock sectors, Manuf, Hitec, and Other, in the U.S. stock market, based on the Fama-French three-factor model, the Carhart four-factor model, and the Fama-French five-factor model, in order to test the validity of the Fama-French three-factor model, the Carhart four-factor model, and the Fama-French five-factor model for the three sectors of the market. French five-factor model for the three sectors of the market. Also, the LSTM model is used to explore the additional factors affecting stock returns. The empirical results show that the Fama-French five-factor model has better validity for the three segments of the market under study, and the LSTM model has the ability to capture the factors affecting the returns of certain industries, and can better regress and predict the stock returns of the relevant industries. Keywords- Fama-French model; Carhart model; Factor model; LSTM model.
Authors: Pierre Fernandez
Abstract: Watermarking embeds information into digital content like images, audio, or text, imperceptible to humans but robustly detectable by specific algorithms. This technology has important applications in many challenges of the industry such as content moderation, tracing AI-generated content, and monitoring the usage of AI models. The contributions of this thesis include the development of new watermarking techniques for images, audio, and text. We first introduce methods for active moderation of images on social platforms. We then develop specific techniques for AI-generated content. We specifically demonstrate methods to adapt latent generative models to embed watermarks in all generated content, identify watermarked sections in speech, and improve watermarking in large language models with tests that ensure low false positive rates. Furthermore, we explore the use of digital watermarking to detect model misuse, including the detection of watermarks in language models fine-tuned on watermarked text, and introduce training-free watermarks for the weights of large transformers. Through these contributions, the thesis provides effective solutions for the challenges posed by the increasing use of generative AI models and the need for model monitoring and content moderation. It finally examines the challenges and limitations of watermarking techniques and discuss potential future directions for research in this area.
Authors: Yitong Duan, Weiran Wang, Jian Li
Abstract: As a fundamental method in economics and finance, the factor model has been extensively utilized in quantitative investment. In recent years, there has been a paradigm shift from traditional linear models with expert-designed factors to more flexible nonlinear machine learning-based models with data-driven factors, aiming to enhance the effectiveness of these factor models. However, due to the low signal-to-noise ratio in market data, mining effective factors in data-driven models remains challenging. In this work, we propose a hypergraph-based factor model with temporal residual contrastive learning (FactorGCL) that employs a hypergraph structure to better capture high-order nonlinear relationships among stock returns and factors. To mine hidden factors that supplement human-designed prior factors for predicting stock returns, we design a cascading residual hypergraph architecture, in which the hidden factors are extracted from the residual information after removing the influence of prior factors. Additionally, we propose a temporal residual contrastive learning method to guide the extraction of effective and comprehensive hidden factors by contrasting stock-specific residual information over different time periods. Our extensive experiments on real stock market data demonstrate that FactorGCL not only outperforms existing state-of-the-art methods but also mines effective hidden factors for predicting stock returns.
Authors: Buyun Liang, Kwan Ho Ryan Chan, Darshan Thaker, Jinqi Luo, Ren\'e Vidal
Abstract: Jailbreak attacks exploit specific prompts to bypass LLM safeguards, causing the LLM to generate harmful, inappropriate, and misaligned content. Current jailbreaking methods rely heavily on carefully designed system prompts and numerous queries to achieve a single successful attack, which is costly and impractical for large-scale red-teaming. To address this challenge, we propose to distill the knowledge of an ensemble of SOTA attackers into a single open-source model, called Knowledge-Distilled Attacker (KDA), which is finetuned to automatically generate coherent and diverse attack prompts without the need for meticulous system prompt engineering. Compared to existing attackers, KDA achieves higher attack success rates and greater cost-time efficiency when targeting multiple SOTA open-source and commercial black-box LLMs. Furthermore, we conducted a quantitative diversity analysis of prompts generated by baseline methods and KDA, identifying diverse and ensemble attacks as key factors behind KDA's effectiveness and efficiency.
Authors: Adam Stooke, Rohit Prabhavalkar, Khe Chai Sim, Pedro Moreno Mengibar
Abstract: Modern systems for automatic speech recognition, including the RNN-Transducer and Attention-based Encoder-Decoder (AED), are designed so that the encoder is not required to alter the time-position of information from the audio sequence into the embedding; alignment to the final text output is processed during decoding. We discover that the transformer-based encoder adopted in recent years is actually capable of performing the alignment internally during the forward pass, prior to decoding. This new phenomenon enables a simpler and more efficient model, the "Aligner-Encoder". To train it, we discard the dynamic programming of RNN-T in favor of the frame-wise cross-entropy loss of AED, while the decoder employs the lighter text-only recurrence of RNN-T without learned cross-attention -- it simply scans embedding frames in order from the beginning, producing one token each until predicting the end-of-message. We conduct experiments demonstrating performance remarkably close to the state of the art, including a special inference configuration enabling long-form recognition. In a representative comparison, we measure the total inference time for our model to be 2x faster than RNN-T and 16x faster than AED. Lastly, we find that the audio-text alignment is clearly visible in the self-attention weights of a certain layer, which could be said to perform "self-transduction".
Authors: Shehzeen Hussain, Paarth Neekhara, Xuesong Yang, Edresson Casanova, Subhankar Ghosh, Mikyas T. Desta, Roy Fejgin, Rafael Valle, Jason Li
Abstract: While autoregressive speech token generation models produce speech with remarkable variety and naturalness, their inherent lack of controllability often results in issues such as hallucinations and undesired vocalizations that do not conform to conditioning inputs. We introduce Koel-TTS, a suite of enhanced encoder-decoder Transformer TTS models that address these challenges by incorporating preference alignment techniques guided by automatic speech recognition and speaker verification models. Additionally, we incorporate classifier-free guidance to further improve synthesis adherence to the transcript and reference speaker audio. Our experiments demonstrate that these optimizations significantly enhance target speaker similarity, intelligibility, and naturalness of synthesized speech. Notably, Koel-TTS directly maps text and context audio to acoustic tokens, and on the aforementioned metrics, outperforms state-of-the-art TTS models, despite being trained on a significantly smaller dataset. Audio samples and demos are available on our website.
Authors: Guanxu Chen, Dongrui Liu, Tao Luo, Jing Shao
Abstract: Explaining the hidden representations of Large Language Models (LLMs) is a perspective to understand LLMs' underlying inference logic and improve their reliability in application scenarios. However, previous methods introduce external ''black-box'' modules to explain ''black-box'' LLMs, increasing the potential uncertainty and failing to provide faithful explanations. In this paper, we propose a self-explaining method SEER, enhancing LLMs' explainability by aggregating the same concept and disentangling the different concepts in the representation space. In this way, SEER provides faithful explanations carried by representations synchronously with the LLMs' output. Additionally, we showcase the applications of SEER on trustworthiness-related tasks (e.g., the safety risks classification and detoxification tasks), where self-explained LLMs achieve consistent improvement in explainability and performance. More crucially, we theoretically analyze the improvement of SEER on LLMs' generalization ability through optimal transport theory.
Authors: Andreas Krause, Jonas H\"ubotter
Abstract: Artificial intelligence commonly refers to the science and engineering of artificial systems that can carry out tasks generally associated with requiring aspects of human intelligence, such as playing games, translating languages, and driving cars. In recent years, there have been exciting advances in learning-based, data-driven approaches towards AI, and machine learning and deep learning have enabled computer systems to perceive the world in unprecedented ways. Reinforcement learning has enabled breakthroughs in complex games such as Go and challenging robotics tasks such as quadrupedal locomotion. A key aspect of intelligence is to not only make predictions, but reason about the uncertainty in these predictions, and to consider this uncertainty when making decisions. This is what this manuscript on "Probabilistic Artificial Intelligence" is about. The first part covers probabilistic approaches to machine learning. We discuss the differentiation between "epistemic" uncertainty due to lack of data and "aleatoric" uncertainty, which is irreducible and stems, e.g., from noisy observations and outcomes. We discuss concrete approaches towards probabilistic inference and modern approaches to efficient approximate inference. The second part of the manuscript is about taking uncertainty into account in sequential decision tasks. We consider active learning and Bayesian optimization -- approaches that collect data by proposing experiments that are informative for reducing the epistemic uncertainty. We then consider reinforcement learning and modern deep RL approaches that use neural network function approximation. We close by discussing modern approaches in model-based RL, which harness epistemic and aleatoric uncertainty to guide exploration, while also reasoning about safety.
Authors: Qi Ye, Shuangyue Geng, Zizhao Han, Weikang Li, L. -M. Duan, Dong-Ling Deng
Abstract: Machine learning is widely believed to be one of the most promising practical applications of quantum computing. Existing quantum machine learning schemes typically employ a quantum-classical hybrid approach that relies crucially on gradients of model parameters. Such an approach lacks provable convergence to global minima and will become infeasible as quantum learning models scale up. Here, we introduce quantum automated learning, where no variational parameter is involved and the training process is converted to quantum state preparation. In particular, we encode training data into unitary operations and iteratively evolve a random initial state under these unitaries and their inverses, with a target-oriented perturbation towards higher prediction accuracy sandwiched in between. Under reasonable assumptions, we rigorously prove that the evolution converges exponentially to the desired state corresponding to the global minimum of the loss function. We show that such a training process can be understood from the perspective of preparing quantum states by imaginary time evolution, where the data-encoded unitaries together with target-oriented perturbations would train the quantum learning model in an automated fashion. We further prove that the quantum automated learning paradigm features good generalization ability with the generalization error upper bounded by the ratio between a logarithmic function of the Hilbert space dimension and the number of training samples. In addition, we carry out extensive numerical simulations on real-life images and quantum data to demonstrate the effectiveness of our approach and validate the assumptions. Our results establish an unconventional quantum learning strategy that is gradient-free with provable and explainable trainability, which would be crucial for large-scale practical applications of quantum computing in machine learning scenarios.
Authors: Purbaditya Bhattacharya, Patrick Nowak
Abstract: Drones or unmanned aerial vehicles are traditionally used for military missions, warfare, and espionage. However, the usage of drones has significantly increased due to multiple industrial applications involving security and inspection, transportation, research purposes, and recreational drone flying. Such an increased volume of drone activity in public spaces requires regulatory actions for purposes of privacy protection and safety. Hence, detection of illegal drone activities such as boundary encroachment becomes a necessity. Such detection tasks are usually automated and performed by deep learning models which are trained on annotated image datasets. This paper builds on a previous work and extends an already published open source dataset. A description and analysis of the entire dataset is provided. The dataset is used to train the YOLOv7 deep learning model and some of its minor variants and the results are provided. Since the detection models are based on a single image input, a simple cross-correlation based tracker is used to reduce detection drops and improve tracking performance in videos. Finally, the entire drone detection system is summarized.
Authors: Liwei Jiang, Abhishek Roy, Krishna Balasubramanian, Damek Davis, Dmitriy Drusvyatskiy, Sen Na
Abstract: We consider applying stochastic approximation (SA) methods to solve nonsmooth variational inclusion problems. Existing studies have shown that the averaged iterates of SA methods exhibit asymptotic normality, with an optimal limiting covariance matrix in the local minimax sense of H\'ajek and Le Cam. However, no methods have been proposed to estimate this covariance matrix in a nonsmooth and potentially non-monotone (nonconvex) setting. In this paper, we study an online batch-means covariance matrix estimator introduced in Zhu et al.(2023). The estimator groups the SA iterates appropriately and computes the sample covariance among batches as an estimate of the limiting covariance. Its construction does not require prior knowledge of the total sample size, and updates can be performed recursively as new data arrives. We establish that, as long as the batch size sequence is properly specified (depending on the stepsize sequence), the estimator achieves a convergence rate of order $O(\sqrt{d}n^{-1/8+\varepsilon})$ for any $\varepsilon>0$, where $d$ and $n$ denote the problem dimensionality and the number of iterations (or samples) used. Although the problem is nonsmooth and potentially non-monotone (nonconvex), our convergence rate matches the best-known rate for covariance estimation methods using only first-order information in smooth and strongly-convex settings. The consistency of this covariance estimator enables asymptotically valid statistical inference, including constructing confidence intervals and performing hypothesis testing.
Authors: Muhammad Imran, Jonathan R. Krebs, Vishal Balaji Sivaraman, Teng Zhang, Amarjeet Kumar, Walker R. Ueland, Michael J. Fassler, Jinlong Huang, Xiao Sun, Lisheng Wang, Pengcheng Shi, Maximilian Rokuss, Michael Baumgartner, Yannick Kirchhof, Klaus H. Maier-Hein, Fabian Isensee, Shuolin Liu, Bing Han, Bong Thanh Nguyen, Dong-jin Shin, Park Ji-Woo, Mathew Choi, Kwang-Hyun Uhm, Sung-Jea Ko, Chanwoong Lee, Jaehee Chun, Jin Sung Kim, Minghui Zhang, Hanxiao Zhang, Xin You, Yun Gu, Zhaohong Pan, Xuan Liu, Xiaokun Liang, Markus Tiefenthaler, Enrique Almar-Munoz, Matthias Schwab, Mikhail Kotyushev, Rostislav Epifanov, Marek Wodzinski, Henning Muller, Abdul Qayyum, Moona Mazher, Steven A. Niederer, Zhiwei Wang, Kaixiang Yang, Jintao Ren, Stine Sofia Korreman, Yuchong Gao, Hongye Zeng, Haoyu Zheng, Rui Zheng, Jinghua Yue, Fugen Zhou, Bo Liu, Alexander Cosman, Muxuan Liang, Chang Zhao, Gilbert R. Upchurch Jr., Jun Ma, Yuyin Zhou, Michol A. Cooper, Wei Shao
Abstract: Multi-class segmentation of the aorta in computed tomography angiography (CTA) scans is essential for diagnosing and planning complex endovascular treatments for patients with aortic dissections. However, existing methods reduce aortic segmentation to a binary problem, limiting their ability to measure diameters across different branches and zones. Furthermore, no open-source dataset is currently available to support the development of multi-class aortic segmentation methods. To address this gap, we organized the AortaSeg24 MICCAI Challenge, introducing the first dataset of 100 CTA volumes annotated for 23 clinically relevant aortic branches and zones. This dataset was designed to facilitate both model development and validation. The challenge attracted 121 teams worldwide, with participants leveraging state-of-the-art frameworks such as nnU-Net and exploring novel techniques, including cascaded models, data augmentation strategies, and custom loss functions. We evaluated the submitted algorithms using the Dice Similarity Coefficient (DSC) and Normalized Surface Distance (NSD), highlighting the approaches adopted by the top five performing teams. This paper presents the challenge design, dataset details, evaluation metrics, and an in-depth analysis of the top-performing algorithms. The annotated dataset, evaluation code, and implementations of the leading methods are publicly available to support further research. All resources can be accessed at https://aortaseg24.grand-challenge.org.
Authors: Chiara Criscuolo, Davide Martinenghi, Giuseppe Piccirillo
Abstract: We address the critical issue of biased algorithms and unfair rankings, which have permeated various sectors, including search engines, recommendation systems, and workforce management. These biases can lead to discriminatory outcomes in a data-driven world, especially against marginalized and underrepresented groups. Efforts towards responsible data science and responsible artificial intelligence aim to mitigate these biases and promote fairness, diversity, and transparency. However, most fairness-aware ranking methods singularly focus on protected attributes such as race, gender, or socio-economic status, neglecting the intersectionality of these attributes, i.e., the interplay between multiple social identities. Understanding intersectionality is crucial to ensure that existing inequalities are not preserved by fair rankings. We offer a description of the main ways to incorporate intersectionality in fair ranking systems through practical examples and provide a comparative overview of existing literature and a synoptic table summarizing the various methodologies. Our analysis highlights the need for intersectionality to attain fairness, while also emphasizing that fairness, alone, does not necessarily imply intersectionality.
Authors: David Islip, Roy H. Kwon, Sanghyeon Bae, Woo Chang Kim
Abstract: Two-stage stochastic programs (2SPs) are important tools for making decisions under uncertainty. Decision-makers use contextual information to generate a set of scenarios to represent the true conditional distribution. However, the number of scenarios required is a barrier to implementing 2SPs, motivating the problem of generating a small set of surrogate scenarios that yield high-quality decisions when they represent uncertainty. Current scenario generation approaches do not leverage contextual information or do not address computational concerns. In response, we propose contextual scenario generation (CSG) to learn a mapping between the context and a set of surrogate scenarios of user-specified size. First, we propose a distributional approach that learns the mapping by minimizing a distributional distance between the predicted surrogate scenarios and the true contextual distribution. Second, we propose a task-based approach that aims to produce surrogate scenarios that yield high-quality decisions. The task-based approach uses neural architectures to approximate the downstream objective and leverages the approximation to search for the mapping. The proposed approaches apply to various problem structures and loosely only require efficient solving of the associated subproblems and 2SPs defined on the reduced scenario sets. Numerical experiments demonstrating the effectiveness of the proposed methods are presented.
Authors: Subhamoy Chatterjee, Andres Munoz-Jaramillo, Anna Malanushenko
Abstract: Deep generative models have shown immense potential in generating unseen data that has properties of real data. These models learn complex data-generating distributions starting from a smaller set of latent dimensions. However, generative models have encountered great skepticism in scientific domains due to the disconnection between generative latent vectors and scientifically relevant quantities. In this study, we integrate three types of machine learning models to generate solar magnetic patches in a physically interpretable manner and use those as a query to find matching patches in real observations. We use the magnetic field measurements from Space-weather HMI Active Region Patches (SHARPs) to train a Generative Adversarial Network (GAN). We connect the physical properties of GAN-generated images with their latent vectors to train Support Vector Machines (SVMs) that do mapping between physical and latent spaces. These produce directions in the GAN latent space along which known physical parameters of the SHARPs change. We train a self-supervised learner (SSL) to make queries with generated images and find matches from real data. We find that the GAN-SVM combination enables users to produce high-quality patches that change smoothly only with a prescribed physical quantity, making generative models physically interpretable. We also show that GAN outputs can be used to retrieve real data that shares the same physical properties as the generated query. This elevates Generative Artificial Intelligence (AI) from a means-to-produce artificial data to a novel tool for scientific data interrogation, supporting its applicability beyond the domain of heliophysics.
Authors: Julian Killingback, Hansi Zeng, Hamed Zamani
Abstract: The vast majority of retrieval models depend on vector inner products to produce a relevance score between a query and a document. This naturally limits the expressiveness of the relevance score that can be employed. We propose a new paradigm, instead of producing a vector to represent the query we produce a small neural network which acts as a learned relevance function. This small neural network takes in a representation of the document, in this paper we use a single vector, and produces a scalar relevance score. To produce the little neural network we use a hypernetwork, a network that produce the weights of other networks, as our query encoder or as we call it a Hypencoder. Experiments on in-domain search tasks show that Hypencoder is able to significantly outperform strong dense retrieval models and has higher metrics then reranking models and models an order of magnitude larger. Hypencoder is also shown to generalize well to out-of-domain search tasks. To assess the extent of Hypencoder's capabilities, we evaluate on a set of hard retrieval tasks including tip-of-the-tongue retrieval and instruction-following retrieval tasks and find that the performance gap widens substantially compared to standard retrieval tasks. Furthermore, to demonstrate the practicality of our method we implement an approximate search algorithm and show that our model is able to search 8.8M documents in under 60ms.
Authors: Almuthanna Alageel, Sergio Maffeis, Imperial College London
Abstract: Advanced Persistent Threats (APTs) are among the most sophisticated threats facing critical organizations worldwide. APTs employ specific tactics, techniques, and procedures (TTPs) which make them difficult to detect in comparison to frequent and aggressive attacks. In fact, current network intrusion detection systems struggle to detect APTs communications, allowing such threats to persist unnoticed on victims' machines for months or even years. In this paper, we present EarlyCrow, an approach to detect APT malware command and control over HTTP(S) using contextual summaries. The design of EarlyCrow is informed by a novel threat model focused on TTPs present in traffic generated by tools recently used as part of APT campaigns. The threat model highlights the importance of the context around the malicious connections, and suggests traffic attributes which help APT detection. EarlyCrow defines a novel multipurpose network flow format called PairFlow, which is leveraged to build the contextual summary of a PCAP capture, representing key behavioral, statistical and protocol information relevant to APT TTPs. We evaluate the effectiveness of EarlyCrow on unseen APTs obtaining a headline macro average F1-score of 93.02% with FPR of $0.74%.
Authors: Toufique Ahmed, Jatin Ganhotra, Rangeet Pan, Avraham Shinnar, Saurabh Sinha, Martin Hirzel
Abstract: While there has been plenty of work on generating tests from existing code, there has been limited work on generating tests from issues. A correct test must validate the code patch that resolves the issue. In this work, we focus on the scenario where the code patch does not exist yet. This approach supports two major use-cases. First, it supports TDD (test-driven development), the discipline of "test first, write code later" that has well-documented benefits for human software engineers. Second, it also validates SWE (software engineering) agents, which generate code patches for resolving issues. This paper introduces Otter, an LLM-based solution for generating tests from issues. Otter augments LLMs with rule-based analysis to check and repair their outputs, and introduces a novel self-reflective action planning stage. Experiments show Otter outperforming state-of-the-art systems for generating tests from issues, in addition to enhancing systems that generate patches from issues. We hope that Otter helps make developers more productive at resolving issues and leads to more robust, well-tested code.
Authors: Alireza Heidari, Amirhossein Ahmadi, Wei Zhang
Abstract: In this paper, we introduce DobLIX, a dual-objective learned index specifically designed for Log-Structured Merge(LSM) tree-based key-value stores. Although traditional learned indexes focus exclusively on optimizing index lookups, they often overlook the impact of data access from storage, resulting in performance bottlenecks. DobLIX addresses this by incorporating a second objective, data access optimization, into the learned index training process. This dual-objective approach ensures that both index lookup efficiency and data access costs are minimized, leading to significant improvements in read performance while maintaining write efficiency in real-world LSM-tree systems. Additionally, DobLIX features a reinforcement learning agent that dynamically tunes the system parameters, allowing it to adapt to varying workloads in real-time. Experimental results using real-world datasets demonstrate that DobLIX reduces indexing overhead and improves throughput by 1.19 to 2.21 times compared to state-of-the-art methods within RocksDB, a widely used LSM-tree-based storage engine.
Authors: Baturay Saglam, Zhuoran Yang, Dionysis Kalogerias, Amin Karbasi
Abstract: Large language models (LLMs) have demonstrated remarkable proficiency in in-context learning (ICL), where models adapt to new tasks through example-based prompts without requiring parameter updates. However, understanding how tasks are internally encoded and generalized remains a challenge. To address some of the empirical and technical gaps in the literature, we introduce an automated formulation for encoding task information in ICL prompts as a function of attention heads within the transformer architecture. This approach computes a single task vector as a weighted sum of attention heads, with the weights optimized causally via gradient descent. Our findings show that existing methods fail to generalize effectively to modalities beyond text. In response, we also design a benchmark to evaluate whether a task vector can preserve task fidelity in functional regression tasks. The proposed method successfully extracts task-specific information from in-context demonstrations and excels in both text and regression tasks, demonstrating its generalizability across modalities. Moreover, ablation studies show that our method's effectiveness stems from aligning the distribution of the last hidden state with that of an optimally performing in-context-learned model.
Authors: Maneesha Wickramasuriya, Beomyeol Yu, Taeyoung Lee, Murray Snyder
Abstract: This paper proposes a vision-in-the-loop simulation environment for deep monocular pose estimation of a UAV operating in an ocean environment. Recently, a deep neural network with a transformer architecture has been successfully trained to estimate the pose of a UAV relative to the flight deck of a research vessel, overcoming several limitations of GPS-based approaches. However, validating the deep pose estimation scheme in an actual ocean environment poses significant challenges due to the limited availability of research vessels and the associated operational costs. To address these issues, we present a photo-realistic 3D virtual environment leveraging recent advancements in Gaussian splatting, a novel technique that represents 3D scenes by modeling image pixels as Gaussian distributions in 3D space, creating a lightweight and high-quality visual model from multiple viewpoints. This approach enables the creation of a virtual environment integrating multiple real-world images collected in situ. The resulting simulation enables the indoor testing of flight maneuvers while verifying all aspects of flight software, hardware, and the deep monocular pose estimation scheme. This approach provides a cost-effective solution for testing and validating the autonomous flight of shipboard UAVs, specifically focusing on vision-based control and estimation algorithms.
Authors: Mohammadreza Baharani, Ghazal Alinezhad Noghre, Armin Danesh Pazho, Gabriel Maldonado, Hamed Tabkhi
Abstract: AFoundation Models (FM) have increasingly drawn the attention of researchers due to their scalability and generalization across diverse tasks. Inspired by the success of FMs and the principles that have driven advancements in Large Language Models (LLMs), we introduce MoFM as a novel Motion Foundation Model. MoFM is designed for the semantic understanding of complex human motions in both time and space. To facilitate large-scale training, MotionBook, a comprehensive human motion dictionary of discretized motions is designed and employed. MotionBook utilizes Thermal Cubes to capture spatio-temporal motion heatmaps, applying principles from discrete variational models to encode human movements into discrete units for a more efficient and scalable representation. MoFM, trained on a large corpus of motion data, provides a foundational backbone adaptable to diverse downstream tasks, supporting paradigms such as one-shot, unsupervised, and supervised tasks. This versatility makes MoFM well-suited for a wide range of motion-based applications.
Authors: Manh Luong, Khai Nguyen, Dinh Phung, Gholamreza Haffari, Lizhen Qu
Abstract: Teacher-forcing training for audio captioning usually leads to exposure bias due to training and inference mismatch. Prior works propose the contrastive method to deal with caption degeneration. However, the contrastive method ignores the temporal information when measuring similarity across acoustic and linguistic modalities, leading to inferior performance. In this work, we develop the temporal-similarity score by introducing the unbiased sliced Wasserstein RBF (USW-RBF) kernel equipped with rotary positional embedding to account for temporal information across modalities. In contrast to the conventional sliced Wasserstein RBF kernel, we can form an unbiased estimation of USW-RBF kernel via Monte Carlo estimation. Therefore, it is well-suited to stochastic gradient optimization algorithms, and its approximation error decreases at a parametric rate of $\mathcal{O}(L^{-1/2})$ with $L$ Monte Carlo samples. Additionally, we introduce an audio captioning framework based on the unbiased sliced Wasserstein kernel, incorporating stochastic decoding methods to mitigate caption degeneration during the generation process. We conduct extensive quantitative and qualitative experiments on two datasets, AudioCaps and Clotho, to illustrate the capability of generating high-quality audio captions. Experimental results show that our framework is able to increase caption length, lexical diversity, and text-to-audio self-retrieval accuracy.
Authors: Weiming Feng, Hongyang Liu, Minji Yang
Abstract: Spin systems form an important class of undirected graphical models. For two Gibbs distributions $\mu$ and $\nu$ induced by two spin systems on the same graph $G = (V, E)$, we study the problem of approximating the total variation distance $d_{TV}(\mu,\nu)$ with an $\epsilon$-relative error. We propose a new reduction that connects the problem of approximating the TV-distance to sampling and approximate counting. Our applications include the hardcore model and the antiferromagnetic Ising model in the uniqueness regime, the ferromagnetic Ising model, and the general Ising model satisfying the spectral condition. Additionally, we explore the computational complexity of approximating the total variation distance $d_{TV}(\mu_S,\nu_S)$ between two marginal distributions on an arbitrary subset $S \subseteq V$. We prove that this problem remains hard even when both $\mu$ and $\nu$ admit polynomial-time sampling and approximate counting algorithms.
Authors: Izunna Okpala, Ashkan Golgoon, Arjun Ravi Kannan
Abstract: The advent of large language models has ushered in a new era of agentic systems, where artificial intelligence programs exhibit remarkable autonomous decision-making capabilities across diverse domains. This paper explores agentic system workflows in the financial services industry. In particular, we build agentic crews that can effectively collaborate to perform complex modeling and model risk management (MRM) tasks. The modeling crew consists of a manager and multiple agents who perform specific tasks such as exploratory data analysis, feature engineering, model selection, hyperparameter tuning, model training, model evaluation, and writing documentation. The MRM crew consists of a manager along with specialized agents who perform tasks such as checking compliance of modeling documentation, model replication, conceptual soundness, analysis of outcomes, and writing documentation. We demonstrate the effectiveness and robustness of modeling and MRM crews by presenting a series of numerical examples applied to credit card fraud detection, credit card approval, and portfolio credit risk modeling datasets.
Authors: Dylan Waldner, Risto Miikkulainen
Abstract: As AI models grow in power and generality, understanding how agents learn and make decisions in complex environments is critical to promoting ethical behavior. This paper examines the ethical implications of implementing biological drives, specifically, self preservation, into three different agents. A Bayesian agent optimized with NEAT, a Bayesian agent optimized with stochastic variational inference, and a GPT 4o agent play a simulated, LLM generated text based adventure game. The agents select actions at each scenario to survive, adapting to increasingly challenging scenarios. Post simulation analysis evaluates the ethical scores of the agent's decisions, uncovering the tradeoffs they navigate to survive. Specifically, analysis finds that when danger increases, agents ignore ethical considerations and opt for unethical behavior. The agents' collective behavior, trading ethics for survival, suggests that prioritizing survival increases the risk of unethical behavior. In the context of AGI, designing agents to prioritize survival may amplify the likelihood of unethical decision making and unintended emergent behaviors, raising fundamental questions about goal design in AI safety research.
Authors: Weizhe Chen, Sven Koenig, Bistra Dilkina
Abstract: The recent release of OpenAI's o1 models and other similar frameworks showcasing test-time scaling laws has demonstrated their exceptional capability to tackle complex reasoning tasks. Inspired by this, subsequent research has revealed that such test-time scaling laws hinge on the model's ability to search both within a single response (intra-response) and across multiple responses (inter-response) during training. Crucially, beyond selecting a single optimal response, the model must also develop robust self-correction capabilities within its own outputs. However, training models to achieve effective self-evaluation and self-correction remains a significant challenge, heavily dependent on the quality of self-reflection data. In this paper, we address this challenge by focusing on enhancing the quality of self-reflection data generation for complex problem-solving, which can subsequently improve the training of next-generation large language models (LLMs). Specifically, we explore how manually triggering a model's self-correction mechanisms can improve performance on challenging reasoning tasks. To this end, we propose a novel iterative deepening sampling algorithm framework designed to enhance self-correction and generate higher-quality samples. Through extensive experiments on Math500 and AIME benchmarks, we demonstrate that our method achieves a higher success rate on difficult tasks and provide detailed ablation studies to analyze its effectiveness across diverse settings.
Authors: Shadab Ahamed, Simon Ghyselincks, Pablo Chang Huang Arias, Julian Kloiber, Yasin Ranjbar, Jingrong Tang, Niloufar Zakariaei, Eldad Haber
Abstract: Magnetic data inversion is an important tool in geophysics, used to infer subsurface magnetic susceptibility distributions from surface magnetic field measurements. This inverse problem is inherently ill-posed, characterized by non-unique solutions, depth ambiguity, and sensitivity to noise. Traditional inversion approaches rely on predefined regularization techniques to stabilize solutions, limiting their adaptability to complex or diverse geological scenarios. In this study, we propose an approach that integrates variable dictionary learning and scale-space methods to address these challenges. Our method employs learned dictionaries, allowing for adaptive representation of complex subsurface features that are difficult to capture with predefined bases. Additionally, we extend classical variational inversion by incorporating multi-scale representations through a scale-space framework, enabling the progressive introduction of structural detail while mitigating overfitting. We implement both fixed and dynamic dictionary learning techniques, with the latter introducing iteration-dependent dictionaries for enhanced flexibility. Using a synthetic dataset to simulate geological scenarios, we demonstrate significant improvements in reconstruction accuracy and robustness compared to conventional variational and dictionary-based methods. Our results highlight the potential of learned dictionaries, especially when coupled with scale-space dynamics, to improve model recovery and noise handling. These findings underscore the promise of our data-driven approach for advance magnetic data inversion and its applications in geophysical exploration, environmental assessment, and mineral prospecting.
Authors: Vivek Myers, Bill Chunyuan Zheng, Anca Dragan, Kuan Fang, Sergey Levine
Abstract: Effective task representations should facilitate compositionality, such that after learning a variety of basic tasks, an agent can perform compound tasks consisting of multiple steps simply by composing the representations of the constituent steps together. While this is conceptually simple and appealing, it is not clear how to automatically learn representations that enable this sort of compositionality. We show that learning to associate the representations of current and future states with a temporal alignment loss can improve compositional generalization, even in the absence of any explicit subtask planning or reinforcement learning. We evaluate our approach across diverse robotic manipulation tasks as well as in simulation, showing substantial improvements for tasks specified with either language or goal images.
Authors: Marianne Ab\'emgnigni Njifon, Tobias Weber, Viktor Bezborodov, Tyll Krueger, Dominic Schuhmacher
Abstract: Accurate tumor classification is essential for selecting effective treatments, but current methods have limitations. Standard tumor grading, which categorizes tumors based on cell differentiation, is not recommended as a stand-alone procedure, as some well-differentiated tumors can be malignant. Tumor heterogeneity assessment via single-cell sequencing offers profound insights but can be costly and may still require significant manual intervention. Many existing statistical machine learning methods for tumor data still require complex pre-processing of MRI and histopathological data. In this paper, we propose to build on a mathematical model that simulates tumor evolution (O\.{z}a\'{n}ski (2017)) and generate artificial datasets for tumor classification. Tumor heterogeneity is estimated using normalized entropy, with a threshold to classify tumors as having high or low heterogeneity. Our contributions are threefold: (1) the cut and graph generation processes from the artificial data, (2) the design of tumor features, and (3) the construction of Block Graph Neural Networks (BGNN), a Graph Neural Network-based approach to predict tumor heterogeneity. The experimental results reveal that the combination of the proposed features and models yields excellent results on artificially generated data ($89.67\%$ accuracy on the test data). In particular, in alignment with the emerging trends in AI-assisted grading and spatial transcriptomics, our results suggest that enriching traditional grading methods with birth (e.g., Ki-67 proliferation index) and death markers can improve heterogeneity prediction and enhance tumor classification.
Authors: Kaushik Bhattacharya, Lianghao Cao, George Stepaniants, Andrew Stuart, Margaret Trautner
Abstract: The theory of homogenization provides a systematic approach to the derivation of macroscale constitutive laws, obviating the need to repeatedly resolve complex microstructure. However, the unit cell problem that defines the constitutive model is typically not amenable to explicit evaluation. It is therefore of interest to learn constitutive models from data generated by the unit cell problem. Many viscoelastic and elastoviscoplastic materials are characterized by memory-dependent constitutive laws. In order to amortize the computational investment in finding such memory-dependent constitutive laws, it is desirable to learn their dependence on the material microstructure. While prior work has addressed learning memory dependence and material dependence separately, their joint learning has not been considered. This paper focuses on the joint learning problem and proposes a novel neural operator framework to address it. In order to provide firm foundations, the homogenization problem for linear Kelvin-Voigt viscoelastic materials is studied. The theoretical properties of the cell problem in this Kelvin-Voigt setting are used to motivate the proposed general neural operator framework; these theoretical properties are also used to prove a universal approximation theorem for the learned macroscale constitutive model. This formulation of learnable constitutive models is then deployed beyond the Kelvin-Voigt setting. Numerical experiments are presented showing that the resulting data-driven methodology accurately learns history- and microstructure-dependent linear viscoelastic and nonlinear elastoviscoplastic constitutive models, and numerical results also demonstrate that the resulting constitutive models can be deployed in macroscale simulation of material deformation.
Authors: Qianteng Zhu, Gert Aarts, Wei Wang, Kai Zhou, Lingxiao Wang
Abstract: We develop diffusion models for simulating lattice gauge theories, where stochastic quantization is explicitly incorporated as a physical condition for sampling. We demonstrate the applicability of this novel sampler to U(1) gauge theory in two spacetime dimensions and find that a model trained at a small inverse coupling constant can be extrapolated to larger inverse coupling regions without encountering the topological freezing problem. Additionally, the trained model can be employed to sample configurations on different lattice sizes without requiring further training. The exactness of the generated samples is ensured by incorporating Metropolis-adjusted Langevin dynamics into the generation process. Furthermore, we demonstrate that this approach enables more efficient sampling of topological quantities compared to traditional algorithms such as Hybrid Monte Carlo and Langevin simulations.
Authors: Brandon Ho, Batuhan Altundas, Matthew Gombolay
Abstract: In fast-paced, ever-changing environments, dynamic Motion Planning for Multi-Agent Systems in the presence of obstacles is a universal and unsolved problem. Be it from path planning around obstacles to the movement of robotic arms, or in planning navigation of robot teams in settings such as Robosoccer, dynamic motion planning is needed to avoid collisions while reaching the targeted destination when multiple agents occupy the same area. In continuous domains where the world changes quickly, existing classical Motion Planning algorithms such as RRT* and A* become computationally expensive to rerun at every time step. Many variations of classical and well-formulated non-learning path-planning methods have been proposed to solve this universal problem but fall short due to their limitations of speed, smoothness, optimally, etc. Deep Learning models overcome their challenges due to their ability to adapt to varying environments based on past experience. However, current learning motion planning models use discretized environments, do not account for heterogeneous agents or replanning, and build up to improve the classical motion planners' efficiency, leading to issues with scalability. To prevent collisions between heterogenous team members and collision to obstacles while trying to reach the target location, we present a learning-based dynamic navigation model and show our model working on a simple environment in the concept of a simple Robosoccer Game.
Authors: Xinsong Feng, Zihan Yu, Yanhai Xiong, Haipeng Chen
Abstract: Reinforcement learning (RL) has emerged as a promising tool for combinatorial optimization (CO) problems due to its ability to learn fast, effective, and generalizable solutions. Nonetheless, existing works mostly focus on one-shot deterministic CO, while sequential stochastic CO (SSCO) has rarely been studied despite its broad applications such as adaptive influence maximization (IM) and infectious disease intervention. In this paper, we study the SSCO problem where we first decide the budget (e.g., number of seed nodes in adaptive IM) allocation for all time steps, and then select a set of nodes for each time step. The few existing studies on SSCO simplify the problems by assuming a uniformly distributed budget allocation over the time horizon, yielding suboptimal solutions. We propose a generic hierarchical RL (HRL) framework called wake-sleep option (WS-option), a two-layer option-based framework that simultaneously decides adaptive budget allocation on the higher layer and node selection on the lower layer. WS-option starts with a coherent formulation of the two-layer Markov decision processes (MDPs), capturing the interdependencies between the two layers of decisions. Building on this, WS-option employs several innovative designs to balance the model's training stability and computational efficiency, preventing the vicious cyclic interference issue between the two layers. Empirical results show that WS-option exhibits significantly improved effectiveness and generalizability compared to traditional methods. Moreover, the learned model can be generalized to larger graphs, which significantly reduces the overhead of computational resources.
Authors: Yixian Shen, Qi Bi, Jia-Hong Huang, Hongyi Zhu, Andy D. Pimentel, Anuj Pathania
Abstract: Low-rank adaptation (LoRA) has been demonstrated effective in reducing the trainable parameter number when fine-tuning a large foundation model (LLM). However, it still encounters computational and memory challenges when scaling to larger models or addressing more complex task adaptation. In this work, we introduce Sparse Spectrum Adaptation via Discrete Hartley Transformation (SSH), a novel approach that significantly reduces the number of trainable parameters while enhancing model performance. It selects the most informative spectral components across all layers, under the guidance of the initial weights after a discrete Hartley transformation (DHT). The lightweight inverse DHT then projects the spectrum back into the spatial domain for updates. Extensive experiments across both single-modality tasks such as language understanding and generation and multi-modality tasks such as video-text understanding demonstrate that SSH outperforms existing parameter-efficient fine-tuning (PEFT) methods while achieving substantial reductions in computational cost and memory requirements.
Authors: Xiaoyang Liu, Kangjie Bao, Jiashuo Zhang, Yunqi Liu, Yu Chen, Yuntian Liu, Yang Jiao, Tao Luo
Abstract: Autoformalization, the process of automatically translating natural language mathematics into machine-verifiable formal language, has demonstrated advancements with the progress of large language models (LLMs). However, a key obstacle to further advancements is the scarcity of paired datasets that align natural language with formal language. To address this challenge, we introduce ATLAS (Autoformalizing Theorems through Lifting, Augmentation, and Synthesis of Data), an iterative data generation framework designed to produce large-scale, high-quality parallel theorem statements. With the proposed ATLAS running for 10 iterations, we construct an undergraduate-level dataset comprising 300k theorem statements and develop the ATLAS translator, achieving accuracies of 80.59% (pass@8) and 92.99% (pass@128) on ProofNet, significantly outperforming the base model (23.99% and 47.17%) and InternLM2-Math-Plus-7B (50.94% and 80.32%). Furthermore, the ATLAS translator also achieves state-of-the-art performance on both the high-school-level miniF2F dataset and the graduate-level MathQual dataset introduced in this work. The datasets, model, and code will be released to the public soon.
Authors: Marian Lupascu, Ana-Cristina Rogoz, Mihai Sorin Stupariu, Radu Tudor Ionescu
Abstract: In this survey, we systematically analyze techniques used to adapt large multimodal models (LMMs) for low-resource (LR) languages, examining approaches ranging from visual enhancement and data creation to cross-modal transfer and fusion strategies. Through a comprehensive analysis of 106 studies across 75 LR languages, we identify key patterns in how researchers tackle the challenges of limited data and computational resources. We find that visual information often serves as a crucial bridge for improving model performance in LR settings, though significant challenges remain in areas such as hallucination mitigation and computational efficiency. We aim to provide researchers with a clear understanding of current approaches and remaining challenges in making LMMs more accessible to speakers of LR (understudied) languages. We complement our survey with an open-source repository available at: https://github.com/marianlupascu/LMM4LRL-Survey.
Authors: Beining Zhang, Aditya Kapoor, Mingfei Sun
Abstract: Multi-agent reinforcement learning (MARL) often relies on \emph{parameter sharing (PS)} to scale efficiently. However, purely shared policies can stifle each agent's unique specialization, reducing overall performance in heterogeneous environments. We propose \textbf{Low-Rank Agent-Specific Adaptation (LoRASA)}, a novel approach that treats each agent's policy as a specialized ``task'' fine-tuned from a shared backbone. Drawing inspiration from parameter-efficient transfer methods, LoRASA appends small, low-rank adaptation matrices to each layer of the shared policy, naturally inducing \emph{parameter-space sparsity} that promotes both specialization and scalability. We evaluate LoRASA on challenging benchmarks including the StarCraft Multi-Agent Challenge (SMAC) and Multi-Agent MuJoCo (MAMuJoCo), implementing it atop widely used algorithms such as MAPPO and A2PO. Across diverse tasks, LoRASA matches or outperforms existing baselines \emph{while reducing memory and computational overhead}. Ablation studies on adapter rank, placement, and timing validate the method's flexibility and efficiency. Our results suggest LoRASA's potential to establish a new norm for MARL policy parameterization: combining a shared foundation for coordination with low-rank agent-specific refinements for individual specialization.
Authors: Rahul Vaze, Abhishek Sinha
Abstract: Auto-bidding problem under a strict return-on-spend constraint (ROSC) is considered, where an algorithm has to make decisions about how much to bid for an ad slot depending on the revealed value, and the hidden allocation and payment function that describes the probability of winning the ad-slot depending on its bid. The objective of an algorithm is to maximize the expected utility (product of ad value and probability of winning the ad slot) summed across all time slots subject to the total expected payment being less than the total expected utility, called the ROSC. A (surprising) impossibility result is derived that shows that no online algorithm can achieve a sub-linear regret even when the value, allocation and payment function are drawn i.i.d. from an unknown distribution. The problem is non-trivial even when the revealed value remains constant across time slots, and an algorithm with regret guarantee that is optimal up to logarithmic factor is derived.
Authors: Yongcheng Zeng, Xinyu Cui, Xuanfa Jin, Guoqing Liu, Zexu Sun, Quan He, Dong Li, Ning Yang, Jianye Hao, Haifeng Zhang, Jun Wang
Abstract: A truly intelligent Large Language Model (LLM) should be capable of correcting errors in its responses through external interactions. However, even the most advanced models often face challenges in improving their outputs. In this paper, we explore how to cultivate LLMs with the self-refinement capability through iterative preference training, and how this ability can be leveraged to improve model performance during inference. To this end, we introduce a novel post-training and inference framework, called ARIES: Adaptive Refinement and Iterative Enhancement Structure. This method iteratively performs preference training and self-refinement-based data collection. During training, ARIES strengthen the model's direct question-answering capability while simultaneously unlocking its self-refinement potential. During inference, ARIES harnesses this self-refinement capability to generate a series of progressively refined responses, which are then filtered using either the Reward Model Scoring or a simple yet effective Rule-Based Selection mechanism, specifically tailored to our approach, to construct a dataset for the next round of preference training. Experimental results demonstrate the remarkable performance of ARIES. When applied to the Llama-3.1-8B model and under the self-refinement setting, ARIES surpasses powerful models such as GPT-4o, achieving 62.3% length-controlled (LC) and a 63.3% raw win rates on AlpacaEval 2, outperforming Iterative DPO by 27.8% and 35.5% respectively, as well as a 50.3% win rate on Arena-Hard, surpassing Iterative DPO by 26.6%. Furthermore, ARIES consistently enhances performance on mathematical reasoning tasks like GSM8K and MATH.
Authors: Alessio Benavoli, Dario Piga, Marco Forgione, Marco Zaffalon
Abstract: In this work, we present a novel approach to system identification for dynamical systems, based on a specific class of Deep Gaussian Processes (Deep GPs). These models are constructed by interconnecting linear dynamic GPs (equivalent to stochastic linear time-invariant dynamical systems) and static GPs (to model static nonlinearities). Our approach combines the strengths of data-driven methods, such as those based on neural network architectures, with the ability to output a probability distribution. This offers a more comprehensive framework for system identification that includes uncertainty quantification. Using both simulated and real-world data, we demonstrate the effectiveness of the proposed approach.
Authors: Andre Wibisono
Abstract: We study the mixing time guarantee for sampling in relative Fisher information via the Proximal Sampler algorithm, which is an approximate proximal discretization of the Langevin dynamics. We show that when the target probability distribution is strongly log-concave, the relative Fisher information converges exponentially fast along the Proximal Sampler; this matches the exponential convergence rate of the relative Fisher information along the continuous-time Langevin dynamics for strongly log-concave target. When combined with a standard implementation of the Proximal Sampler via rejection sampling, this exponential convergence rate provides a high-accuracy iteration complexity guarantee for the Proximal Sampler in relative Fisher information when the target distribution is strongly log-concave and log-smooth. Our proof proceeds by establishing a strong data processing inequality for relative Fisher information along the Gaussian channel under strong log-concavity, and a data processing inequality along the reverse Gaussian channel for a special distribution. The forward and reverse Gaussian channels compose to form the Proximal Sampler, and these data processing inequalities imply the exponential convergence rate of the relative Fisher information along the Proximal Sampler.
Authors: Chun-Yi Kuan, Hung-yi Lee
Abstract: Recent advancements in controllable expressive speech synthesis, especially in text-to-speech (TTS) models, have allowed for the generation of speech with specific styles guided by textual descriptions, known as style prompts. While this development enhances the flexibility and naturalness of synthesized speech, there remains a significant gap in understanding how these models handle vague or abstract style prompts. This study investigates the potential gender bias in how models interpret occupation-related prompts, specifically examining their responses to instructions like "Act like a nurse". We explore whether these models exhibit tendencies to amplify gender stereotypes when interpreting such prompts. Our experimental results reveal the model's tendency to exhibit gender bias for certain occupations. Moreover, models of different sizes show varying degrees of this bias across these occupations.
Authors: Miroslav \v{S}trupl, Oleg Szehr, Francesco Faccio, Dylan R. Ashley, Rupesh Kumar Srivastava, J\"urgen Schmidhuber
Abstract: This article provides a rigorous analysis of convergence and stability of Episodic Upside-Down Reinforcement Learning, Goal-Conditioned Supervised Learning and Online Decision Transformers. These algorithms performed competitively across various benchmarks, from games to robotic tasks, but their theoretical understanding is limited to specific environmental conditions. This work initiates a theoretical foundation for algorithms that build on the broad paradigm of approaching reinforcement learning through supervised learning or sequence modeling. At the core of this investigation lies the analysis of conditions on the underlying environment, under which the algorithms can identify optimal solutions. We also assess whether emerging solutions remain stable in situations where the environment is subject to tiny levels of noise. Specifically, we study the continuity and asymptotic convergence of command-conditioned policies, values and the goal-reaching objective depending on the transition kernel of the underlying Markov Decision Process. We demonstrate that near-optimal behavior is achieved if the transition kernel is located in a sufficiently small neighborhood of a deterministic kernel. The mentioned quantities are continuous (with respect to a specific topology) at deterministic kernels, both asymptotically and after a finite number of learning cycles. The developed methods allow us to present the first explicit estimates on the convergence and stability of policies and values in terms of the underlying transition kernels. On the theoretical side we introduce a number of new concepts to reinforcement learning, like working in segment spaces, studying continuity in quotient topologies and the application of the fixed-point theory of dynamical systems. The theoretical study is accompanied by a detailed investigation of example environments and numerical experiments.
Authors: Lars van der Laan, Ahmed Alaa
Abstract: Ensuring model calibration is critical for reliable predictions, yet popular distribution-free methods, such as histogram binning and isotonic regression, provide only asymptotic guarantees. We introduce a unified framework for Venn and Venn-Abers calibration, generalizing Vovk's binary classification approach to arbitrary prediction tasks and loss functions. Venn calibration leverages binning calibrators to construct prediction sets that contain at least one marginally perfectly calibrated point prediction in finite samples, capturing epistemic uncertainty in the calibration process. The width of these sets shrinks asymptotically to zero, converging to a conditionally calibrated point prediction. Furthermore, we propose Venn multicalibration, a novel methodology for finite-sample calibration across subpopulations. For quantile loss, group-conditional and multicalibrated conformal prediction arise as special cases of Venn multicalibration, and Venn calibration produces novel conformal prediction intervals that achieve quantile-conditional coverage. As a separate contribution, we extend distribution-free conditional calibration guarantees of histogram binning and isotonic calibration to general losses.
Authors: Wenhao Ding, Sushant Veer, Karen Leung, Yulong Cao, Marco Pavone
Abstract: Validating the safety and performance of an autonomous vehicle (AV) requires benchmarking on real-world driving logs. However, typical driving logs contain mostly uneventful scenarios with minimal interactions between road users. Identifying interactive scenarios in real-world driving logs enables the curation of datasets that amplify critical signals and provide a more accurate assessment of an AV's performance. In this paper, we present a novel metric that identifies interactive scenarios by measuring an AV's surprise potential on others. First, we identify three dimensions of the design space to describe a family of surprise potential measures. Second, we exhaustively evaluate and compare different instantiations of the surprise potential measure within this design space on the nuScenes dataset. To determine how well a surprise potential measure correctly identifies an interactive scenario, we use a reward model learned from human preferences to assess alignment with human intuition. Our proposed surprise potential, arising from this exhaustive comparative study, achieves a correlation of more than 0.82 with the human-aligned reward function, outperforming existing approaches. Lastly, we validate motion planners on curated interactive scenarios to demonstrate downstream applications.
Authors: Shaoshuai Du, Yiyi Tao, Yixian Shen, Hang Zhang, Yanxin Shen, Xinyu Qiu, Chuanqi Shi
Abstract: This study investigates the performance of various large language models (LLMs) on zero-shot end-to-end relation extraction (RE) in Chinese, a task that integrates entity recognition and relation extraction without requiring annotated data. While LLMs show promise for RE, most prior work focuses on English or assumes pre-annotated entities, leaving their effectiveness in Chinese RE largely unexplored. To bridge this gap, we evaluate ChatGPT, Gemini, and LLaMA based on accuracy, efficiency, and adaptability. ChatGPT demonstrates the highest overall performance, balancing precision and recall, while Gemini achieves the fastest inference speed, making it suitable for real-time applications. LLaMA underperforms in both accuracy and latency, highlighting the need for further adaptation. Our findings provide insights into the strengths and limitations of LLMs for zero-shot Chinese RE, shedding light on trade-offs between accuracy and efficiency. This study serves as a foundation for future research aimed at improving LLM adaptability to complex linguistic tasks in Chinese NLP.
Authors: Zijiang Yan, Jianhua Pei, Hongda Wu, Hina Tabassum, Ping Wang
Abstract: This paper proposes a novel framework for real-time adaptive-bitrate video streaming by integrating latent diffusion models (LDMs) within the FFmpeg techniques. This solution addresses the challenges of high bandwidth usage, storage inefficiencies, and quality of experience (QoE) degradation associated with traditional constant bitrate streaming (CBS) and adaptive bitrate streaming (ABS). The proposed approach leverages LDMs to compress I-frames into a latent space, offering significant storage and semantic transmission savings without sacrificing high visual quality. While it keeps B-frames and P-frames as adjustment metadata to ensure efficient video reconstruction at the user side, the proposed framework is complemented with the most state-of-the-art denoising and video frame interpolation (VFI) techniques. These techniques mitigate semantic ambiguity and restore temporal coherence between frames, even in noisy wireless communication environments. Experimental results demonstrate the proposed method achieves high-quality video streaming with optimized bandwidth usage, outperforming state-of-the-art solutions in terms of QoE and resource efficiency. This work opens new possibilities for scalable real-time video streaming in 5G and future post-5G networks.
Authors: Anupama Sridhar, Alexander Johansen
Abstract: Theoretical work on Temporal Difference (TD) learning has provided finite-sample and high-probability guarantees for data generated from Markov chains. However, these bounds typically require linear function approximation, instance-dependent step sizes, algorithmic modifications, and restrictive mixing rates. We present theoretical findings for TD learning under more applicable assumptions, including instance-independent step sizes, full data utilization, and polynomial ergodicity, applicable to both linear and non-linear functions. \textbf{To our knowledge, this is the first proof of TD(0) convergence on Markov data under universal and instance-independent step sizes.} While each contribution is significant on its own, their combination allows these bounds to be effectively utilized in practical application settings. Our results include bounds for linear models and non-linear under generalized gradients and H\"older continuity.
Authors: Kang Yang, Yuning Chen, Wan Du
Abstract: We present Generalizable Wireless Radiance Fields (GWRF), a framework for modeling wireless signal propagation at arbitrary 3D transmitter and receiver positions. Unlike previous methods that adapt vanilla Neural Radiance Fields (NeRF) from the optical to the wireless signal domain, requiring extensive per-scene training, GWRF generalizes effectively across scenes. First, a geometry-aware Transformer encoder-based wireless scene representation module incorporates information from geographically proximate transmitters to learn a generalizable wireless radiance field. Second, a neural-driven ray tracing algorithm operates on this field to automatically compute signal reception at the receiver. Experimental results demonstrate that GWRF outperforms existing methods on single scenes and achieves state-of-the-art performance on unseen scenes.
Authors: An Zhao, Moucheng Xu, Ahmed H. Shahin, Wim Wuyts, Mark G. Jones, Joseph Jacob, Daniel C. Alexander
Abstract: Understanding the progression trajectories of diseases is crucial for early diagnosis and effective treatment planning. This is especially vital for life-threatening conditions such as Idiopathic Pulmonary Fibrosis (IPF), a chronic, progressive lung disease with a prognosis comparable to many cancers. Computed tomography (CT) imaging has been established as a reliable diagnostic tool for IPF. Accurately predicting future CT scans of early-stage IPF patients can aid in developing better treatment strategies, thereby improving survival outcomes. In this paper, we propose 4D Vector Quantised Generative Adversarial Networks (4D-VQ-GAN), a model capable of generating realistic CT volumes of IPF patients at any time point. The model is trained using a two-stage approach. In the first stage, a 3D-VQ-GAN is trained to reconstruct CT volumes. In the second stage, a Neural Ordinary Differential Equation (ODE) based temporal model is trained to capture the temporal dynamics of the quantised embeddings generated by the encoder in the first stage. We evaluate different configurations of our model for generating longitudinal CT scans and compare the results against ground truth data, both quantitatively and qualitatively. For validation, we conduct survival analysis using imaging biomarkers derived from generated CT scans and achieve a C-index comparable to that of biomarkers derived from the real CT scans. The survival analysis results demonstrate the potential clinical utility inherent to generated longitudinal CT scans, showing that they can reliably predict survival outcomes.
Authors: Quinn Dougherty, Ronak Mehta
Abstract: We introduce the Formally Verified Automated Programming Progress Standards, or FVAPPS, a benchmark of 4715 samples for writing programs and proving their correctness, the largest formal verification benchmark, including 1083 curated and quality controlled samples. Previously, APPS provided a benchmark and dataset for programming puzzles to be completed in Python and checked against unit tests, of the kind seen in technical assessments in the software engineering industry. Building upon recent approaches for benchmarks in interactive theorem proving, we generalize the unit tests to Lean 4 theorems given without proof (i.e., using Lean's "sorry" keyword). On the 406 theorems of 100 randomly selected samples, Sonnet correctly proves 30% and Gemini correctly proves 18%. We challenge the machine learning and program synthesis communities to solve both each general purpose programming problem and its associated correctness specifications. The benchmark is available at https://huggingface.co/datasets/quinn-dougherty/fvapps.
URLs: https://huggingface.co/datasets/quinn-dougherty/fvapps.
Authors: Rabia Asghar, Simon Mooney, Eoin O Neill, Paul Hynds
Abstract: Around 50 percent of Irelands rural population relies on unregulated private wells vulnerable to agricultural runoff and untreated wastewater. High national rates of Shiga toxin-producing Escherichia coli (STEC) and other waterborne illnesses have been linked to well water exposure. Periodic well testing is essential for public health, yet the lack of government incentives places the financial burden on households. Understanding environmental, cognitive, and material factors influencing well-testing behavior is critical. This study employs Agent-Based Modeling (ABM) to simulate policy interventions based on national survey data. The ABM framework, designed for private well-testing behavior, integrates a Deep Q-network reinforcement learning model and Explainable AI (XAI) for decision-making insights. Key features were selected using Recursive Feature Elimination (RFE) with 10-fold cross-validation, while SHAP (Shapley Additive Explanations) provided further interpretability for policy recommendations. Fourteen policy scenarios were tested. The most effective, Free Well Testing plus Communication Campaign, increased participation to 435 out of 561 agents, from a baseline of approximately 5 percent, with rapid behavioral adaptation. Free Well Testing plus Regulation also performed well, with 433 out of 561 agents initiating well testing. Free testing alone raised participation to over 75 percent, with some agents testing multiple times annually. Scenarios with free well testing achieved faster learning efficiency, converging in 1000 episodes, while others took 2000 episodes, indicating slower adaptation. This research demonstrates the value of ABM and XAI in public health policy, providing a framework for evaluating behavioral interventions in environmental health.
Authors: Danial Khatamsaz, Joseph Wagner, Brent Vela, Raymundo Arroyave, Douglas L. Allaire
Abstract: Accelerated discovery in materials science demands autonomous systems capable of dynamically formulating and solving design problems. In this work, we introduce a novel framework that leverages Bayesian optimization over a problem formulation space to identify optimal design formulations in line with decision-maker preferences. By mapping various design scenarios to a multi attribute utility function, our approach enables the system to balance conflicting objectives such as ductility, yield strength, density, and solidification range without requiring an exact problem definition at the outset. We demonstrate the efficacy of our method through an in silico case study on a Mo-Nb-Ti-V-W alloy system targeted for gas turbine engine blade applications. The framework converges on a sweet spot that satisfies critical performance thresholds, illustrating that integrating problem formulation discovery into the autonomous design loop can significantly streamline the experimental process. Future work will incorporate human feedback to further enhance the adaptability of the system in real-world experimental settings.
Authors: Cameron Armijo, Pablo Rivas
Abstract: This study examines the capabilities of the Vision Transformer (ViT) model in generating visual embeddings for images of auto parts sourced from online marketplaces, such as Craigslist and OfferUp. By focusing exclusively on single-modality data, the analysis evaluates ViT's potential for detecting patterns indicative of illicit activities. The workflow involves extracting high-dimensional embeddings from images, applying dimensionality reduction techniques like Uniform Manifold Approximation and Projection (UMAP) to visualize the embedding space, and using K-Means clustering to categorize similar items. Representative posts nearest to each cluster centroid provide insights into the composition and characteristics of the clusters. While the results highlight the strengths of ViT in isolating visual patterns, challenges such as overlapping clusters and outliers underscore the limitations of single-modal approaches in this domain. This work contributes to understanding the role of Vision Transformers in analyzing online marketplaces and offers a foundation for future advancements in detecting fraudulent or illegal activities.
Authors: Daniele Bracale, Moulinath Banerjee, Yuekai Sun, Kevin Stoll, Salam Turki
Abstract: We propose a shape-constrained approach to dynamic pricing for censored data in the linear valuation model that eliminates the need for tuning parameters commonly required in existing methods. Previous works have addressed the challenge of unknown market noise distribution F using strategies ranging from kernel methods to reinforcement learning algorithms, such as bandit techniques and upper confidence bounds (UCB), under the Lipschitz (and stronger) assumption(s) on $F_0$. In contrast, our method relies on isotonic regression under the weaker assumption that $F_0$ is $\alpha$-Holder continuous for some $\alpha \in (0,1]$. We obtain an upper bound on the asymptotic expected regret that matches existing bounds in the literature for $\alpha = 1$ (the Lipschitz case). Simulations and experiments with real-world data obtained by Welltower Inc (a major healthcare Real Estate Investment Trust) consistently demonstrate that our method attains better empirical regret in comparison to several existing methods in the literature while offering the advantage of being completely tuning-parameter free.
Authors: Ying Lei, Yancheng Cao, Will Wang, Yuanzhe Dong, Changchang Yin, Weidan Cao, Ping Zhang, Jingzhen Yang, Bingsheng Yao, Yifan Peng, Chunhua Weng, Randy Auerbach, Lena Mamykina, Dakuo Wang, Yuntao Wang, Xuhai Xu
Abstract: While just-in-time interventions (JITIs) have effectively targeted common health behaviors, individuals often have unique needs to intervene in personal undesirable actions that can negatively affect physical, mental, and social well-being. We present WatchGuardian, a smartwatch-based JITI system that empowers users to define custom interventions for these personal actions with a small number of samples. For the model to detect new actions based on limited new data samples, we developed a few-shot learning pipeline that finetuned a pre-trained inertial measurement unit (IMU) model on public hand-gesture datasets. We then designed a data augmentation and synthesis process to train additional classification layers for customization. Our offline evaluation with 26 participants showed that with three, five, and ten examples, our approach achieved an average accuracy of 76.8%, 84.7%, and 87.7%, and an F1 score of 74.8%, 84.2%, and 87.2% We then conducted a four-hour intervention study to compare WatchGuardian against a rule-based intervention. Our results demonstrated that our system led to a significant reduction by 64.0 +- 22.6% in undesirable actions, substantially outperforming the baseline by 29.0%. Our findings underscore the effectiveness of a customizable, AI-driven JITI system for individuals in need of behavioral intervention in personal undesirable actions. We envision that our work can inspire broader applications of user-defined personalized intervention with advanced AI solutions.
Authors: Atsushi Nitanda, Anzelle Lee, Damian Tan Xing Kai, Mizuki Sakaguchi, Taiji Suzuki
Abstract: Mean-field Langevin dynamics (MFLD) is an optimization method derived by taking the mean-field limit of noisy gradient descent for two-layer neural networks in the mean-field regime. Recently, the propagation of chaos (PoC) for MFLD has gained attention as it provides a quantitative characterization of the optimization complexity in terms of the number of particles and iterations. A remarkable progress by Chen et al. (2022) showed that the approximation error due to finite particles remains uniform in time and diminishes as the number of particles increases. In this paper, by refining the defective log-Sobolev inequality -- a key result from that earlier work -- under the neural network training setting, we establish an improved PoC result for MFLD, which removes the exponential dependence on the regularization coefficient from the particle approximation term of the optimization complexity. As an application, we propose a PoC-based model ensemble strategy with theoretical guarantees.
Authors: Nasser A Alsadhan
Abstract: Alzheimer's disease is an untreatable, progressive brain disorder that slowly robs people of their memory, thinking abilities, and ultimately their capacity to complete even the most basic tasks. Among older adults, it is the most frequent cause of dementia. Although there is presently no treatment for Alzheimer's disease, scientific trials are ongoing to discover drugs to combat the condition. Treatments to slow the signs of dementia are also available. Many researchers throughout the world became interested in developing computer-aided diagnosis systems to aid in the early identification of this deadly disease and assure an accurate diagnosis. In particular, image based approaches have been coupled with machine learning techniques to address the challenges of Alzheimer's disease detection. This study proposes a computer aided diagnosis system to detect Alzheimer's disease from biomarkers captured using neuroimaging techniques. The proposed approach relies on deep learning techniques to extract the relevant visual features from the image collection to accurately predict the Alzheimer's class value. In the experiments, standard datasets and pre-trained deep learning models were investigated. Moreover, standard performance measures were used to assess the models' performances. The obtained results proved that VGG16-based models outperform the state of the art performance.
Authors: Shubham Kumar Nigam, Tanmay Dubey, Govind Sharma, Noel Shallum, Kripabandhu Ghosh, Arnab Bhattacharya
Abstract: In this paper, we address the task of semantic segmentation of legal documents through rhetorical role classification, with a focus on Indian legal judgments. We introduce LegalSeg, the largest annotated dataset for this task, comprising over 7,000 documents and 1.4 million sentences, labeled with 7 rhetorical roles. To benchmark performance, we evaluate multiple state-of-the-art models, including Hierarchical BiLSTM-CRF, TransformerOverInLegalBERT (ToInLegalBERT), Graph Neural Networks (GNNs), and Role-Aware Transformers, alongside an exploratory RhetoricLLaMA, an instruction-tuned large language model. Our results demonstrate that models incorporating broader context, structural relationships, and sequential sentence information outperform those relying solely on sentence-level features. Additionally, we conducted experiments using surrounding context and predicted or actual labels of neighboring sentences to assess their impact on classification accuracy. Despite these advancements, challenges persist in distinguishing between closely related roles and addressing class imbalance. Our work underscores the potential of advanced techniques for improving legal document understanding and sets a strong foundation for future research in legal NLP.
Authors: Zhiqiang Que, Jose G. F. Coutinho, Ce Guo, Hongxiang Fan, Wayne Luk
Abstract: This paper presents a unified framework for codifying and automating optimization strategies to efficiently deploy deep neural networks (DNNs) on resource-constrained hardware, such as FPGAs, while maintaining high performance, accuracy, and resource efficiency. Deploying DNNs on such platforms involves addressing the significant challenge of balancing performance, resource usage (e.g., DSPs and LUTs), and inference accuracy, which often requires extensive manual effort and domain expertise. Our novel approach addresses two key issues: cross-stage co-optimization and optimization search. By seamlessly integrating programmatic DNN optimization techniques with high-level synthesis (HLS)-based metaprogramming and leveraging advanced design space exploration (DSE) strategies like Bayesian optimization, the framework automates both top-down and bottom-up design flows, reducing the need for manual intervention and domain expertise. The proposed framework introduces customizable optimization, transformation, and control blocks to enhance DNN accelerator performance and resource efficiency. Experimental results demonstrate up to a 92\% DSP and 89\% LUT usage reduction for select networks, while preserving accuracy, along with a 15.6-fold reduction in optimization time compared to grid search. These results underscore the novelty and potential of the proposed framework for automated, resource-efficient DNN accelerator designs.
Authors: Lu Chen, Yizhou Wang, Shixiang Tang, Qianhong Ma, Tong He, Wanli Ouyang, Xiaowei Zhou, Hujun Bao, Sida Peng
Abstract: This paper addresses the task of learning an agent model behaving like humans, which can jointly perceive, predict, and act in egocentric worlds. Previous methods usually train separate models for these three abilities, leading to information silos among them, which prevents these abilities from learning from each other and collaborating effectively. In this paper, we propose a joint predictive agent model, named EgoAgent, that simultaneously learns to represent the world, predict future states, and take reasonable actions with a single transformer. EgoAgent unifies the representational spaces of the three abilities by mapping them all into a sequence of continuous tokens. Learnable query tokens are appended to obtain current states, future states, and next actions. With joint supervision, our agent model establishes the internal relationship among these three abilities and effectively mimics the human inference and learning processes. Comprehensive evaluations of EgoAgent covering image classification, egocentric future state prediction, and 3D human motion prediction tasks demonstrate the superiority of our method. The code and trained model will be released for reproducibility.
Authors: Zhifei Yang, Keyang Lu, Chao Zhang, Jiaxing Qi, Hanqi Jiang, Ruifei Ma, Shenglin Yin, Yifan Xu, Mingzhe Xing, Zhen Xiao, Jieyi Long, Xiangde Liu, Guangyao Zhai
Abstract: Controllable 3D scene generation has extensive applications in virtual reality and interior design, where the generated scenes should exhibit high levels of realism and controllability in terms of geometry. Scene graphs provide a suitable data representation that facilitates these applications. However, current graph-based methods for scene generation are constrained to text-based inputs and exhibit insufficient adaptability to flexible user inputs, hindering the ability to precisely control object geometry. To address this issue, we propose MMGDreamer, a dual-branch diffusion model for scene generation that incorporates a novel Mixed-Modality Graph, visual enhancement module, and relation predictor. The mixed-modality graph allows object nodes to integrate textual and visual modalities, with optional relationships between nodes. It enhances adaptability to flexible user inputs and enables meticulous control over the geometry of objects in the generated scenes. The visual enhancement module enriches the visual fidelity of text-only nodes by constructing visual representations using text embeddings. Furthermore, our relation predictor leverages node representations to infer absent relationships between nodes, resulting in more coherent scene layouts. Extensive experimental results demonstrate that MMGDreamer exhibits superior control of object geometry, achieving state-of-the-art scene generation performance. Project page: https://yangzhifeio.github.io/project/MMGDreamer.
Authors: Idan Achituve, Hai Victor Habi, Amir Rosenfeld, Arnon Netzer, Idit Diamant, Ethan Fetaya
Abstract: In image processing, solving inverse problems is the task of finding plausible reconstructions of an image that was corrupted by some (usually known) degradation model. Commonly, this process is done using a generative image model that can guide the reconstruction towards solutions that appear natural. The success of diffusion models over the last few years has made them a leading candidate for this task. However, the sequential nature of diffusion models makes this conditional sampling process challenging. Furthermore, since diffusion models are often defined in the latent space of an autoencoder, the encoder-decoder transformations introduce additional difficulties. Here, we suggest a novel sampling method based on sequential Monte Carlo (SMC) in the latent space of diffusion models. We use the forward process of the diffusion model to add additional auxiliary observations and then perform an SMC sampling as part of the backward process. Empirical evaluations on ImageNet and FFHQ show the benefits of our approach over competing methods on various inverse problem tasks.
Authors: Aran Nayebi
Abstract: Under what conditions can capable AI agents efficiently align their actions with human preferences? More specifically, when they are proficient enough to collaborate with us, how long does coordination take, and when is it computationally feasible? These foundational questions of AI alignment help define what makes an AI agent ``sufficiently safe'' and valuable to humans. Since such generally capable systems do not yet exist, a theoretical analysis is needed to establish when guarantees hold -- and what they even are. We introduce a game-theoretic framework that generalizes prior alignment approaches with fewer assumptions, allowing us to analyze the computational complexity of alignment across $M$ objectives and $N$ agents, providing both upper and lower bounds. Unlike previous work, which often assumes common priors, idealized communication, or implicit tractability, our framework formally characterizes the difficulty of alignment under minimal assumptions. Our main result shows that even when agents are fully rational and computationally \emph{unbounded}, alignment can be achieved with high probability in time \emph{linear} in the task space size. Therefore, in real-world settings, where task spaces are often \emph{exponential} in input length, this remains impractical. More strikingly, our lower bound demonstrates that alignment is \emph{impossible} to speed up when scaling to exponentially many tasks or agents, highlighting a fundamental computational barrier to scalable alignment. Relaxing these idealized assumptions, we study \emph{computationally bounded} agents with noisy messages (representing obfuscated intent), showing that while alignment can still succeed with high probability, it incurs additional \emph{exponential} slowdowns in the task space size, number of agents, and number of tasks. We conclude by identifying conditions that make alignment more feasible.
Authors: Md. Saif Hassan Onim, Himanshu Thapliyal
Abstract: The widespread use of cloud-based medical devices and wearable sensors has made physiological data susceptible to tampering. These attacks can compromise the reliability of healthcare systems which can be critical and life-threatening. Detection of such data tampering is of immediate need. Machine learning has been used to detect anomalies in datasets but the performance of Quantum Machine Learning (QML) is still yet to be evaluated for physiological sensor data. Thus, our study compares the effectiveness of QML for detecting physiological data tampering, focusing on two types of white-box attacks: data poisoning and adversarial perturbation. The results show that QML models are better at identifying label-flipping attacks, achieving accuracy rates of 75%-95% depending on the data and attack severity. This superior performance is due to the ability of quantum algorithms to handle complex and high-dimensional data. However, both QML and classical models struggle to detect more sophisticated adversarial perturbation attacks, which subtly alter data without changing its statistical properties. Although QML performed poorly against this attack with around 45%-65% accuracy, it still outperformed classical algorithms in some cases.
Authors: Yingying Fan, Lan Gao, Jinchi Lv, Xiaocong Xu
Abstract: We propose a unified theoretical framework for studying the robustness of the model-X knockoffs framework by investigating the asymptotic false discovery rate (FDR) control of the practically implemented approximate knockoffs procedure. This procedure deviates from the model-X knockoffs framework by substituting the true covariate distribution with a user-specified distribution that can be learned using in-sample observations. By replacing the distributional exchangeability condition of the model-X knockoff variables with three conditions on the approximate knockoff statistics, we establish that the approximate knockoffs procedure achieves the asymptotic FDR control. Using our unified framework, we further prove that an arguably most popularly used knockoff variable generation method--the Gaussian knockoffs generator based on the first two moments matching--achieves the asymptotic FDR control when the two-moment-based knockoff statistics are employed in the knockoffs inference procedure. For the first time in the literature, our theoretical results justify formally the effectiveness and robustness of the Gaussian knockoffs generator. Simulation and real data examples are conducted to validate the theoretical findings.
Authors: Alessandro Micheli, M\'elodie Monod, Samir Bhatt
Abstract: Diffusion models have emerged as powerful tools for solving inverse problems, yet prior work has primarily focused on observations with Gaussian measurement noise, restricting their use in real-world scenarios. This limitation persists due to the intractability of the likelihood score, which until now has only been approximated in the simpler case of Gaussian likelihoods. In this work, we extend diffusion models to handle inverse problems where the observations follow a distribution from the exponential family, such as a Poisson or a Binomial distribution. By leveraging the conjugacy properties of exponential family distributions, we introduce the evidence trick, a method that provides a tractable approximation to the likelihood score. In our experiments, we demonstrate that our methodology effectively performs Bayesian inference on spatially inhomogeneous Poisson processes with intensities as intricate as ImageNet images. Furthermore, we demonstrate the real-world impact of our methodology by showing that it performs competitively with the current state-of-the-art in predicting malaria prevalence estimates in Sub-Saharan Africa.
Authors: Rahul Porwal, Alice Rozet, Pryce Houck, Jotsna Gowda, Sarah Moeller, Kevin Tang
Abstract: African American English (AAE) presents unique challenges in natural language processing (NLP). This research systematically compares the performance of available NLP models--rule-based, transformer-based, and large language models (LLMs)--capable of identifying key grammatical features of AAE, namely Habitual Be and Multiple Negation. These features were selected for their distinct grammatical complexity and frequency of occurrence. The evaluation involved sentence-level binary classification tasks, using both zero-shot and few-shot strategies. The analysis reveals that while LLMs show promise compared to the baseline, they are influenced by biases such as recency and unrelated features in the text such as formality. This study highlights the necessity for improved model training and architectural adjustments to better accommodate AAE's unique linguistic characteristics. Data and code are available.
Authors: Yihan He, Hong-Yu Chen, Yuan Cao, Jianqing Fan, Han Liu
Abstract: LLMs demonstrate significant inference capacities in complicated machine learning tasks, using the Transformer model as its backbone. Motivated by the limited understanding of such models on the unsupervised learning problems, we study the learning guarantees of Transformers in performing multi-class clustering of the Gaussian Mixture Models. We develop a theory drawing strong connections between the Softmax Attention layers and the workflow of the EM algorithm on clustering the mixture of Gaussians. Our theory provides approximation bounds for the Expectation and Maximization steps by proving the universal approximation abilities of multivariate mappings by Softmax functions. In addition to the approximation guarantees, we also show that with a sufficient number of pre-training samples and an initialization, Transformers can achieve the minimax optimal rate for the problem considered. Our extensive simulations empirically verified our theory by revealing the strong learning capacities of Transformers even beyond the assumptions in the theory, shedding light on the powerful inference capacities of LLMs.
Authors: Muhammad Faaiz Taufiq
Abstract: Off-policy evaluation (OPE) is a critical challenge in robust decision-making that seeks to assess the performance of a new policy using data collected under a different policy. However, the existing OPE methodologies suffer from several limitations arising from statistical uncertainty as well as causal considerations. In this thesis, we address these limitations by presenting three different works. Firstly, we consider the problem of high variance in the importance-sampling-based OPE estimators. We introduce the Marginal Ratio (MR) estimator, a novel OPE method that reduces variance by focusing on the marginal distribution of outcomes rather than direct policy shifts, improving robustness in contextual bandits. Next, we propose Conformal Off-Policy Prediction (COPP), a principled approach for uncertainty quantification in OPE that provides finite-sample predictive intervals, ensuring robust decision-making in risk-sensitive applications. Finally, we address causal unidentifiability in off-policy decision-making by developing novel bounds for sequential decision settings, which remain valid under arbitrary unmeasured confounding. We apply these bounds to assess the reliability of digital twin models, introducing a falsification framework to identify scenarios where model predictions diverge from real-world behaviour. Our contributions provide new insights into robust decision-making under uncertainty and establish principled methods for evaluating policies in both static and dynamic settings.
Authors: Tom Szwagier, Xavier Pennec
Abstract: Many machine learning methods look for low-dimensional representations of the data. The underlying subspace can be estimated by first choosing a dimension $q$ and then optimizing a certain objective function over the space of $q$-dimensional subspaces (the Grassmannian). Trying different $q$ yields in general non-nested subspaces, which raises an important issue of consistency between the data representations. In this paper, we propose a simple trick to enforce nestedness in subspace learning methods. It consists in lifting Grassmannian optimization problems to flag manifolds (the space of nested subspaces of increasing dimension) via nested projectors. We apply the flag trick to several classical machine learning methods and show that it successfully addresses the nestedness issue.
Authors: Safaa Menssouri, El Mehdi Amhoud
Abstract: Internet of things (IoT) networks, boosted by 6G technology, are transforming various industries. However, their widespread adoption introduces significant security risks, particularly in detecting rare but potentially damaging cyber-attacks. This makes the development of robust IDS crucial for monitoring network traffic and ensuring their safety. Traditional IDS often struggle with detecting rare attacks due to severe class imbalances in IoT data. In this paper, we propose a novel two-stage system called conditional tabular generative synthetic minority data generation with deep neural network (CTGSM-DNN). In the first stage, a conditional tabular generative adversarial network (CTGAN) is employed to generate synthetic data for rare attack classes. In the second stage, the SMOTEENN method is applied to improve dataset quality. The full study was conducted using the CSE-CIC-IDS2018 dataset, and we assessed the performance of the proposed IDS using different evaluation metrics. The experimental results demonstrated the effectiveness of the proposed multiclass classifier, achieving an overall accuracy of 99.90% and 80% accuracy in detecting rare attacks.
Authors: Getoar Sopa, Juraj Marusic, Marco Avella-Medina, John P. Cunningham
Abstract: In recent years, there has been much work on scaling Bayesian Optimization to high-dimensional problems, for example hyperparameter tuning in large neural network models. These scalable methods have been successful, finding high objective values much more quickly than traditional global Bayesian Optimization or random search-based methods. At the same time, these large neural network models often use sensitive data, but preservation of Differential Privacy has not scaled alongside these modern Bayesian Optimization procedures. Here we develop a method to privately estimate potentially high-dimensional parameter spaces using Gradient Informative Bayesian Optimization. Our theoretical results prove that under suitable conditions, our method converges exponentially fast to a ball around the optimal parameter configuration. Moreover, regardless of whether the assumptions are satisfied, we show that our algorithm maintains privacy and empirically demonstrates superior performance to existing methods in the high-dimensional hyperparameter setting.
Authors: Bidipta Sarkar, Warren Xia, C. Karen Liu, Dorsa Sadigh
Abstract: Communicating in natural language is a powerful tool in multi-agent settings, as it enables independent agents to share information in partially observable settings and allows zero-shot coordination with humans. However, most prior works are limited as they either rely on training with large amounts of human demonstrations or lack the ability to generate natural and useful communication strategies. In this work, we train language models to have productive discussions about their environment in natural language without any human demonstrations. We decompose the communication problem into listening and speaking. Our key idea is to leverage the agent's goal to predict useful information about the world as a dense reward signal that guides communication. Specifically, we improve a model's listening skills by training them to predict information about the environment based on discussions, and we simultaneously improve a model's speaking skills with multi-agent reinforcement learning by rewarding messages based on their influence on other agents. To investigate the role and necessity of communication in complex social settings, we study an embodied social deduction game based on Among Us, where the key question to answer is the identity of an adversarial imposter. We analyze emergent behaviors due to our technique, such as accusing suspects and providing evidence, and find that it enables strong discussions, doubling the win rates compared to standard RL. We release our code and models at https://socialdeductionllm.github.io/
Authors: David R. Burt, Renato Berlinghieri, Stephen Bates, Tamara Broderick
Abstract: Linear models remain ubiquitous in modern spatial applications - including climate science, public health, and economics - due to their interpretability, speed, and reproducibility. While practitioners generally report a form of uncertainty, popular spatial uncertainty quantification methods do not jointly handle model misspecification and distribution shift - despite both being essentially always present in spatial problems. In the present paper, we show that existing methods for constructing confidence (or credible) intervals in spatial linear models fail to provide correct coverage due to unaccounted-for bias. In contrast to classical methods that rely on an i.i.d. assumption that is inappropriate in spatial problems, in the present work we instead make a spatial smoothness (Lipschitz) assumption. We are then able to propose a new confidence-interval construction that accounts for bias in the estimation procedure. We demonstrate that our new method achieves nominal coverage via both theory and experiments. Code to reproduce experiments is available at https://github.com/DavidRBurt/Lipschitz-Driven-Inference.
URLs: https://github.com/DavidRBurt/Lipschitz-Driven-Inference.
Authors: Aytijhya Saha, Aaditya Ramdas
Abstract: This paper addresses a fundamental but largely unexplored challenge in sequential changepoint analysis: conducting inference following a detected change. We study the problem of localizing the changepoint using only the data observed up to a data-dependent stopping time at which a sequential detection algorithm $\mathcal A$ declares a change. We first construct confidence sets for the unknown changepoint when pre- and post-change distributions are assumed to be known. We then extend our framework to composite pre- and post-change scenarios. We impose no conditions on the observation space or on $\mathcal A$ -- we only need to be able to run $\mathcal A$ on simulated data sequences. In summary, this work offers both theoretically sound and practically effective tools for sequential changepoint localization.
Authors: Yijia Xiao, Runhui Wang, Luyang Kong, Davor Golac, Wei Wang
Abstract: The increasing complexity of computer science research projects demands more effective tools for deploying code repositories. Large Language Models (LLMs), such as Anthropic Claude and Meta Llama, have demonstrated significant advancements across various fields of computer science research, including the automation of diverse software engineering tasks. To evaluate the effectiveness of LLMs in handling complex code development tasks of research projects, particularly for NLP/CV/AI/ML/DM topics, we introduce CSR-Bench, a benchmark for Computer Science Research projects. This benchmark assesses LLMs from various aspects including accuracy, efficiency, and deployment script quality, aiming to explore their potential in conducting computer science research autonomously. We also introduce a novel framework, CSR-Agents, that utilizes multiple LLM agents to automate the deployment of GitHub code repositories of computer science research projects. Specifically, by checking instructions from markdown files and interpreting repository structures, the model generates and iteratively improves bash commands that set up the experimental environments and deploy the code to conduct research tasks. Preliminary results from CSR-Bench indicate that LLM agents can significantly enhance the workflow of repository deployment, thereby boosting developer productivity and improving the management of developmental workflows.
Authors: Hieu Trung Nguyen, Bao Nguyen, Binh Nguyen, Viet Anh Nguyen
Abstract: Modern language models (LMs) have significantly advanced generative modeling in natural language processing (NLP). Despite their success, LMs often struggle with adaptation to new contexts in real-time applications. A promising approach to task adaptation is activation intervention, which steers the LMs' generation process by identifying and manipulating the activations. However, existing interventions are highly dependent on heuristic rules or require many prompt inputs to determine effective interventions. This paper proposes a layer-wise additive activation intervention framework that optimizes the intervention process, thus enhancing the sample efficiency. We benchmark our framework on various datasets, demonstrating improvements in the accuracy of pre-trained LMs and competing intervention baselines.
Authors: Wonyoung Kim, Sungwoo Park, Garud Iyengar, Assaf Zeevi, Min-hwan Oh
Abstract: We introduce a novel linear bandit problem with partially observable features, resulting in partial reward information and spurious estimates. Without proper address for latent part, regret possibly grows linearly in decision horizon $T$, as their influence on rewards are unknown. To tackle this, we propose a novel analysis to handle the latent features and an algorithm that achieves sublinear regret. The core of our algorithm involves (i) augmenting basis vectors orthogonal to the observed feature space, and (ii) introducing an efficient doubly robust estimator. Our approach achieves a regret bound of $\tilde{O}(\sqrt{(d + d_h)T})$, where $d$ is the dimension of observed features, and $d_h$ is the unknown dimension of the subspace of the unobserved features. Notably, our algorithm requires no prior knowledge of the unobserved feature space, which may expand as more features become hidden. Numerical experiments confirm that our algorithm outperforms both non-contextual multi-armed bandits and linear bandit algorithms depending solely on observed features.
Authors: Ziyang Guo, Yifan Wu, Jason Hartline, Jessica Hullman
Abstract: Humans and AIs are often paired on decision tasks with the expectation of achieving complementary performance, where the combination of human and AI outperforms either one alone. However, how to improve performance of a human-AI team is often not clear without knowing more about what particular information and strategies each agent employs. We provide a decision-theoretic framework for characterizing the value of information -- and consequently, opportunities for agents to better exploit available information--in AI-assisted decision workflow. We demonstrate the use of the framework for model selection, empirical evaluation of human-AI performance, and explanation design. We propose a novel information-based instance-level explanation technique that adapts a conventional saliency-based explanation to explain information value in decision making.
Authors: Jianyu Xu, Yining Wang, Xi Chen, Yu-Xiang Wang
Abstract: We study an online dynamic pricing problem where the potential demand at each time period $t=1,2,\ldots, T$ is stochastic and dependent on the price. However, a perishable inventory is imposed at the beginning of each time $t$, censoring the potential demand if it exceeds the inventory level. To address this problem, we introduce a pricing algorithm based on the optimistic estimates of derivatives. We show that our algorithm achieves $\tilde{O}(\sqrt{T})$ optimal regret even with adversarial inventory series. Our findings advance the state-of-the-art in online decision-making problems with censored feedback, offering a theoretically optimal solution against adversarial observations.
Authors: Siqi Du, Hongsheng Huang, Kaixin Shen, Ziqi Liu, Shengjun Tang
Abstract: In Earth sciences, unobserved factors exhibit non-stationary spatial distributions, causing the relationships between features and targets to display spatial heterogeneity. In geographic machine learning tasks, conventional statistical learning methods often struggle to capture spatial heterogeneity, leading to unsatisfactory prediction accuracy and unreliable interpretability. While approaches like Geographically Weighted Regression (GWR) capture local variations, they fall short of uncovering global patterns and tracking the continuous evolution of spatial heterogeneity. Motivated by this limitation, we propose a novel perspective - that is, simultaneously modeling common features across different locations alongside spatial differences using deep neural networks. The proposed method is a dual-branch neural network with an encoder-decoder structure. In the encoding stage, the method aggregates node information in a spatiotemporal conditional graph using GCN and LSTM, encoding location-specific spatiotemporal heterogeneity as an implicit conditional vector. Additionally, a self-attention-based encoder is used to extract location-invariant common features from the data. In the decoding stage, the approach employs a conditional generation strategy that predicts response variables and interpretative weights based on data features under spatiotemporal conditions. The approach is validated by predicting vegetation gross primary productivity (GPP) using global climate and land cover data from 2001 to 2020. Trained on 50 million samples and tested on 2.8 million, the proposed model achieves an RMSE of 0.836, outperforming LightGBM (1.063) and TabNet (0.944). Visualization analyses indicate that our method can reveal the distribution differences of the dominant factors of GPP across various times and locations.
Authors: Tansheng Zhu, Hongyu Zhou, Ke Jin, Xusheng Xu, Qiufan Yuan, Lijie Ji
Abstract: Bayesian optimization is highly effective for optimizing expensive-to-evaluate black-box functions, but it faces significant computational challenges due to the high computational complexity of Gaussian processes, which results in a total time complexity that is quartic with respect to the number of iterations. To address this limitation, we propose the Bayesian Optimization by Kernel regression and density-based Exploration (BOKE) algorithm. BOKE uses kernel regression for efficient function approximation, kernel density for exploration, and the improved kernel regression upper confidence bound criteria to guide the optimization process, thus reducing computational costs to quadratic. Our theoretical analysis rigorously establishes the global convergence of BOKE and ensures its robustness. Through extensive numerical experiments on both synthetic and real-world optimization tasks, we demonstrate that BOKE not only performs competitively compared to Gaussian process-based methods but also exhibits superior computational efficiency. These results highlight BOKE's effectiveness in resource-constrained environments, providing a practical approach for optimization problems in engineering applications.
Authors: Yuchen He, Chihao Zhang
Abstract: We study the problem of sampling from a $d$-dimensional distribution with density $p(x)\propto e^{-f(x)}$, which does not necessarily satisfy good isoperimetric conditions. Specifically, we show that for any $L,M$ satisfying $LM\ge d\ge 5$, $\epsilon\in \left\{0,\frac{1}{32}\right\}$, and any algorithm with query accesses to the value of $f(x)$ and $\nabla f(x)$, there exists an $L$-log-smooth distribution with second moment at most $M$ such that the algorithm requires $\left\{\frac{LM}{d\epsilon}\right\}^{\Omega(d)}$ queries to compute a sample whose distribution is within $\epsilon$ in total variation distance to the target distribution. We complement the lower bound with an algorithm requiring $\left\{\frac{LM}{d\epsilon}\right\}^{\mathcal O(d)}$ queries, thereby characterizing the tight (up to the constant in the exponent) query complexity for sampling from the family of non-log-concave distributions. Our results are in sharp contrast with the recent work of Huang et al. (COLT'24), where an algorithm with quasi-polynomial query complexity was proposed for sampling from a non-log-concave distribution when $M=\mathtt{poly}(d)$. Their algorithm works under the stronger condition that all distributions along the trajectory of the Ornstein-Uhlenbeck process, starting from the target distribution, are $\mathcal O(1)$-log-smooth. We investigate this condition and prove that it is strictly stronger than requiring the target distribution to be $\mathcal O(1)$-log-smooth. Additionally, we study this condition in the context of mixtures of Gaussians. Finally, we place our results within the broader theme of ``sampling versus optimization'', as studied in Ma et al. (PNAS'19). We show that for a wide range of parameters, sampling is strictly easier than optimization by a super-exponential factor in the dimension $d$.
Authors: Guoxin Chen, Minpeng Liao, Peiying Yu, Dingmin Wang, Zile Qiao, Chao Yang, Xin Zhao, Kai Fan
Abstract: Retrieval-augmented generation (RAG) systems face a fundamental challenge in aligning independently developed retrievers and large language models (LLMs). Existing approaches typically involve modifying either component or introducing simple intermediate modules, resulting in practical limitations and sub-optimal performance. Inspired by human search behavior -- typically involving a back-and-forth process of proposing search queries and reviewing documents, we propose C-3PO, a proxy-centric framework that facilitates communication between retrievers and LLMs through a lightweight multi-agent system. Our framework implements three specialized agents that collaboratively optimize the entire RAG pipeline without altering the retriever and LLMs. These agents work together to assess the need for retrieval, generate effective queries, and select information suitable for the LLMs. To enable effective multi-agent coordination, we develop a tree-structured rollout approach for reward credit assignment in reinforcement learning. Extensive experiments in both in-domain and out-of-distribution scenarios demonstrate that C-3PO significantly enhances RAG performance while maintaining plug-and-play flexibility and superior generalization capabilities.
Authors: Rickard K. A. Karlsson, Jesse H. Krijthe
Abstract: A major challenge in estimating treatment effects in observational studies is the reliance on untestable conditions such as the assumption of no unmeasured confounding. In this work, we propose an algorithm that can falsify the assumption of no unmeasured confounding in a setting with observational data from multiple heterogeneous sources, which we refer to as environments. Our proposed falsification strategy leverages a key observation that unmeasured confounding can cause observed causal mechanisms to appear dependent. Building on this observation, we develop a novel two-stage procedure that detects these dependencies with high statistical power while controlling false positives. The algorithm does not require access to randomized data and, in contrast to other falsification approaches, functions even under transportability violations when the environment has a direct effect on the outcome of interest. To showcase the practical relevance of our approach, we show that our method is able to efficiently detect confounding on both simulated and real-world data.
Authors: Jiacheng Hu, Yanlin Xiang, Yang Lin, Junliang Du, Hanchao Zhang, Houze Liu
Abstract: This study introduces an AI-driven skin lesion classification algorithm built on an enhanced Transformer architecture, addressing the challenges of accuracy and robustness in medical image analysis. By integrating a multi-scale feature fusion mechanism and refining the self-attention process, the model effectively extracts both global and local features, enhancing its ability to detect lesions with ambiguous boundaries and intricate structures. Performance evaluation on the ISIC 2017 dataset demonstrates that the improved Transformer surpasses established AI models, including ResNet50, VGG19, ResNext, and Vision Transformer, across key metrics such as accuracy, AUC, F1-Score, and Precision. Grad-CAM visualizations further highlight the interpretability of the model, showcasing strong alignment between the algorithm's focus areas and actual lesion sites. This research underscores the transformative potential of advanced AI models in medical imaging, paving the way for more accurate and reliable diagnostic tools. Future work will explore the scalability of this approach to broader medical imaging tasks and investigate the integration of multimodal data to enhance AI-driven diagnostic frameworks for intelligent healthcare.
Authors: Zhichen Dong, Zhanhui Zhou, Zhixuan Liu, Chao Yang, Chaochao Lu
Abstract: In this work, we argue that large language models (LLMs), though trained to predict only the next token, exhibit emergent planning behaviors: $\textbf{their hidden representations encode future outputs beyond the next token}$. Through simple probing, we demonstrate that LLM prompt representations encode global attributes of their entire responses, including $\textit{structural attributes}$ (response length, reasoning steps), $\textit{content attributes}$ (character choices in storywriting, multiple-choice answers at the end of response), and $\textit{behavioral attributes}$ (answer confidence, factual consistency). In addition to identifying response planning, we explore how it scales with model size across tasks and how it evolves during generation. The findings that LLMs plan ahead for the future in their hidden representations suggests potential applications for improving transparency and generation control.
Authors: Wu Lin, Felix Dangel, Runa Eschenhagen, Juhan Bae, Richard E. Turner, Roger B. Grosse
Abstract: Many training methods, such as Adam(W) and Shampoo, learn a positive-definite curvature matrix and apply an inverse root before preconditioning. Recently, non-diagonal training methods, such as Shampoo, have gained significant attention; however, they remain computationally inefficient and are limited to specific types of curvature information due to the costly matrix root computation via matrix decomposition. To address this, we propose a Riemannian optimization approach that dynamically adapts spectral-factorized positive-definite curvature estimates, enabling the efficient application of arbitrary matrix roots and generic curvature learning. We demonstrate the efficacy and versatility of our approach in positive-definite matrix optimization and covariance adaptation for gradient-free optimization, as well as its efficiency in curvature learning for neural net training.
Authors: Utkarsh Tiwari, Aryan Seth, Adi Mukherjee, Kaavya Mer, Kavish, Dhruv Kumar
Abstract: We introduce DebateBench, a novel dataset consisting of an extensive collection of transcripts and metadata from some of the world's most prestigious competitive debates. The dataset consists of British Parliamentary debates from prestigious debating tournaments on diverse topics, annotated with detailed speech-level scores and house rankings sourced from official adjudication data. We curate 256 speeches across 32 debates with each debate being over 1 hour long with each input being an average of 32,000 tokens. Designed to capture long-context, large-scale reasoning tasks, DebateBench provides a benchmark for evaluating modern large language models (LLMs) on their ability to engage in argumentation, deliberation, and alignment with human experts. To do well on DebateBench, the LLMs must perform in-context learning to understand the rules and evaluation criteria of the debates, then analyze 8 seven minute long speeches and reason about the arguments presented by all speakers to give the final results. Our preliminary evaluation using GPT o1, GPT-4o, and Claude Haiku, shows that LLMs struggle to perform well on DebateBench, highlighting the need to develop more sophisticated techniques for improving their performance.
Authors: Xavier Vasques, Hanhee Paik, Laura Cif
Abstract: The functional characterization of different neuronal types has been a longstanding and crucial challenge. With the advent of physical quantum computers, it has become possible to apply quantum machine learning algorithms to translate theoretical research into practical solutions. Previous studies have shown the advantages of quantum algorithms on artificially generated datasets, and initial experiments with small binary classification problems have yielded comparable outcomes to classical algorithms. However, it is essential to investigate the potential quantum advantage using real-world data. To the best of our knowledge, this study is the first to propose the utilization of quantum systems to classify neuron morphologies, thereby enhancing our understanding of the performance of automatic multiclass neuron classification using quantum kernel methods. We examined the influence of feature engineering on classification accuracy and found that quantum kernel methods achieved similar performance to classical methods, with certain advantages observed in various configurations.
Authors: Haiduo Huang, Fuwei Yang, Zhenhua Liu, Yixing Xu, Jinze Li, Yang Liu, Xuanwu Yin, Dong Li, Pengju Ren, Emad Barsoum
Abstract: Speculative decoding (SD) accelerates large language model inference by using a smaller draft model to predict multiple tokens, which are then verified in parallel by the larger target model. However, the limited capacity of the draft model often necessitates tree-based sampling to improve prediction accuracy, where multiple candidates are generated at each step. We identify a key limitation in this approach: the candidates at the same step are derived from the same representation, limiting diversity and reducing overall effectiveness. To address this, we propose Jakiro, leveraging Mixture of Experts (MoE), where independent experts generate diverse predictions, effectively decoupling correlations among candidates. Furthermore, we introduce a hybrid inference strategy, combining autoregressive decoding for initial tokens with parallel decoding for subsequent stages, and enhance the latter with contrastive mechanism in features to improve accuracy. Our method significantly boosts prediction accuracy and achieves higher inference speedups. Extensive experiments across diverse models validate the effectiveness and robustness of our approach, establishing a new SOTA in speculative decoding. Our codes are available at https://github.com/haiduo/Jakiro.
Authors: Ivan Grega, F\'elix Therrien, Abhishek Soni, Karry Ocean, Kevan Dettelbach, Ribwar Ahmadi, Mehrdad Mokhtari, Curtis P. Berlinguette, Yoshua Bengio
Abstract: The electrochemical reduction of atmospheric CO$_2$ into high-energy molecules with renewable energy is a promising avenue for energy storage that can take advantage of existing infrastructure especially in areas where sustainable alternatives to fossil fuels do not exist. Automated laboratories are currently being developed and used to optimize the composition and operating conditions of gas diffusion electrodes (GDEs), the device in which this reaction takes place. Improving the efficiency of GDEs is crucial for this technology to become viable. Here we present a modeling framework to efficiently explore the high-dimensional parameter space of GDE designs in an active learning context. At the core of the framework is an uncertainty-aware physics model calibrated with experimental data. The model has the flexibility to capture various input parameter spaces and any carbon products which can be modeled with Tafel kinetics. It is interpretable, and a Gaussian process layer can capture deviations of real data from the function space of the physical model itself. We deploy the model in a simulated active learning setup with real electrochemical data gathered by the AdaCarbon automated laboratory and show that it can be used to efficiently traverse the multi-dimensional parameter space.
Authors: Michele Caprio, Yusuf Sale, Eyke H\"ullermeier
Abstract: Recently, Cella and Martin proved how, under an assumption called consonance, a credal set (i.e. a closed and convex set of probabilities) can be derived from the conformal transducer associated with transductive conformal prediction. We show that the Imprecise Highest Density Region (IHDR) associated with such a credal set corresponds to the classical Conformal Prediction Region. In proving this result, we relate the set of probability density/mass functions (pdf/pmf's) associated with the elements of the credal set to the imprecise probabilistic concept of a cloud. As a result, we establish new relationships between Conformal Prediction and Imprecise Probability (IP) theories. A byproduct of our presentation is the discovery that consonant plausibility functions are monoid homomorphisms, a new algebraic property of an IP tool.
Authors: Marco Rondina, Fabiana Vinci, Antonio Vetr\`o, Juan Carlos De Martin
Abstract: The ethical, social and legal issues surrounding facial analysis technologies have been widely debated in recent years. Key critics have argued that these technologies can perpetuate bias and discrimination, particularly against marginalized groups. We contribute to this field of research by reporting on the limitations of facial analysis systems with the faces of people with Down syndrome: this particularly vulnerable group has received very little attention in the literature so far. This study involved the creation of a specific dataset of face images. An experimental group with faces of people with Down syndrome, and a control group with faces of people who are not affected by the syndrome. Two commercial tools were tested on the dataset, along three tasks: gender recognition, age prediction and face labelling. The results show an overall lower accuracy of prediction in the experimental group, and other specific patterns of performance differences: i) high error rates in gender recognition in the category of males with Down syndrome; ii) adults with Down syndrome were more often incorrectly labelled as children; iii) social stereotypes are propagated in both the control and experimental groups, with labels related to aesthetics more often associated with women, and labels related to education level and skills more often associated with men. These results, although limited in scope, shed new light on the biases that alter face classification when applied to faces of people with Down syndrome. They confirm the structural limitation of the technology, which is inherently dependent on the datasets used to train the models.
Authors: Timo Fudala, Vasileios Tsouvalas, Nirvana Meratnia
Abstract: Multimodal transformers integrate diverse data types like images, audio, and text, advancing tasks such as audio-visual understanding and image-text retrieval; yet their high parameterization limits deployment on resource-constrained edge devices. Split Learning (SL), which partitions models at a designated cut-layer to offload compute-intensive operations to the server, offers a promising approach for distributed training of multimodal transformers, though its application remains underexplored. We present MPSL, a parallel SL approach for computational efficient fine-tuning of multimodal transformers in a distributed manner, while eliminating label sharing, client synchronization, and per-client sub-model management. MPSL employs lightweight client-side tokenizers and a unified modality-agnostic encoder, allowing flexible adaptation to task-specific needs. Our evaluation across 7 multimodal datasets demonstrates that MPSL matches or outperforms Federated Learning, reduces client-side computations by 250x, and achieves superior scalability in communication cost with model growth. Through extensive analysis, we highlight task suitability, trade-offs, and scenarios where MPSL excels, inspiring further exploration.
Authors: Huw Cheston, Jan Van Balen, Simon Durand
Abstract: Sampling, the practice of reusing recorded music or sounds from another source in a new work, is common in popular music genres like hip-hop and rap. Numerous services have emerged that allow users to identify connections between samples and the songs that incorporate them, with the goal of enhancing music discovery. Designing a system that can perform the same task automatically is challenging, as samples are commonly altered with audio effects like pitch- and time-stretching and may only be seconds long. Progress on this task has been minimal and is further blocked by the limited availability of training data. Here, we show that a convolutional neural network trained on an artificial dataset can identify real-world samples in commercial hip-hop music. We extract vocal, harmonic, and percussive elements from several databases of non-commercial music recordings using audio source separation, and train the model to fingerprint a subset of these elements in transformed versions of the original audio. We optimize the model using a joint classification and metric learning loss and show that it achieves 13% greater precision on real-world instances of sampling than a fingerprinting system using acoustic landmarks, and that it can recognize samples that have been both pitch shifted and time stretched. We also show that, for half of the commercial music recordings we tested, our model is capable of locating the position of a sample to within five seconds.
Authors: Yongqi An, Xu Zhao, Tao Yu, Ming Tang, Jinqiao Wang
Abstract: Outliers have been widely observed in Large Language Models (LLMs), significantly impacting model performance and posing challenges for model compression. Understanding the functionality and formation mechanisms of these outliers is critically important. Existing works, however, largely focus on reducing the impact of outliers from an algorithmic perspective, lacking an in-depth investigation into their causes and roles. In this work, we provide a detailed analysis of the formation process, underlying causes, and functions of outliers in LLMs. We define and categorize three types of outliers-activation outliers, weight outliers, and attention outliers-and analyze their distributions across different dimensions, uncovering inherent connections between their occurrences and their ultimate influence on the attention mechanism. Based on these observations, we hypothesize and explore the mechanisms by which these outliers arise and function, demonstrating through theoretical derivations and experiments that they emerge due to the self-attention mechanism's softmax operation. These outliers act as implicit context-aware scaling factors within the attention mechanism. As these outliers stem from systematic influences, we term them systematic outliers. Our study not only enhances the understanding of Transformer-based LLMs but also shows that structurally eliminating outliers can accelerate convergence and improve model compression. The code is avilable at https://github.com/an-yongqi/systematic-outliers.
Authors: Lingao Xiao, Songhua Liu, Yang He, Xinchao Wang
Abstract: Dataset distillation and dataset pruning are two prominent techniques for compressing datasets to improve computational and storage efficiency. Despite their overlapping objectives, these approaches are rarely compared directly. Even within each field, the evaluation protocols are inconsistent across various methods, which complicates fair comparisons and hinders reproducibility. Considering these limitations, we introduce in this paper a benchmark that equitably evaluates methodologies across both distillation and pruning literatures. Notably, our benchmark reveals that in the mainstream dataset distillation setting for large-scale datasets, which heavily rely on soft labels from pre-trained models, even randomly selected subsets can achieve surprisingly competitive performance. This finding suggests that an overemphasis on soft labels may be diverting attention from the intrinsic value of the image data, while also imposing additional burdens in terms of generation, storage, and application. To address these issues, we propose a new framework for dataset compression, termed Prune, Combine, and Augment (PCA), which focuses on leveraging image data exclusively, relies solely on hard labels for evaluation, and achieves state-of-the-art performance in this setup. By shifting the emphasis back to the images, our benchmark and PCA framework pave the way for more balanced and accessible techniques in dataset compression research. Our code is available at: https://github.com/ArmandXiao/Rethinking-Dataset-Compression
URLs: https://github.com/ArmandXiao/Rethinking-Dataset-Compression
Authors: Marco Rondina, Antonio Vetr\`o, Riccardo Coppola, Oumaima Regragrui, Alessandro Fabris, Gianmaria Silvello, Gian Antonio Susto, Juan Carlos De Martin
Abstract: Context. As software systems become more integrated into society's infrastructure, the responsibility of software professionals to ensure compliance with various non-functional requirements increases. These requirements include security, safety, privacy, and, increasingly, non-discrimination. Motivation. Fairness in pricing algorithms grants equitable access to basic services without discriminating on the basis of protected attributes. Method. We replicate a previous empirical study that used black box testing to audit pricing algorithms used by Italian car insurance companies, accessible through a popular online system. With respect to the previous study, we enlarged the number of tests and the number of demographic variables under analysis. Results. Our work confirms and extends previous findings, highlighting the problematic permanence of discrimination across time: demographic variables significantly impact pricing to this day, with birthplace remaining the main discriminatory factor against individuals not born in Italian cities. We also found that driver profiles can determine the number of quotes available to the user, denying equal opportunities to all. Conclusion. The study underscores the importance of testing for non-discrimination in software systems that affect people's everyday lives. Performing algorithmic audits over time makes it possible to evaluate the evolution of such algorithms. It also demonstrates the role that empirical software engineering can play in making software systems more accountable.
Authors: Filip Ekstr\"om Kelvinius, Oskar B. Andersson, Abhijith S. Parackal, Dong Qian, Rickard Armiento, Fredrik Lindsten
Abstract: Crystalline materials often exhibit a high level of symmetry. However, most generative models do not account for symmetry, but rather model each atom without any constraints on its position or element. We propose a generative model, Wyckoff Diffusion (WyckoffDiff), which generates symmetry-based descriptions of crystals. This is enabled by considering a crystal structure representation that encodes all symmetry, and we design a novel neural network architecture which enables using this representation inside a discrete generative model framework. In addition to respecting symmetry by construction, the discrete nature of our model enables fast generation. We additionally present a new metric, Fr\'echet Wrenformer Distance, which captures the symmetry aspects of the materials generated, and we benchmark WyckoffDiff against recently proposed generative models for crystal generation.
Authors: Christophe Vauthier, Quentin M\'erigot, Anna Korba
Abstract: In this paper, we investigate the properties of the Sliced Wasserstein Distance (SW) when employed as an objective functional. The SW metric has gained significant interest in the optimal transport and machine learning literature, due to its ability to capture intricate geometric properties of probability distributions while remaining computationally tractable, making it a valuable tool for various applications, including generative modeling and domain adaptation. Our study aims to provide a rigorous analysis of the critical points arising from the optimization of the SW objective. By computing explicit perturbations, we establish that stable critical points of SW cannot concentrate on segments. This stability analysis is crucial for understanding the behaviour of optimization algorithms for models trained using the SW objective. Furthermore, we investigate the properties of the SW objective, shedding light on the existence and convergence behavior of critical points. We illustrate our theoretical results through numerical experiments.
Authors: Jean Vassoyan, Nathana\"el Beau, Roman Plaud
Abstract: The ability to achieve long-term goals is a key challenge in the current development of large language models (LLMs). To address this, pre-trained LLMs can be fine-tuned with reinforcement learning (RL) to explore solutions that optimize a given goal. However, exploration with LLMs is difficult, as a balance has to be struck between discovering new solutions and staying close enough to the pre-trained model, so as not to degrade basic capabilities. This is typically controlled with a Kullback-Leibler (KL) penalty. In this paper, we investigate the exploration dynamics of a small language model on a simple arithmetic task. We show how varying degrees of pre-training influence exploration and demonstrate the importance of "critical tokens" which have a dramatic impact on the final outcome. Consequently, we introduce a simple modification to the KL penalty that favors exploration on critical tokens, increasing the efficiency of the RL fine-tuning stage.
Authors: Hidde Fokkema, Tim van Erven, Sara Magliacane
Abstract: Machine learning is a vital part of many real-world systems, but several concerns remain about the lack of interpretability, explainability and robustness of black-box AI systems. Concept-based models (CBM) address some of these challenges by learning interpretable concepts from high-dimensional data, e.g. images, which are used to predict labels. An important issue in CBMs is concept leakage, i.e., spurious information in the learned concepts, which effectively leads to learning "wrong" concepts. Current mitigating strategies are heuristic, have strong assumptions, e.g., they assume that the concepts are statistically independent of each other, or require substantial human interaction in terms of both interventions and labels provided by annotators. In this paper, we describe a framework that provides theoretical guarantees on the correctness of the learned concepts and on the number of required labels, without requiring any interventions. Our framework leverages causal representation learning (CRL) to learn high-level causal variables from low-level data, and learns to align these variables with interpretable concepts. We propose a linear and a non-parametric estimator for this mapping, providing a finite-sample high probability result in the linear case and an asymptotic consistency result for the non-parametric estimator. We implement our framework with state-of-the-art CRL methods, and show its efficacy in learning the correct concepts in synthetic and image benchmarks.
Authors: Oskar Nordenfors, Axel Flinth
Abstract: In many machine learning tasks, known symmetries can be used as an inductive bias to improve model performance. In this paper, we consider learning group equivariance through training with data augmentation. We summarize results from a previous paper of our own, and extend the results to show that equivariance of the trained model can be achieved through training on augmented data in tandem with regularization.
Authors: Gleb Novikov
Abstract: We study the problem of computationally efficient robust estimation of scatter matrices of elliptical distributions under the strong contamination model. We design polynomial time algorithms that achieve dimension-independent error in Frobenius norm. Our first result is a sequence of efficient algorithms that approaches nearly optimal error. Specifically, under a mild assumption on the eigenvalues of the scatter matrix $\Sigma$, for every $t \in \mathbb{N}$, we design an estimator that, given $n = d^{O(t)}$ samples, in time $n^{O(t)}$ finds $\hat{\Sigma}$ such that $ \Vert{\Sigma^{-1/2}\, ({\hat{\Sigma} - \Sigma})\, \Sigma^{-1/2}}\Vert_{\text{F}} \le O(t \cdot \varepsilon^{1-\frac{1}{t}})$, where $\varepsilon$ is the fraction of corruption. We do not require any assumptions on the moments of the distribution, while all previously known computationally efficient algorithms for robust covariance/scatter estimation with dimension-independent error rely on strong assumptions on the moments, such as sub-Gaussianity or (certifiable) hypercontractivity. Furthermore, under a stronger assumption on the eigenvalues of $\Sigma$ (that, in particular, is satisfied by all matrices with constant condition number), we provide a fast (sub-quadratic in the input size) algorithm that, given nearly optimal number of samples $n = \tilde{O}(d^2/\varepsilon)$, in time $\tilde{O}({nd^2 poly(1/\varepsilon)})$ finds $\hat{\Sigma}$ such that $\Vert\hat{\Sigma} - \Sigma\Vert_{\text{F}} \le O(\Vert{\Sigma}\Vert \cdot \sqrt{\varepsilon})$. Our approach is based on robust covariance estimation of the spatial sign (the projection onto the sphere of radius $\sqrt{d}$) of elliptical distributions.
Authors: Eric Aubinais, Philippe Formont, Pablo Piantanida, Elisabeth Gassiat
Abstract: Quantizing machine learning models has demonstrated its effectiveness in lowering memory and inference costs while maintaining performance levels comparable to the original models. In this work, we investigate the impact of quantization procedures on the privacy of data-driven models, specifically focusing on their vulnerability to membership inference attacks. We derive an asymptotic theoretical analysis of Membership Inference Security (MIS), characterizing the privacy implications of quantized algorithm weights against the most powerful (and possibly unknown) attacks. Building on these theoretical insights, we propose a novel methodology to empirically assess and rank the privacy levels of various quantization procedures. Using synthetic datasets, we demonstrate the effectiveness of our approach in assessing the MIS of different quantizers. Furthermore, we explore the trade-off between privacy and performance using real-world data and models in the context of molecular modeling.
Authors: M\'elissa Tamine, Benjamin Heymann, Patrick Loiseau, Maxime Vono
Abstract: Semivalue-based data valuation in machine learning (ML) quantifies the contribution of individual data points to a downstream ML task by leveraging principles from cooperative game theory and the notion of utility. While this framework has been used in practice for assessing data quality, our experiments reveal inconsistent valuation outcomes across different utilities, albeit all related to ML performance. Beyond raising concerns about the reliability of data valuation, this inconsistency is challenging to interpret, as it stems from the complex interaction of the utility with data points and semivalue weights, which has barely been studied in prior work. In this paper, we take a first step toward clarifying the utility impact on semivalue-based data valuation. Specifically, we provide geometric interpretations of this impact for a broad family of classification utilities, which includes the accuracy and the arithmetic mean. We introduce the notion of spatial signatures: given a semivalue, data points can be embedded into a two-dimensional space, and utility functions map to the dual of this space. This geometric perspective separates the influence of the dataset and semivalue from that of the utility, providing a theoretical explanation for the experimentally observed sensitivity of valuation outcomes to the utility choice.
Authors: Anirudha Majumdar, Mohit Sharma, Dmitry Kalashnikov, Sumeet Singh, Pierre Sermanet, Vikas Sindhwani
Abstract: Visuomotor policies trained via imitation learning are capable of performing challenging manipulation tasks, but are often extremely brittle to lighting, visual distractors, and object locations. These vulnerabilities can depend unpredictably on the specifics of training, and are challenging to expose without time-consuming and expensive hardware evaluations. We propose the problem of predictive red teaming: discovering vulnerabilities of a policy with respect to environmental factors, and predicting the corresponding performance degradation without hardware evaluations in off-nominal scenarios. In order to achieve this, we develop RoboART: an automated red teaming (ART) pipeline that (1) modifies nominal observations using generative image editing to vary different environmental factors, and (2) predicts performance under each variation using a policy-specific anomaly detector executed on edited observations. Experiments across 500+ hardware trials in twelve off-nominal conditions for visuomotor diffusion policies demonstrate that RoboART predicts performance degradation with high accuracy (less than 0.19 average difference between predicted and real success rates). We also demonstrate how predictive red teaming enables targeted data collection: fine-tuning with data collected under conditions predicted to be adverse boosts baseline performance by 2-7x.
Authors: Linxiao Gong, Hao Yang, Gaoyun Fang, Bobo Ju, Juncen Guo, Xiaoguang Zhu, Yan Wang, Xiping Hu, Peng Sun, Azzedine Boukerche
Abstract: The explosive growth of video data has driven the development of distributed video analytics in cloud-edge-terminal collaborative (CETC) systems, enabling efficient video processing, real-time inference, and privacy-preserving analysis. Among multiple advantages, CETC systems can distribute video processing tasks and enable adaptive analytics across cloud, edge, and terminal devices, leading to breakthroughs in video surveillance, autonomous driving, and smart cities. In this survey, we first analyze fundamental architectural components, including hierarchical, distributed, and hybrid frameworks, alongside edge computing platforms and resource management mechanisms. Building upon these foundations, edge-centric approaches emphasize on-device processing, edge-assisted offloading, and edge intelligence, while cloud-centric methods leverage powerful computational capabilities for complex video understanding and model training. Our investigation also covers hybrid video analytics incorporating adaptive task offloading and resource-aware scheduling techniques that optimize performance across the entire system. Beyond conventional approaches, recent advances in large language models and multimodal integration reveal both opportunities and challenges in platform scalability, data protection, and system reliability. Future directions also encompass explainable systems, efficient processing mechanisms, and advanced video analytics, offering valuable insights for researchers and practitioners in this dynamic field.
Authors: Armel Soubeiga, Violaine Antoine
Abstract: A recent developing trend in clustering is the advancement of algorithms that not only identify clusters within data, but also express and capture the uncertainty of cluster membership. Evidential clustering addresses this by using the Dempster-Shafer theory of belief functions, a framework designed to manage and represent uncertainty. This approach results in a credal partition, a structured set of mass functions that quantify the uncertain assignment of each object to potential groups. The Python framework evclust, presented in this paper, offers a suite of efficient evidence clustering algorithms as well as tools for visualizing, evaluating and analyzing credal partitions.
Authors: Yuchen Zhuang, Jingfeng Yang, Haoming Jiang, Xin Liu, Kewei Cheng, Sanket Lokegaonkar, Yifan Gao, Qing Ping, Tianyi Liu, Binxuan Huang, Zheng Li, Zhengyang Wang, Pei Chen, Ruijie Wang, Rongzhi Zhang, Nasser Zalmout, Priyanka Nigam, Bing Yin, Chao Zhang
Abstract: Due to the scarcity of agent-oriented pre-training data, LLM-based autonomous agents typically rely on complex prompting or extensive fine-tuning, which often fails to introduce new capabilities while preserving strong generalizability. We introduce Hephaestus-Forge, the first large-scale pre-training corpus designed to enhance the fundamental capabilities of LLM agents in API function calling, intrinsic reasoning and planning, and adapting to environmental feedback. Hephaestus-Forge comprises 103B agent-specific data encompassing 76,537 APIs, including both tool documentation to introduce knowledge of API functions and function calling trajectories to strengthen intrinsic reasoning. To explore effective training protocols, we investigate scaling laws to identify the optimal recipe in data mixing ratios. By continual pre-training on Hephaestus-Forge, Hephaestus outperforms small- to medium-scale open-source LLMs and rivals commercial LLMs on three agent benchmarks, demonstrating the effectiveness of our pre-training corpus in enhancing fundamental agentic capabilities and generalization of LLMs to new tasks or environments.
Authors: Mohammed Abdul Hafeez Khan, Samuel Morries Boddepalli, Siddhartha Bhattacharyya, Debasis Mitra
Abstract: Accurate classification and anatomical localization are essential for effective medical diagnostics and research, which may be efficiently performed using deep learning techniques. However, availability of limited labeled data poses a significant challenge. To address this, we adapted Prototypical Networks and the Propagation-Reconstruction Network (PRNet) for few-shot classification and localization, respectively, in Single Photon Emission Computed Tomography (SPECT) images. For the proof of concept we used a 2D-sliced image cropped around heart. The Prototypical Network, with a pre-trained ResNet-18 backbone, classified ventricles, myocardium, and liver tissues with 96.67% training and 93.33% validation accuracy. PRNet, adapted for 2D imaging with an encoder-decoder architecture and skip connections, achieved a training loss of 1.395, accurately reconstructing patches and capturing spatial relationships. These results highlight the potential of Prototypical Networks for tissue classification with limited labeled data and PRNet for anatomical landmark localization, paving the way for improved performance in deep learning frameworks.
Authors: Ioannis Levi, Konstantinos Kyritsis, Vasileios Papapanagiotou, Georgios Tsakiridis, Anastasios Delopoulos
Abstract: Accurate monitoring of eating behavior is crucial for managing obesity and eating disorders such as bulimia nervosa. At the same time, existing methods rely on multiple and/or specialized sensors, greatly harming adherence and ultimately, the quality and continuity of data. This paper introduces a novel approach for estimating the weight of a bite, from a commercial smartwatch. Our publicly-available dataset contains smartwatch inertial data from ten participants, with manually annotated start and end times of each bite along with their corresponding weights from a smart scale, under semi-controlled conditions. The proposed method combines extracted behavioral features such as the time required to load the utensil with food, with statistical features of inertial signals, that serve as input to a Support Vector Regression model to estimate bite weights. Under a leave-one-subject-out cross-validation scheme, our approach achieves a mean absolute error (MAE) of 3.99 grams per bite. To contextualize this performance, we introduce the improvement metric, that measures the relative MAE difference compared to a baseline model. Our method demonstrates a 17.41% improvement, while the adapted state-of-the art method shows a -28.89% performance against that same baseline. The results presented in this work establish the feasibility of extracting meaningful bite weight estimates from commercial smartwatch inertial sensors alone, laying the groundwork for future accessible, non-invasive dietary monitoring systems.
Authors: Camille Little, Lili Zheng, Genevera Allen
Abstract: Feature importance measures are widely studied and are essential for understanding model behavior, guiding feature selection, and enhancing interpretability. However, many machine learning fitted models involve complex, higher-order interactions between features. Existing feature importance metrics fail to capture these higher-order effects while existing interaction metrics often suffer from limited applicability or excessive computation; no methods exist to conduct statistical inference for feature interactions. To bridge this gap, we first propose a new model-agnostic metric, interaction Leave-One-Covariate-Out iLOCO, for measuring the importance of higher-order feature interactions. Next, we leverage recent advances in LOCO inference to develop distribution-free and assumption-light confidence intervals for our iLOCO metric. To address computational challenges, we also introduce an ensemble learning method for calculating the iLOCO metric and confidence intervals that we show is both computationally and statistically efficient. We validate our iLOCO metric and our confidence intervals on both synthetic and real data sets, showing that our approach outperforms existing methods and provides the first inferential approach to detecting feature interactions.
Authors: Fabian Gr\"oger, Pascal Baumann, Ludovic Amruthalingam, Laurent Simon, Ruksana Giurda, Simone Lionetti
Abstract: Effectively steering hearable devices requires understanding the acoustic environment around the user. In the computational analysis of sound scenes, foundation models have emerged as the state of the art to produce high-performance, robust, multi-purpose audio representations. We introduce and release Deep Evaluation of Audio Representations (DEAR), the first dataset and benchmark to evaluate the efficacy of foundation models in capturing essential acoustic properties for hearables. The dataset includes 1,158 audio tracks, each 30 seconds long, created by spatially mixing proprietary monologues with commercial, high-quality recordings of everyday acoustic scenes. Our benchmark encompasses eight tasks that assess the general context, speech sources, and technical acoustic properties of the audio scenes. Through our evaluation of four general-purpose audio representation models, we demonstrate that the BEATs model significantly surpasses its counterparts. This superiority underscores the advantage of models trained on diverse audio collections, confirming their applicability to a wide array of auditory tasks, including encoding the environment properties necessary for hearable steering. The DEAR dataset and associated code are available at https://dear-dataset.github.io.
Authors: Cyril Shih-Huan Hsu, Chrysa Papagianni, Paola Grosso
Abstract: The emergence of the fifth generation (5G) technology has transformed mobile networks into multi-service environments, necessitating efficient network slicing to meet diverse Service Level Agreements (SLAs). SLA decomposition across multiple network domains, each potentially managed by different service providers, poses a significant challenge due to limited visibility into real-time underlying domain conditions. This paper introduces Risk-Aware Iterated Local Search (RAILS), a novel risk model-driven meta-heuristic framework designed to jointly address SLA decomposition and service provider selection in multi-domain networks. By integrating online risk modeling with iterated local search principles, RAILS effectively navigates the complex optimization landscape, utilizing historical feedback from domain controllers. We formulate the joint problem as a Mixed-Integer Nonlinear Programming (MINLP) problem and prove its NP-hardness. Extensive simulations demonstrate that RAILS achieves near-optimal performance, offering an efficient, real-time solution for adaptive SLA management in modern multi-domain networks.
Authors: Ivan Lau, Jonathan Scarlett
Abstract: In this paper, we study a variant of best-arm identification involving elements of risk sensitivity and communication constraints. Specifically, the goal of the learner is to identify the arm with the highest quantile reward, while the communication from an agent (who observes rewards) and the learner (who chooses actions) is restricted to only one bit of feedback per arm pull. We propose an algorithm that utilizes noisy binary search as a subroutine, allowing the learner to estimate quantile rewards through 1-bit feedback. We derive an instance-dependent upper bound on the sample complexity of our algorithm and provide an algorithm-independent lower bound for specific instances, with the two matching to within logarithmic factors under mild conditions, or even to within constant factors in certain low error probability scaling regimes. The lower bound is applicable even in the absence of communication constraints, and thus we conclude that restricting to 1-bit feedback has a minimal impact on the scaling of the sample complexity.
Authors: Bessie Dominguez-Dager, Felix Escalona, Francisco Gomez-Donoso, Miguel Cazorla
Abstract: Person re-identification (Re-ID) is a key challenge in computer vision, requiring the matching of individuals across different cameras, locations, and time periods. While most research focuses on short-term scenarios with minimal appearance changes, real-world applications demand robust Re-ID systems capable of handling long-term scenarios, where persons' appearances can change significantly due to variations in clothing and physical characteristics. In this paper, we present CHIRLA, Comprehensive High-resolution Identification and Re-identification for Large-scale Analysis, a novel dataset specifically designed for long-term person Re-ID. CHIRLA consists of recordings from strategically placed cameras over a seven-month period, capturing significant variations in both temporal and appearance attributes, including controlled changes in participants' clothing and physical features. The dataset includes 22 individuals, four connected indoor environments, and seven cameras. We collected more than five hours of video that we semi-automatically labeled to generate around one million bounding boxes with identity annotations. By introducing this comprehensive benchmark, we aim to facilitate the development and evaluation of Re-ID algorithms that can reliably perform in challenging, long-term real-world scenarios.
Authors: Shashank Sule, Wojciech Czaja
Abstract: We present Neumann eigenmaps (NeuMaps), a novel approach for enhancing the standard diffusion map embedding using landmarks, i.e distinguished samples within the dataset. By interpreting these landmarks as a subgraph of the larger data graph, NeuMaps are obtained via the eigendecomposition of a renormalized Neumann Laplacian. We show that NeuMaps offer two key advantages: (1) they provide a computationally efficient embedding that accurately recovers the diffusion distance associated with the reflecting random walk on the subgraph, and (2) they naturally incorporate the Nystr\"om extension within the diffusion map framework through the discrete Neumann boundary condition. Through examples in digit classification and molecular dynamics, we demonstrate that NeuMaps not only improve upon existing landmark-based embedding methods but also enhance the stability of diffusion map embeddings to the removal of highly significant points.
Authors: Marina Sheshukova, Sergey Samsonov, Denis Belomestny, Eric Moulines, Qi-Man Shao, Zhuo-Song Zhang, Alexey Naumov
Abstract: In this paper, we establish non-asymptotic convergence rates in the central limit theorem for Polyak-Ruppert-averaged iterates of stochastic gradient descent (SGD). Our analysis builds on the result of the Gaussian approximation for nonlinear statistics of independent random variables of Shao and Zhang (2022). Using this result, we prove the non-asymptotic validity of the multiplier bootstrap for constructing the confidence sets for the optimal solution of an optimization problem. In particular, our approach avoids the need to approximate the limiting covariance of Polyak-Ruppert SGD iterates, which allows us to derive approximation rates in convex distance of order up to $1/\sqrt{n}$.
Authors: Valia Efthymiou, Chara Podimata, Diptangshu Sen, Juba Ziani
Abstract: We study strategic classification in binary decision-making settings where agents can modify their features in order to improve their classification outcomes. Importantly, our work considers the causal structure across different features, acknowledging that effort in a given feature may affect other features. The main goal of our work is to understand \emph{when and how much agent effort is invested towards desirable features}, and how this is influenced by the deployed classifier, the causal structure of the agent's features, their ability to modify them, and the information available to the agent about the classifier and the feature causal graph. In the complete information case, when agents know the classifier and the causal structure of the problem, we derive conditions ensuring that rational agents focus on features favored by the principal. We show that designing classifiers to induce desirable behavior is generally non-convex, though tractable in special cases. We also extend our analysis to settings where agents have incomplete information about the classifier or the causal graph. While optimal effort selection is again a non-convex problem under general uncertainty, we highlight special cases of partial uncertainty where this selection problem becomes tractable. Our results indicate that uncertainty drives agents to favor features with higher expected importance and lower variance, potentially misaligning with principal preferences. Finally, numerical experiments based on a cardiovascular disease risk study illustrate how to incentivize desirable modifications under uncertainty.
Authors: Elizaveta Semenova
Abstract: Surrogate models are widely used in natural sciences, engineering, and machine learning to approximate complex systems and reduce computational costs. However, the current landscape lacks standardisation across key stages of the pipeline, including data collection, sampling design, model class selection, evaluation metrics, and downstream task performance analysis. This fragmentation limits reproducibility, reliability, and cross-domain applicability. The issue has only been exacerbated by the AI revolution and a new suite of surrogate model classes that it offers. In this position paper, we argue for the urgent need for a unified framework to guide the development and evaluation of surrogate models. We outline essential steps for constructing a comprehensive pipeline and discuss alternative perspectives, such as the benefits of domain-specific frameworks. By advocating for a standardised approach, this paper seeks to improve the reliability of surrogate modelling, foster cross-disciplinary knowledge transfer, and, as a result, accelerate scientific progress.
Authors: Manuel M. M\"uller, Yuetian Luo, Rina Foygel Barber
Abstract: In statistics and machine learning, when we train a fitted model on available data, we typically want to ensure that we are searching within a model class that contains at least one accurate model -- that is, we would like to ensure an upper bound on the model class risk (the lowest possible risk that can be attained by any model in the class). However, it is also of interest to establish lower bounds on the model class risk, for instance so that we can determine whether our fitted model is at least approximately optimal within the class, or, so that we can decide whether the model class is unsuitable for the particular task at hand. Particularly in the setting of interpolation learning where machine learning models are trained to reach zero error on the training data, we might ask if, at the very least, a positive lower bound on the model class risk is possible -- or are we unable to detect that "all models are wrong"? In this work, we answer these questions in a distribution-free setting by establishing a model-agnostic, fundamental hardness result for the problem of constructing a lower bound on the best test error achievable over a model class, and examine its implications on specific model classes such as tree-based methods and linear regression.
Authors: Emanuele Coradin (INFN, Sezione di Padova, Italy), Fabio Cufino (Universit\`a di Bologna, Dipartimento di Fisica, Italy), Muhammad Awais (INFN, Sezione di Padova, Italy), Tommaso Dorigo (INFN, Sezione di Padova, Italy), Enrico Lupi (INFN, Sezione di Padova, Italy), Eleonora Porcu (Universit\`a di Bologna, Dipartimento di Fisica, Italy), Jinu Raj (Central University of Tamil Nadu, India), Fredrik Sandin (Lule{\aa} University of Technology, Sweden), Mia Tosi (INFN, Sezione di Padova, Italy)
Abstract: We study the application of a neural network architecture for identifying charged particle trajectories via unsupervised learning of delays and synaptic weights using a spike-time-dependent plasticity rule. In the considered model, the neurons receive time-encoded information on the position of particle hits in a tracking detector for a particle collider, modeled according to the geometry of the Compact Muon Solenoid Phase II detector. We show how a spiking neural network is capable of successfully identifying in a completely unsupervised way the signal left by charged particles in the presence of conspicuous noise from accidental or combinatorial hits. These results open the way to applications of neuromorphic computing to particle tracking, motivating further studies into its potential for real-time, low-power particle tracking in future high-energy physics experiments.
Authors: Guanghao Ye, Khiem Duc Pham, Xinzhi Zhang, Sivakanth Gopi, Baolin Peng, Beibin Li, Janardhan Kulkarni, Huseyin A. Inan
Abstract: Recent AI advancements, such as OpenAI's new models, are transforming LLMs into LRMs (Large Reasoning Models) that perform reasoning during inference, taking extra time and compute for higher-quality outputs. We aim to uncover the algorithmic framework for training LRMs. Methods like self-consistency, PRM, and AlphaZero suggest reasoning as guided search. We ask: what is the simplest, most scalable way to enable search in LLMs? We propose a post-training framework called Reinforcement Learning via Self-Play (RLSP). RLSP involves three steps: (1) supervised fine-tuning with human or synthetic demonstrations of the reasoning process, (2) using an exploration reward signal to encourage diverse and efficient reasoning behaviors, and (3) RL training with an outcome verifier to ensure correctness while preventing reward hacking. Our key innovation is to decouple exploration and correctness signals during PPO training, carefully balancing them to improve performance and efficiency. Empirical studies in the math domain show that RLSP improves reasoning. On the Llama-3.1-8B-Instruct model, RLSP can boost performance by 23% in MATH-500 test set; On AIME 2024 math problems, Qwen2.5-32B-Instruct improved by 10% due to RLSP. However, a more important finding of this work is that the models trained using RLSP, even with the simplest exploration reward that encourages the model to take more intermediate steps, showed several emergent behaviors such as backtracking, exploration of ideas, and verification. These findings demonstrate that RLSP framework might be enough to enable emergence of complex reasoning abilities in LLMs when scaled. Lastly, we propose a theory as to why RLSP search strategy is more suitable for LLMs inspired by a remarkable result that says CoT provably increases computational power of LLMs, which grows as the number of steps in CoT \cite{li2024chain,merrill2023expresssive}.
Authors: Yuxuan Han, Han Zhong, Miao Lu, Jose Blanchet, Zhengyuan Zhou
Abstract: We study the fundamental problem of offline assortment optimization under the Multinomial Logit (MNL) model, where sellers must determine the optimal subset of the products to offer based solely on historical customer choice data. While most existing approaches to learning-based assortment optimization focus on the online learning of the optimal assortment through repeated interactions with customers, such exploration can be costly or even impractical in many real-world settings. In this paper, we consider the offline learning paradigm and investigate the minimal data requirements for efficient offline assortment optimization. To this end, we introduce Pessimistic Rank-Breaking (PRB), an algorithm that combines rank-breaking with pessimistic estimation. We prove that PRB is nearly minimax optimal by establishing the tight suboptimality upper bound and a nearly matching lower bound. This further shows that "optimal item coverage" - where each item in the optimal assortment appears sufficiently often in the historical data - is both sufficient and necessary for efficient offline learning. This significantly relaxes the previous requirement of observing the complete optimal assortment in the data. Our results provide fundamental insights into the data requirements for offline assortment optimization under the MNL model.
Authors: Chengqi Lyu, Songyang Gao, Yuzhe Gu, Wenwei Zhang, Jianfei Gao, Kuikun Liu, Ziyi Wang, Shuaibin Li, Qian Zhao, Haian Huang, Weihan Cao, Jiangning Liu, Hongwei Liu, Junnan Liu, Songyang Zhang, Dahua Lin, Kai Chen
Abstract: Reasoning abilities, especially those for solving complex math problems, are crucial components of general intelligence. Recent advances by proprietary companies, such as o-series models of OpenAI, have made remarkable progress on reasoning tasks. However, the complete technical details remain unrevealed, and the techniques that are believed certainly to be adopted are only reinforcement learning (RL) and the long chain of thoughts. This paper proposes a new RL framework, termed OREAL, to pursue the performance limit that can be achieved through \textbf{O}utcome \textbf{RE}w\textbf{A}rd-based reinforcement \textbf{L}earning for mathematical reasoning tasks, where only binary outcome rewards are easily accessible. We theoretically prove that behavior cloning on positive trajectories from best-of-N (BoN) sampling is sufficient to learn the KL-regularized optimal policy in binary feedback environments. This formulation further implies that the rewards of negative samples should be reshaped to ensure the gradient consistency between positive and negative samples. To alleviate the long-existing difficulties brought by sparse rewards in RL, which are even exacerbated by the partial correctness of the long chain of thought for reasoning tasks, we further apply a token-level reward model to sample important tokens in reasoning trajectories for learning. With OREAL, for the first time, a 7B model can obtain 94.0 pass@1 accuracy on MATH-500 through RL, being on par with 32B models. OREAL-32B also surpasses previous 32B models trained by distillation with 95.0 pass@1 accuracy on MATH-500. Our investigation also indicates the importance of initial policy models and training queries for RL. Code, models, and data will be released to benefit future research\footnote{https://github.com/InternLM/OREAL}.
Authors: Zhe Yu, Xiaoyin Xi
Abstract: With AI systems widely applied to assist human in decision-making processes such as talent hiring, school admission, and loan approval; there is an increasing need to ensure that the decisions made are fair. One major challenge for analyzing fairness in decisions is that the standards are highly subjective and contextual -- there is no consensus for what absolute fairness means for every scenario. Not to say that different fairness standards often conflict with each other. To bypass this issue, this work aims to test relative fairness in decisions. That is, instead of defining what are ``absolutely'' fair decisions, we propose to test the relative fairness of one decision set against another with differential parity -- the difference between two sets of decisions should be independent from a certain sensitive attribute. This proposed differential parity fairness notion has the following benefits: (1) it avoids the ambiguous and contradictory definition of ``absolutely'' fair decisions; (2) it reveals the relative preference and bias between two decision sets; (3) differential parity can serve as a new group fairness notion when a reference set of decisions (ground truths) is provided. One limitation for differential parity is that, it requires the two sets of decisions under comparison to be made on the same data subjects. To overcome this limitation, we propose to utilize a machine learning model to bridge the gap between the two decisions sets made on difference data and estimate the differential parity.
Authors: Tao Li, Zhehao Huang, Yingwen Wu, Zhengbao He, Qinghua Tao, Xiaolin Huang, Chih-Jen Lin
Abstract: Weight averaging is a widely used technique for accelerating training and improving the generalization of deep neural networks (DNNs). While existing approaches like stochastic weight averaging (SWA) rely on pre-set weighting schemes, they can be suboptimal when handling diverse weights. We introduce Trainable Weight Averaging (TWA), a novel optimization method that operates within a reduced subspace spanned by candidate weights and learns optimal weighting coefficients through optimization. TWA offers greater flexibility and can be applied to different training scenarios. For large-scale applications, we develop a distributed training framework that combines parallel computation with low-bit compression for the projection matrix, effectively managing memory and computational demands. TWA can be implemented using either training data (TWA-t) or validation data (TWA-v), with the latter providing more effective averaging. Extensive experiments showcase TWA's advantages: (i) it consistently outperforms SWA in generalization performance and flexibility, (ii) when applied during early training, it reduces training time by over 40\% on CIFAR datasets and 30\% on ImageNet while maintaining comparable performance, and (iii) during fine-tuning, it significantly enhances generalization by weighted averaging of model checkpoints. In summary, we present an efficient and effective framework for trainable weight averaging. The code is available at https://github.com/nblt/TWA.
Authors: Xuefeng Liu, Fangfang Xia, Rick L. Stevens, Yuxin Chen
Abstract: While training models and labeling data are resource-intensive, a wealth of pre-trained models and unlabeled data exists. To effectively utilize these resources, we present an approach to actively select pre-trained models while minimizing labeling costs. We frame this as an online contextual active model selection problem: At each round, the learner receives an unlabeled data point as a context. The objective is to adaptively select the best model to make a prediction while limiting label requests. To tackle this problem, we propose CAMS, a contextual active model selection algorithm that relies on two novel components: (1) a contextual model selection mechanism, which leverages context information to make informed decisions about which model is likely to perform best for a given context, and (2) an active query component, which strategically chooses when to request labels for data points, minimizing the overall labeling cost. We provide rigorous theoretical analysis for the regret and query complexity under both adversarial and stochastic settings. Furthermore, we demonstrate the effectiveness of our algorithm on a diverse collection of benchmark classification tasks. Notably, CAMS requires substantially less labeling effort (less than 10%) compared to existing methods on CIFAR10 and DRIFT benchmarks, while achieving similar or better accuracy. Our code is publicly available at: https://github.com/xuefeng-cs/Contextual-Active-Model-Selection.
URLs: https://github.com/xuefeng-cs/Contextual-Active-Model-Selection.
Authors: Florian E. Dorner, Nikola Konstantinov, Georgi Pashaliev, Martin Vechev
Abstract: Collaborative learning techniques have the potential to enable training machine learning models that are superior to models trained on a single entity's data. However, in many cases, potential participants in such collaborative schemes are competitors on a downstream task, such as firms that each aim to attract customers by providing the best recommendations. This can incentivize dishonest updates that damage other participants' models, potentially undermining the benefits of collaboration. In this work, we formulate a game that models such interactions and study two learning tasks within this framework: single-round mean estimation and multi-round SGD on strongly-convex objectives. For a natural class of player actions, we show that rational clients are incentivized to strongly manipulate their updates, preventing learning. We then propose mechanisms that incentivize honest communication and ensure learning quality comparable to full cooperation. Lastly, we empirically demonstrate the effectiveness of our incentive scheme on a standard non-convex federated learning benchmark. Our work shows that explicitly modeling the incentives and actions of dishonest clients, rather than assuming them malicious, can enable strong robustness guarantees for collaborative learning.
Authors: Andr\'es R. Masegosa, Luis A. Ortega
Abstract: This paper introduces a distribution-dependent PAC-Chernoff bound that exhibits perfect tightness for interpolators, even within over-parameterized model classes. This bound, which relies on basic principles of Large Deviation Theory, defines a natural measure of the smoothness of a model, characterized by simple real-valued functions. Building upon this bound and the new concept of smoothness, we present an unified theoretical framework revealing why certain interpolators show an exceptional generalization, while others falter. We theoretically show how a wide spectrum of modern learning methodologies, encompassing techniques such as $\ell_2$-norm, distance-from-initialization and input-gradient regularization, in combination with data augmentation, invariant architectures, and over-parameterization, collectively guide the optimizer toward smoother interpolators, which, according to our theoretical framework, are the ones exhibiting superior generalization performance. This study shows that distribution-dependent bounds serve as a powerful tool to understand the complex dynamics behind the generalization capabilities of over-parameterized interpolators.
Authors: Afsana Khan, Marijn ten Thij, Frank Thuijsman, Anna Wilbik
Abstract: Vertical federated learning (VFL) is a promising approach for collaboratively training machine learning models using private data partitioned vertically across different parties. Ideally in a VFL setting, the active party (party possessing features of samples with labels) benefits by improving its machine learning model through collaboration with some passive parties (parties possessing additional features of the same samples without labels) in a privacy preserving manner. However, motivating passive parties to participate in VFL can be challenging. In this paper, we focus on the problem of allocating incentives to the passive parties by the active party based on their contributions to the VFL process. We address this by formulating the incentive allocation problem as a bankruptcy game, a concept from cooperative game theory. Using the Talmudic division rule, which leads to the Nucleolus as its solution, we ensure a fair distribution of incentives. We evaluate our proposed method on synthetic and real-world datasets and show that it ensures fairness and stability in incentive allocation among passive parties who contribute their data to the federated model. Additionally, we compare our method to the existing solution of calculating Shapley values and show that our approach provides a more efficient solution with fewer computations.
Authors: Hao Lin, Ke Wu, Jie Li, Jun Li, Wu-Jun Li
Abstract: Distributed learning is commonly used for training deep learning models, especially large models. In distributed learning, manual parallelism (MP) methods demand considerable human effort and have limited flexibility. Hence, automatic parallelism (AP) methods have recently been proposed for automating the parallel strategy optimization process. Existing AP methods suffer from sub-optimal solutions because they do not jointly optimize the two categories of parallel strategies (i.e., inter-layer parallelism and intra-layer parallelism). In this paper, we propose a novel AP method called UniAP, which unifies inter- and intra-layer automatic parallelism by mixed integer quadratic programming. To the best of our knowledge, UniAP is the first parallel method that can jointly optimize the two categories of parallel strategies to find an optimal solution. Experimental results show that UniAP outperforms state-of-the-art methods by up to 3.80$\times$ in throughput and reduces strategy optimization time by up to 107$\times$ across five Transformer-based models.
Authors: Miguel Abreu, Luis Paulo Reis, Nuno Lau
Abstract: Symmetry, a fundamental concept to understand our environment, often oversimplifies reality from a mathematical perspective. Humans are a prime example, deviating from perfect symmetry in terms of appearance and cognitive biases (e.g. having a dominant hand). Nevertheless, our brain can easily overcome these imperfections and efficiently adapt to symmetrical tasks. The driving motivation behind this work lies in capturing this ability through reinforcement learning. To this end, we introduce Adaptive Symmetry Learning (ASL), a model-minimization actor-critic extension that addresses incomplete or inexact symmetry descriptions by adapting itself during the learning process. ASL consists of a symmetry fitting component and a modular loss function that enforces a common symmetric relation across all states while adapting to the learned policy. The performance of ASL is compared to existing symmetry-enhanced methods in a case study involving a four-legged ant model for multidirectional locomotion tasks. The results show that ASL can recover from large perturbations and generalize knowledge to hidden symmetric states. It achieves comparable or better performance than alternative methods in most scenarios, making it a valuable approach for leveraging model symmetry while compensating for inherent perturbations.
Authors: Yupeng Wu, Wenjie Huang, Chin Pang Ho
Abstract: One of the main challenges in reinforcement learning (RL) is that the agent has to make decisions that would influence the future performance without having complete knowledge of the environment. Dynamically adjusting the level of epistemic risk during the learning process can help to achieve reliable policies in safety-critical settings with better efficiency. In this work, we propose a new framework, Distributional RL with Online Risk Adaptation (DRL-ORA). This framework quantifies both epistemic and implicit aleatory uncertainties in a unified manner and dynamically adjusts the epistemic risk levels by solving a total variation minimization problem online. The selection of risk levels is performed efficiently via a grid search using a Follow-The-Leader-type algorithm, where the offline oracle corresponds to a "satisficing measure" under a specially modified loss function. We show that DRL-ORA outperforms existing methods that rely on fixed risk levels or manually designed risk level adaptation in multiple classes of tasks.
Authors: Luigi Sbail\`o, Luca Ghiringhelli
Abstract: The configuration of latent representations plays a critical role in determining the performance of deep neural network classifiers. In particular, the emergence of well-separated class embeddings in the latent space has been shown to improve both generalization and robustness. In this paper, we propose a method to induce the collapse of latent representations belonging to the same class into a single point, which enhances class separability in the latent space while enforcing Lipschitz continuity in the network. We demonstrate that this phenomenon, which we call \textit{latent point collapse}, is achieved by adding a strong $L_2$ penalty on the penultimate-layer representations and is the result of a push-pull tension developed with the cross-entropy loss function. In addition, we show the practical utility of applying this compressing loss term to the latent representations of a low-dimensional linear penultimate layer. The proposed approach is straightforward to implement and yields substantial improvements in discriminative feature embeddings, along with remarkable gains in robustness to input perturbations.
Authors: Debolina Halder Lina, Arlei Silva
Abstract: Graph Neural Networks (GNNs) have achieved state-of-the-art results in many relevant tasks where decisions might disproportionately impact specific communities. However, existing work on fair GNNs often assumes that either protected attributes are fully observed or that the missing protected attribute imputation is fair. In practice, biases in the imputation will propagate to the model outcomes, leading them to overestimate the fairness of their predictions. We address this challenge by proposing Better Fair than Sorry (BFtS), a fair missing data imputation model for protected attributes. The key design principle behind BFtS is that imputations should approximate the worst-case scenario for fairness -- i.e. when optimizing fairness is the hardest. We implement this idea using a 3-player adversarial scheme where two adversaries collaborate against a GNN-based classifier, and the classifier minimizes the maximum bias. Experiments using synthetic and real datasets show that BFtS often achieves a better fairness x accuracy trade-off than existing alternatives.
Authors: Giovanni Luca Marchetti, Gabriele Cesa, Pratik Kumar, Arash Behboodi
Abstract: Lattice reduction is a combinatorial optimization problem aimed at finding the most orthogonal basis in a given lattice. The Lenstra-Lenstra-Lov\'asz (LLL) algorithm is the best algorithm in the literature for solving this problem. In light of recent research on algorithm discovery, in this work, we would like to answer this question: is it possible to parametrize the algorithm space for lattice reduction problem with neural networks and find an algorithm without supervised data? Our strategy is to use equivariant and invariant parametrizations and train in a self-supervised way. We design a deep neural model outputting factorized unimodular matrices and train it in a self-supervised manner by penalizing non-orthogonal lattice bases. We incorporate the symmetries of lattice reduction into the model by making it invariant to isometries and scaling of the ambient space and equivariant with respect to the hyperocrahedral group permuting and flipping the lattice basis elements. We show that this approach yields an algorithm with comparable complexity and performance to the LLL algorithm on a set of benchmarks. Additionally, motivated by certain applications for wireless communication, we extend our method to a convolutional architecture which performs joint reduction of spatially-correlated lattices arranged in a grid, thereby amortizing its cost over multiple lattices.
Authors: Hikari Otsuka, Daiki Chijiwa, \'Angel L\'opez Garc\'ia-Arias, Yasuyuki Okoshi, Kazushi Kawamura, Thiem Van Chu, Daichi Fujiki, Susumu Takeuchi, Masato Motomura
Abstract: Randomly initialized dense networks contain subnetworks that achieve high accuracy without weight learning--strong lottery tickets (SLTs). Recently, Gadhikar et al. (2023) demonstrated that SLTs could also be found within a randomly pruned source network. This phenomenon can be exploited to further compress the small memory size required by SLTs. However, their method is limited to SLTs that are even sparser than the source, leading to worse accuracy due to unintentionally high sparsity. This paper proposes a method for reducing the SLT memory size without restricting the sparsity of the SLTs that can be found. A random subset of the initial weights is frozen by either permanently pruning them or locking them as a fixed part of the SLT, resulting in a smaller model size. Experimental results show that Edge-Popup (Ramanujan et al., 2020; Sreenivasan et al., 2022) finds SLTs with better accuracy-to-model size trade-off within frozen networks than within dense or randomly pruned source networks. In particular, freezing $70\%$ of a ResNet on ImageNet provides $3.3 \times$ compression compared to the SLT found within a dense counterpart, raises accuracy by up to $14.12$ points compared to the SLT found within a randomly pruned counterpart, and offers a better accuracy-model size trade-off than both.
Authors: Haeun Jeon, Hyunglip Bae, Minsu Park, Chanyeong Kim, Woo Chang Kim
Abstract: In decision-making problems under uncertainty, predicting unknown parameters is often considered independent of the optimization part. Decision-focused learning (DFL) is a task-oriented framework that integrates prediction and optimization by adapting the predictive model to give better decisions for the corresponding task. Here, an inevitable challenge arises when computing the gradients of the optimal decision with respect to the parameters. Existing research copes with this issue by smoothly reforming surrogate optimization or constructing surrogate loss functions that mimic task loss. However, they are applied to restricted optimization domains. In this paper, we propose Locally Convex Global Loss Network (LCGLN), a global surrogate loss model that can be implemented in a general DFL paradigm. LCGLN learns task loss via a partial input convex neural network which is guaranteed to be convex for chosen inputs while keeping the non-convex global structure for the other inputs. This enables LCGLN to admit general DFL through only a single surrogate loss without any sense for choosing appropriate parametric forms. We confirm the effectiveness and flexibility of LCGLN by evaluating our proposed model with three stochastic decision-making problems.
Authors: Th\'eo Vincent, Daniel Palenicek, Boris Belousov, Jan Peters, Carlo D'Eramo
Abstract: The vast majority of Reinforcement Learning methods is largely impacted by the computation effort and data requirements needed to obtain effective estimates of action-value functions, which in turn determine the quality of the overall performance and the sample-efficiency of the learning procedure. Typically, action-value functions are estimated through an iterative scheme that alternates the application of an empirical approximation of the Bellman operator and a subsequent projection step onto a considered function space. It has been observed that this scheme can be potentially generalized to carry out multiple iterations of the Bellman operator at once, benefiting the underlying learning algorithm. However, till now, it has been challenging to effectively implement this idea, especially in high-dimensional problems. In this paper, we introduce iterated $Q$-Network (i-QN), a novel principled approach that enables multiple consecutive Bellman updates by learning a tailored sequence of action-value functions where each serves as the target for the next. We show that i-QN is theoretically grounded and that it can be seamlessly used in value-based and actor-critic methods. We empirically demonstrate the advantages of i-QN in Atari $2600$ games and MuJoCo continuous control problems.
Authors: Zhanke Zhou, Yongqi Zhang, Jiangchao Yao, Quanming Yao, Bo Han
Abstract: To deduce new facts on a knowledge graph (KG), a link predictor learns from the graph structure and collects local evidence to find the answer to a given query. However, existing methods suffer from a severe scalability problem due to the utilization of the whole KG for prediction, which hinders their promise on large scale KGs and cannot be directly addressed by vanilla sampling methods. In this work, we propose the one-shot-subgraph link prediction to achieve efficient and adaptive prediction. The design principle is that, instead of directly acting on the whole KG, the prediction procedure is decoupled into two steps, i.e., (i) extracting only one subgraph according to the query and (ii) predicting on this single, query dependent subgraph. We reveal that the non-parametric and computation-efficient heuristics Personalized PageRank (PPR) can effectively identify the potential answers and supporting evidence. With efficient subgraph-based prediction, we further introduce the automated searching of the optimal configurations in both data and model spaces. Empirically, we achieve promoted efficiency and leading performances on five large-scale benchmarks. The code is publicly available at: https://github.com/tmlr-group/one-shot-subgraph.
Authors: Muhammad Aneeq uz Zaman, Alec Koppel, Mathieu Lauri\`ere, Tamer Ba\c{s}ar
Abstract: We address in this paper Reinforcement Learning (RL) among agents that are grouped into teams such that there is cooperation within each team but general-sum (non-zero sum) competition across different teams. To develop an RL method that provably achieves a Nash equilibrium, we focus on a linear-quadratic structure. Moreover, to tackle the non-stationarity induced by multi-agent interactions in the finite population setting, we consider the case where the number of agents within each team is infinite, i.e., the mean-field setting. This results in a General-Sum LQ Mean-Field Type Game (GS-MFTG). We characterize the Nash equilibrium (NE) of the GS-MFTG, under a standard invertibility condition. This MFTG NE is then shown to be $O(1/M)$-NE for the finite population game where $M$ is a lower bound on the number of agents in each team. These structural results motivate an algorithm called Multi-player Receding-horizon Natural Policy Gradient (MRNPG), where each team minimizes its cumulative cost \emph{independently} in a receding-horizon manner. Despite the non-convexity of the problem, we establish that the resulting algorithm converges to a global NE through a novel problem decomposition into sub-problems using backward recursive discrete-time Hamilton-Jacobi-Isaacs (HJI) equations, in which \emph{independent natural policy gradient} is shown to exhibit linear convergence under time-independent diagonal dominance. Numerical studies included corroborate the theoretical results.
Authors: Tyler Chang, Andrew Gillette, Romit Maulik
Abstract: Effective verification and validation techniques for modern scientific machine learning workflows are challenging to devise. Statistical methods are abundant and easily deployed, but often rely on speculative assumptions about the data and methods involved. Error bounds for classical interpolation techniques can provide mathematically rigorous estimates of accuracy, but often are difficult or impractical to determine computationally. In this work, we present a best-of-both-worlds approach to verifiable scientific machine learning by demonstrating that (1) multiple standard interpolation techniques have informative error bounds that can be computed or estimated efficiently; (2) comparative performance among distinct interpolants can aid in validation goals; (3) deploying interpolation methods on latent spaces generated by deep learning techniques enables some interpretability for black-box models. We present a detailed case study of our approach for predicting lift-drag ratios from airfoil images. Code developed for this work is available in a public Github repository.
Authors: Artem Vysogorets, Kartik Ahuja, Julia Kempe
Abstract: In the era of exceptionally data-hungry models, careful selection of the training data is essential to mitigate the extensive costs of deep learning. Data pruning offers a solution by removing redundant or uninformative samples from the dataset, which yields faster convergence and improved neural scaling laws. However, little is known about its impact on classification bias of the trained models. We conduct the first systematic study of this effect and reveal that existing data pruning algorithms can produce highly biased classifiers. We present theoretical analysis of the classification risk in a mixture of Gaussians to argue that choosing appropriate class pruning ratios, coupled with random pruning within classes has potential to improve worst-class performance. We thus propose DRoP, a distributionally robust approach to pruning and empirically demonstrate its performance on standard computer vision benchmarks. In sharp contrast to existing algorithms, our proposed method continues improving distributional robustness at a tolerable drop of average performance as we prune more from the datasets.
Authors: Ali Al-Lawati, Elsayed Eshra, Prasenjit Mitra
Abstract: Trajectory generation is an important task in movement studies; it circumvents the privacy, ethical, and technical challenges of collecting real trajectories from the target population. In particular, real trajectories in the wildlife domain are scarce as a result of ethical and environmental constraints of the collection process. In this paper, we consider the problem of generating long-horizon trajectories, akin to wildlife migration, based on a small set of real samples. We propose a hierarchical approach to learn the global movement characteristics of the real dataset and recursively refine localized regions. Our solution, WildGraph, discretizes the geographic path into a prototype network of H3 (https://www.uber.com/blog/h3/) regions and leverages a recurrent variational auto-encoder to probabilistically generate paths over the regions, based on occupancy. WildGraph successfully generates realistic months-long trajectories using a sample size as small as 60. Experiments performed on two wildlife migration datasets demonstrate that our proposed method improves the generalization of the generated trajectories in comparison to existing work while achieving superior or comparable performance in several benchmark metrics. Our code is published on the following repository: https://github.com/aliwister/wildgraph.
URLs: https://www.uber.com/blog/h3/), https://github.com/aliwister/wildgraph.
Authors: William Fleshman, Aleem Khan, Marc Marone, Benjamin Van Durme
Abstract: Large language models (LLMs) are increasingly capable of completing knowledge intensive tasks by recalling information from a static pretraining corpus. Here we are concerned with LLMs in the context of evolving data requirements. For instance: batches of new data that are introduced periodically; subsets of data with user-based access controls; or requirements on dynamic removal of documents with guarantees that associated knowledge cannot be recalled. We wish to satisfy these requirements while at the same time ensuring a model does not forget old information when new data becomes available. To address these issues, we introduce AdapterSwap, a training and inference scheme that organizes knowledge from a data collection into a set of low-rank adapters, which are dynamically composed during inference. Our experiments demonstrate AdapterSwap's ability to support efficient continual learning, while also enabling organizations to have fine-grained control over data access and deletion.
Authors: Yanwen Wang, Mahdi Khodadadzadeh, Raul Zurita-Milla
Abstract: Recent geospatial machine learning studies have shown that the results of model evaluation via cross-validation (CV) are strongly affected by the dissimilarity between the sample data and the prediction locations. In this paper, we propose a method to quantify such a dissimilarity in the interval 0 to 100% and from the perspective of the data feature space. The proposed method is based on adversarial validation, which is an approach that can check whether sample data and prediction locations can be separated with a binary classifier. The proposed method is called dissimilarity quantification by adversarial validation (DAV). To study the effectiveness and general?ity of DAV, we tested it on a series of experiments based on both synthetic and real datasets and with gradually increasing dissimilarities. Results show that DAV effectively quantified dissimilarity across the entire range of values. Next to this, we studied how dissimilarity affects CV methods' evaluations by comparing the results of random CV method (RDM-CV) and of two geospatial CV methods, namely, block and spatial+ CV (BLK-CV and SP-CV). Our results showed the evaluations follow similar patterns in all datasets and predictions: when dissimilarity is low (usually lower than 30%), RDM-CV provides the most accurate evaluation results. As dissimilarity increases, geospatial CV methods, especially SP-CV, become more and more accurate and even outperform RDM-CV. When dissimilarity is high (>=90%), no CV method provides accurate evaluations. These results show the importance of considering feature space dissimilarity when working with geospatial machine learning predictions and can help researchers and practitioners to select more suitable CV methods for evaluating their predictions.
Authors: Jiyoun Kim, Junu Kim, Kyunghoon Hur, Edward Choi
Abstract: The increasing volume of electronic health records (EHRs) presents the opportunity to improve the accuracy and robustness of models in clinical prediction tasks. Unlike traditional centralized approaches, federated learning enables training on data from multiple institutions while preserving patient privacy and complying with regulatory constraints. However, most federated learning research focuses on building a global model to serve multiple clients, overlooking the practical need for a client-specific model. In this work, we introduce EHRFL, a federated learning framework using EHRs, designed to develop a model tailored to a single client (i.e., healthcare institution). Our framework addresses two key challenges: (1) enabling federated learning across clients with heterogeneous EHR systems using text-based EHR modeling, and (2) reducing the cost of federated learning by selecting suitable participating clients using averaged patient embeddings. Our experiment results on multiple open-source EHR datasets demonstrate the effectiveness of EHRFL in addressing the two challenges, establishing it as a practical solution for building a client-specific model in federated learning.
Authors: Ziming Liu, Yixuan Wang, Sachin Vaidya, Fabian Ruehle, James Halverson, Marin Solja\v{c}i\'c, Thomas Y. Hou, Max Tegmark
Abstract: Inspired by the Kolmogorov-Arnold representation theorem, we propose Kolmogorov-Arnold Networks (KANs) as promising alternatives to Multi-Layer Perceptrons (MLPs). While MLPs have fixed activation functions on nodes ("neurons"), KANs have learnable activation functions on edges ("weights"). KANs have no linear weights at all -- every weight parameter is replaced by a univariate function parametrized as a spline. We show that this seemingly simple change makes KANs outperform MLPs in terms of accuracy and interpretability. For accuracy, much smaller KANs can achieve comparable or better accuracy than much larger MLPs in data fitting and PDE solving. Theoretically and empirically, KANs possess faster neural scaling laws than MLPs. For interpretability, KANs can be intuitively visualized and can easily interact with human users. Through two examples in mathematics and physics, KANs are shown to be useful collaborators helping scientists (re)discover mathematical and physical laws. In summary, KANs are promising alternatives for MLPs, opening opportunities for further improving today's deep learning models which rely heavily on MLPs.
Authors: Liran Nochumsohn, Omri Azencot
Abstract: Data augmentation serves as a popular regularization technique to combat overfitting challenges in neural networks. While automatic augmentation has demonstrated success in image classification tasks, its application to time-series problems, particularly in long-term forecasting, has received comparatively less attention. To address this gap, we introduce a time-series automatic augmentation approach named TSAA, which is both efficient and easy to implement. The solution involves tackling the associated bilevel optimization problem through a two-step process: initially training a non-augmented model for a limited number of epochs, followed by an iterative split procedure. During this iterative process, we alternate between identifying a robust augmentation policy through Bayesian optimization and refining the model while discarding suboptimal runs. Extensive evaluations on challenging univariate and multivariate forecasting benchmark problems demonstrate that TSAA consistently outperforms several robust baselines, suggesting its potential integration into prediction pipelines. Code is available at this repository: https://github.com/azencot-group/TSAA.
Authors: Sharmita Dey, Benjamin Paassen, Sarath Ravindran Nair, Sabri Boughorbel, Arndt F. Schilling
Abstract: Lower limb amputations and neuromuscular impairments severely restrict mobility, necessitating advancements beyond conventional prosthetics. While motorized bionic limbs show promise, their effectiveness depends on replicating the dynamic coordination of human movement across diverse environments. In this paper, we introduce a model for human behavior in the context of bionic prosthesis control. Our approach leverages human locomotion demonstrations to learn the synergistic coupling of the lower limbs, enabling the prediction of the kinematic behavior of a missing limb during tasks such as walking, climbing inclines, and stairs. We propose a multitasking, continually adaptive model that anticipates and refines movements over time. At the core of our method is a technique called multitask prospective rehearsal, that anticipates and synthesizes future movements based on the previous prediction and employs a corrective mechanism for subsequent predictions. Our evolving architecture merges lightweight, task-specific modules on a shared backbone, ensuring both specificity and scalability. We validate our model through experiments on real-world human gait datasets, including transtibial amputees, across a wide range of locomotion tasks. Results demonstrate that our approach consistently outperforms baseline models, particularly in scenarios with distributional shifts, adversarial perturbations, and noise.
Authors: Yijin Ni, Xiaoming Huo
Abstract: In many contemporary statistical and machine learning methods, one needs to optimize an objective function that depends on the discrepancy between two probability distributions. The discrepancy can be referred to as a metric for distributions. Widely adopted examples of such a metric include Energy Distance (ED), distance Covariance (dCov), Maximum Mean Discrepancy (MMD), and the Hilbert-Schmidt Independence Criterion (HSIC). We show that these metrics can be unified under a general framework of kernel-based two-sample statistics. This paper establishes a novel uniform concentration inequality for the aforementioned kernel-based statistics. Our results provide upper bounds for estimation errors in the associated optimization problems, thereby offering both finite-sample and asymptotic performance guarantees. As illustrative applications, we demonstrate how these bounds facilitate the derivation of error bounds for procedures such as distance covariance-based dimension reduction, distance covariance-based independent component analysis, MMD-based fairness-constrained inference, MMD-based generative model search, and MMD-based generative adversarial networks.
Authors: Alex Morehead, Nabin Giri, Jian Liu, Pawan Neupane, Jianlin Cheng
Abstract: The effects of ligand binding on protein structures and their in vivo functions carry numerous implications for modern biomedical research and biotechnology development efforts such as drug discovery. Although several deep learning (DL) methods and benchmarks designed for protein-ligand docking have recently been introduced, to date no prior works have systematically studied the behavior of the latest docking and structure prediction methods within the broadly applicable context of (1) using predicted (apo) protein structures for docking (e.g., for applicability to new proteins); (2) binding multiple (cofactor) ligands concurrently to a given target protein (e.g., for enzyme design); and (3) having no prior knowledge of binding pockets (e.g., for generalization to unknown pockets). To enable a deeper understanding of docking methods' real-world utility, we introduce PoseBench, the first comprehensive benchmark for broadly applicable protein-ligand docking. PoseBench enables researchers to rigorously and systematically evaluate DL methods for apo-to-holo protein-ligand docking and protein-ligand structure prediction using both primary ligand and multi-ligand benchmark datasets, the latter of which we introduce for the first time to the DL community. Empirically, using PoseBench, we find that (1) DL co-folding methods generally outperform comparable conventional and DL docking baselines, yet popular methods such as AlphaFold 3 are still challenged by prediction targets with novel protein sequences; (2) certain DL co-folding methods are highly sensitive to their input multiple sequence alignments, while others are not; and (3) DL methods struggle to strike a balance between structural accuracy and chemical specificity when predicting novel or multi-ligand protein targets. Code, data, tutorials, and benchmark results are available at https://github.com/BioinfoMachineLearning/PoseBench.
Authors: Shu Wei, Yanjie Li, Lina Yu, Weijun Li, Min Wu, Linjun Sun, Jufeng Han, Yan Pang
Abstract: The pursuit of analytical solutions for differential equations has historically been limited by the need for extensive prior knowledge and mathematical prowess; however, machine learning methods like genetic algorithms have recently been applied to this end, albeit with issues of significant time consumption and complexity. This paper presents a novel machine learning-based solver, SSDE (Symbolic Solver for Differential Equations), which employs reinforcement learning to derive symbolic closed-form solutions for various differential equations. Our evaluations on a range of ordinary and partial differential equations demonstrate that SSDE provides superior performance in achieving analytical solutions compared to other machine learning approaches.
Authors: Dimitrios Bachtis, Giulio Biroli, Aur\'elien Decelle, Beatriz Seoane
Abstract: In this paper, we investigate the feature encoding process in a prototypical energy-based generative model, the Restricted Boltzmann Machine (RBM). We start with an analytical investigation using simplified architectures and data structures, and end with numerical analysis of real trainings on real datasets. Our study tracks the evolution of the model's weight matrix through its singular value decomposition, revealing a series of phase transitions associated to a progressive learning of the principal modes of the empirical probability distribution. The model first learns the center of mass of the modes and then progressively resolve all modes through a cascade of phase transitions. We first describe this process analytically in a controlled setup that allows us to study analytically the training dynamics. We then validate our theoretical results by training the Bernoulli-Bernoulli RBM on real data sets. By using data sets of increasing dimension, we show that learning indeed leads to sharp phase transitions in the high-dimensional limit. Moreover, we propose and test a mean-field finite-size scaling hypothesis. This shows that the first phase transition is in the same universality class of the one we studied analytically, and which is reminiscent of the mean-field paramagnetic-to-ferromagnetic phase transition.
Authors: Anna Vettoruzzo, Lorenzo Braccaioli, Joaquin Vanschoren, Marlena Nowaczyk
Abstract: Unsupervised meta-learning aims to learn feature representations from unsupervised datasets that can transfer to downstream tasks with limited labeled data. In this paper, we propose a novel approach to unsupervised meta-learning that leverages the generalization abilities of in-context learning observed in transformer architectures. Our method reframes meta-learning as a sequence modeling problem, enabling the transformer encoder to learn task context from support images and utilize it to predict query images. At the core of our approach lies the creation of diverse tasks generated using a combination of data augmentations and a mixing strategy that challenges the model during training while fostering generalization to unseen tasks at test time. Experimental results on benchmark datasets showcase the superiority of our approach over existing unsupervised meta-learning baselines, establishing it as the new state-of-the-art. Remarkably, our method achieves competitive results with supervised and self-supervised approaches, underscoring its efficacy in leveraging generalization over memorization.
Authors: Riccardo Cadei, Lukas Lindorfer, Sylvia Cremer, Cordelia Schmid, Francesco Locatello
Abstract: Machine Learning and AI have the potential to transform data-driven scientific discovery, enabling accurate predictions for several scientific phenomena. As many scientific questions are inherently causal, this paper looks at the causal inference task of treatment effect estimation, where the outcome of interest is recorded in high-dimensional observations in a Randomized Controlled Trial (RCT). Despite being the simplest possible causal setting and a perfect fit for deep learning, we theoretically find that many common choices in the literature may lead to biased estimates. To test the practical impact of these considerations, we recorded ISTAnt, the first real-world benchmark for causal inference downstream tasks on high-dimensional observations as an RCT studying how garden ants (Lasius neglectus) respond to microparticles applied onto their colony members by hygienic grooming. Comparing 6 480 models fine-tuned from state-of-the-art visual backbones, we find that the sampling and modeling choices significantly affect the accuracy of the causal estimate, and that classification accuracy is not a proxy thereof. We further validated the analysis, repeating it on a synthetically generated visual data set controlling the causal model. Our results suggest that future benchmarks should carefully consider real downstream scientific questions, especially causal ones. Further, we highlight guidelines for representation learning methods to help answer causal questions in the sciences.
Authors: Saber Malekmohammadi, Afaf Taik, Golnoosh Farnadi
Abstract: Federated learning (FL), which is a decentralized machine learning (ML) approach, often incorporates differential privacy (DP) to provide rigorous data privacy guarantees. Previous works attempted to address high structured data heterogeneity in vanilla FL settings through clustering clients (a.k.a clustered FL), but these methods remain sensitive and prone to errors, further exacerbated by the DP noise. This vulnerability makes the previous methods inappropriate for differentially private FL (DPFL) settings with structured data heterogeneity. To address this gap, we propose an algorithm for differentially private clustered FL, which is robust to the DP noise in the system and identifies the underlying clients' clusters correctly. To this end, we propose to cluster clients based on both their model updates and training loss values. Furthermore, for clustering clients' model updates at the end of the first round, our proposed approach addresses the server's uncertainties by employing large batch sizes as well as Gaussian Mixture Models (GMM) to reduce the impact of DP and stochastic noise and avoid potential clustering errors. This idea is efficient especially in privacy-sensitive scenarios with more DP noise. We provide theoretical analysis to justify our approach and evaluate it across diverse data distributions and privacy budgets. Our experimental results show its effectiveness in addressing large structured data heterogeneity in DPFL.
Authors: Jianan Zhao, Zhaocheng Zhu, Mikhail Galkin, Hesham Mostafa, Michael Bronstein, Jian Tang
Abstract: One fundamental challenge in graph machine learning is generalizing to new graphs. Many existing methods following the inductive setup can generalize to test graphs with new structures, but assuming the feature and label spaces remain the same as the training ones. This paper introduces the fully-inductive setup, where models should perform inference on arbitrary test graphs with new structures, feature and label spaces. We propose GraphAny as the first attempt at this challenging setup. GraphAny models inference on a new graph as an analytical solution to a LinearGNN, which can be naturally applied to graphs with any feature and label spaces. To further build a stronger model with learning capacity, we fuse multiple LinearGNN predictions with learned inductive attention scores. Specifically, the attention module is carefully parameterized as a function of the entropy-normalized distance features between pairs of LinearGNN predictions to ensure generalization to new graphs. Empirically, GraphAny trained on a single Wisconsin dataset with only 120 labeled nodes can generalize to 30 new graphs with an average accuracy of 67.26%, surpassing not only all inductive baselines, but also strong transductive methods trained separately on each of the 30 test graphs.
Authors: Wen-Pu Cai, Ming-Yang Li, Wu-Jun Li
Abstract: Large language models~(LLMs) have recently demonstrated promising performance in many tasks. However, the high storage and computational cost of LLMs has become a challenge for deploying LLMs. Weight quantization has been widely used for model compression, which can reduce both storage and computational cost. Most existing weight quantization methods for LLMs use a rank-one codebook for quantization, which results in substantial accuracy loss when the compression ratio is high. In this paper, we propose a novel weight quantization method, called low-rank codebook based quantization~(LCQ), for LLMs. LCQ adopts a low-rank codebook, the rank of which can be larger than one, for quantization. Experiments show that LCQ can achieve better accuracy than existing methods with a negligibly extra storage cost.
Authors: Zhouxing Shi, Qirui Jin, Zico Kolter, Suman Jana, Cho-Jui Hsieh, Huan Zhang
Abstract: Branch-and-bound (BaB) is among the most effective techniques for neural network (NN) verification. However, existing works on BaB for NN verification have mostly focused on NNs with piecewise linear activations, especially ReLU networks. In this paper, we develop a general framework, named GenBaB, to conduct BaB on general nonlinearities to verify NNs with general architectures, based on linear bound propagation for NN verification. To decide which neuron to branch, we design a new branching heuristic which leverages linear bounds as shortcuts to efficiently estimate the potential improvement after branching. To decide nontrivial branching points for general nonlinear functions, we propose to pre-optimize branching points, which can be efficiently leveraged during verification with a lookup table. We demonstrate the effectiveness of our GenBaB on verifying a wide range of NNs, including NNs with activation functions such as Sigmoid, Tanh, Sine and GeLU, as well as NNs involving multi-dimensional nonlinear operations such as multiplications in LSTMs and Vision Transformers. Our framework also allows the verification of general nonlinear computation graphs and enables verification applications beyond simple NNs, particularly for AC Optimal Power Flow (ACOPF). GenBaB is part of the latest $\alpha$,$\beta$-CROWN, the winner of the 4th and the 5th International Verification of Neural Networks Competition (VNN-COMP 2023 and 2024). Code for reproducing the experiments is available at https://github.com/shizhouxing/GenBaB.
Authors: Valerii Iakovlev, Harri L\"ahdesm\"aki
Abstract: Spatiotemporal dynamics models are fundamental for various domains, from heat propagation in materials to oceanic and atmospheric flows. However, currently available neural network-based spatiotemporal modeling approaches fall short when faced with data that is collected randomly over time and space, as is often the case with sensor networks in real-world applications like crowdsourced earthquake detection or pollution monitoring. In response, we developed a new method that can effectively learn spatiotemporal dynamics from such point process observations. Our model integrates techniques from neural differential equations, neural point processes, implicit neural representations and amortized variational inference to model both the dynamics of the system and the probabilistic locations and timings of observations. It outperforms existing methods on challenging spatiotemporal datasets by offering substantial improvements in predictive accuracy and computational efficiency, making it a useful tool for modeling and understanding complex dynamical systems observed under realistic, unconstrained conditions.
Authors: Sofie Goethals, Eoin Delaney, Brent Mittelstadt, Chris Russell
Abstract: Access to resources strongly constrains the decisions we make. While we might wish to offer every student a scholarship, or schedule every patient for follow-up meetings with a specialist, limited resources mean that this is not possible. When deploying machine learning systems, these resource constraints are simply enforced by varying the threshold of a classifier. However, these finite resource limitations are disregarded by most existing tools for fair machine learning, which do not allow the specification of resource limitations and do not remain fair when varying thresholds. This makes them ill-suited for real-world deployment. Our research introduces the concept of "resource-constrained fairness" and quantifies the cost of fairness within this framework. We demonstrate that the level of available resources significantly influences this cost, a factor overlooked in previous evaluations.
Authors: Zijing Ou, Mingtian Zhang, Andi Zhang, Tim Z. Xiao, Yingzhen Li, David Barber
Abstract: The probabilistic diffusion model has become highly effective across various domains. Typically, sampling from a diffusion model involves using a denoising distribution characterized by a Gaussian with a learned mean and either fixed or learned covariances. In this paper, we leverage the recently proposed covariance moment matching technique and introduce a novel method for learning the diagonal covariance. Unlike traditional data-driven diagonal covariance approximation approaches, our method involves directly regressing the optimal diagonal analytic covariance using a new, unbiased objective named Optimal Covariance Matching (OCM). This approach can significantly reduce the approximation error in covariance prediction. We demonstrate how our method can substantially enhance the sampling efficiency, recall rate and likelihood of commonly used diffusion models.
Authors: Georgi Ganev, Meenatchi Sundaram Muthu Selva Annamalai, Emiliano De Cristofaro
Abstract: Synthetic data created by differentially private (DP) generative models is increasingly used in real-world settings. In this context, PATE-GAN has emerged as one of the most popular algorithms, combining Generative Adversarial Networks (GANs) with the private training approach of PATE (Private Aggregation of Teacher Ensembles). In this paper, we set out to reproduce the utility evaluation from the original PATE-GAN paper, compare available implementations, and conduct a privacy audit. More precisely, we analyze and benchmark six open-source PATE-GAN implementations, including three by (a subset of) the original authors. First, we shed light on architecture deviations and empirically demonstrate that none reproduce the utility performance reported in the original paper. We then present an in-depth privacy evaluation, which includes DP auditing, and show that all implementations leak more privacy than intended. Furthermore, we uncover 19 privacy violations and 5 other bugs in these six open-source implementations. Lastly, our codebase is available from: https://github.com/spalabucr/pategan-audit.
Authors: Pedro Valdeira, Jo\~ao Xavier, Cl\'audia Soares, Yuejie Chi
Abstract: Communication overhead is a known bottleneck in federated learning (FL). To address this, lossy compression is commonly used on the information communicated between the server and clients during training. In horizontal FL, where each client holds a subset of the samples, such communication-compressed training methods have recently seen significant progress. However, in their vertical FL counterparts, where each client holds a subset of the features, our understanding remains limited. To address this, we propose an error feedback compressed vertical federated learning (EF-VFL) method to train split neural networks. In contrast to previous communication-compressed methods for vertical FL, EF-VFL does not require a vanishing compression error for the gradient norm to converge to zero for smooth nonconvex problems. By leveraging error feedback, our method can achieve a $\mathcal{O}(1/T)$ convergence rate for a sufficiently large batch size, improving over the state-of-the-art $\mathcal{O}(1/\sqrt{T})$ rate under $\mathcal{O}(1/\sqrt{T})$ compression error, and matching the rate of uncompressed methods. Further, when the objective function satisfies the Polyak-{\L}ojasiewicz inequality, our method converges linearly. In addition to improving convergence, our method also supports the use of private labels. Numerical experiments show that EF-VFL significantly improves over the prior art, confirming our theoretical results. The code for this work can be found at https://github.com/Valdeira/EF-VFL.
Authors: Dong Liu
Abstract: Graph Neural Networks (GNNs) have become fundamental in semi-supervised learning for graph representation, leveraging their ability to capture complex node relationships. A recent trend in GNN research focuses on \textbf{adaptive k-hop structure learning}, moving beyond fixed-hop aggregation to more flexible and dynamic neighborhood selection. While GAMLP \cite{Zhang_2022} employs separate MLP layers for each k-hop domain and ImprovingTE \cite{Yao2023ImprovingTE} enhances this by injecting contextualized substructure information, these methods still rely heavily on predefined sampling strategies, which may limit their ability to generalize and maintain stable accuracy. To address these limitations, we propose an \textbf{adaptive reconstruction framework} that dynamically refines k-hop structure learning. Inspired by "coreset selection" \cite{guo2022deepcore}, our approach adaptively \textbf{reconstructs} node neighborhoods to optimize message passing, ensuring more \textbf{effective and context-aware information flow} across the graph. To further enhance structural robustness, we introduce two key modules: the \textbf{Distance Recomputator} and the \textbf{Topology Reconstructor} (\textcolor{blue}{DRTR}). The Distance Recomputator \textbf{reassesses and recalibrates} node distances based on adaptive graph properties, leading to \textbf{improved node embeddings} that better reflect latent relationships. Meanwhile, the Topology Reconstructor \textbf{dynamically refines local graph structures}, enabling the model to \textbf{adapt to evolving graph topologies} and mitigate the impact of noise and mislabeled data. Empirical evaluations demonstrate that our \textbf{adaptive reconstruction framework} achieves \textbf{significant improvements} over existing k-hop-based models, providing more \textbf{stable and accurate} performance in various graph learning benchmarks.
Authors: Dong Liu, Yanxuan Yu
Abstract: Efficient machine learning (ML) has become increasingly important as models grow larger and data volumes expand. In this work, we address the trade-off between generalization in multi-task learning (MTL) and precision in single-task learning (STL) by introducing the Multi-Task to Single-Task (MT2ST) framework. MT2ST is designed to enhance training efficiency and accuracy in word embedding tasks, showcasing its value as a practical application of efficient ML. Our framework employs two strategies: *Diminish*, which gradually reduces the influence of auxiliary tasks, and *Switch*, which transitions training from MTL to STL at a specific point. Empirical results show that MT2ST reduces training time by 67\% compared to STL and by 13\% compared to traditional MTL, while maintaining high accuracy. These findings highlight MT2ST as an efficient ML solution tailored for optimizing word embedding training. Code is available at https://github.com/NoakLiu/MT2ST.
Authors: Gaojie Jin, Ronghui Mu, Xinping Yi, Xiaowei Huang, Lijun Zhang
Abstract: The Invariant Risk Minimization (IRM) approach aims to address the challenge of domain generalization by training a feature representation that remains invariant across multiple environments. However, in noisy environments, IRM-related techniques such as IRMv1 and VREx may be unable to achieve the optimal IRM solution, primarily due to erroneous optimization directions. To address this issue, we introduce ICorr (an abbreviation for Invariant Correlation), a novel approach designed to surmount the above challenge in noisy settings. Additionally, we dig into a case study to analyze why previous methods may lose ground while ICorr can succeed. Through a theoretical lens, particularly from a causality perspective, we illustrate that the invariant correlation of representation with label is a necessary condition for the optimal invariant predictor in noisy environments, whereas the optimization motivations for other methods may not be. Furthermore, we empirically demonstrate the effectiveness of ICorr by comparing it with other domain generalization methods on various noisy datasets. The code is available at https://github.com/Alexkael/ICorr.
Authors: Weiwei Kong, M\'onica Ribero
Abstract: Differentially-private stochastic gradient descent (DP-SGD) is a family of iterative machine learning training algorithms that privatize gradients to generate a sequence of differentially-private (DP) model parameters. It is also the standard tool used to train DP models in practice, even though most users are only interested in protecting the privacy of the final model. Tight DP accounting for the last iterate would minimize the amount of noise required while maintaining the same privacy guarantee and potentially increasing model utility. However, last-iterate accounting is challenging, and existing works require strong assumptions not satisfied by most implementations. These include assuming (i) the global sensitivity constant is known - to avoid gradient clipping; (ii) the loss function is Lipschitz or convex; and (iii) input batches are sampled randomly. In this work, we forego any unrealistic assumptions and provide privacy bounds for the most commonly used variant of DP-SGD, in which data is traversed cyclically, gradients are clipped, and only the last model is released. More specifically, we establish new Renyi differential privacy (RDP) upper bounds for the last iterate under realistic assumptions of small stepsize and Lipschitz smoothness of the loss function. Our general bounds also recover the special-case convex bounds when the weak-convexity parameter of the objective function approaches zero and no clipping is performed. The approach itself leverages optimal transport techniques for last iterate bounds, which is a nontrivial task when the data is traversed cyclically and the loss function is nonconvex.
Authors: Luoxiao Yang, Yun Wang, Xinqi Fan, Israel Cohen, Jingdong Chen, Yue Zhao, Zijun Zhang
Abstract: Time series forecasting (TSF) possesses great practical values in various fields, including power and energy, transportation, etc. TSF methods have been studied based on knowledge from classical statistics to modern deep learning. Yet, all of them were developed based on one fundamental concept, the numerical data fitting. Thus, the models developed have been long known for being problem-specific and lacking application generalizability. A TSF foundation model serving TSF tasks across different applications can reverse such an impression. The central question is then how to develop such a TSF foundation model. This paper offers a pioneering study in developing a TSF foundation model and proposes a vision intelligence-powered framework, ViTime, for the first time. In ViTime, a method synthesizing authentic time series periodic and trend patterns is developed to enrich sample pattern diversity. A deep architecture operating TSF in image metric space is designed to achieve significantly enhanced TSF generalizability. Extensive experiments demonstrate ViTime's SOTA performance across multiple settings. In zero-shot scenarios, ViTime outperforms TimesFM by 9-15%. With just 10% fine-tuning data, ViTime surpasses both foundation models and fully-supervised benchmarks trained on complete datasets, with this performance gap widening further at 100\% fine-tuning. Additionally, ViTime exhibits exceptional robustness, handling missing data without imputation and outperforming TimesFM by 20-30% under various data perturbations.
Authors: Jiawei Huang, Vinzenz Thoma, Zebang Shen, Heinrich H. Nax, Niao He
Abstract: Designing incentives for an adapting population is a ubiquitous problem in a wide array of economic applications and beyond. In this work, we study how to design additional rewards to steer multi-agent systems towards desired policies \emph{without} prior knowledge of the agents' underlying learning dynamics. Motivated by the limitation of existing works, we consider a new and general category of learning dynamics called \emph{Markovian agents}. We introduce a model-based non-episodic Reinforcement Learning (RL) formulation for our steering problem. Importantly, we focus on learning a \emph{history-dependent} steering strategy to handle the inherent model uncertainty about the agents' learning dynamics. We introduce a novel objective function to encode the desiderata of achieving a good steering outcome with reasonable cost. Theoretically, we identify conditions for the existence of steering strategies to guide agents to the desired policies. Complementing our theoretical contributions, we provide empirical algorithms to approximately solve our objective, which effectively tackles the challenge in learning history-dependent strategies. We demonstrate the efficacy of our algorithms through empirical evaluations.
Authors: Jung Hyun Lee, Jeonghoon Kim, June Yong Yang, Se Jung Kwon, Eunho Yang, Kang Min Yoo, Dongsoo Lee
Abstract: With the commercialization of large language models (LLMs), weight-activation quantization has emerged to compress and accelerate LLMs, achieving high throughput while reducing inference costs. However, existing post-training quantization (PTQ) techniques for quantizing weights and activations of LLMs still suffer from non-negligible accuracy drops, especially on massive multitask language understanding. To address this issue, we propose Low-Rank Quantization (LRQ) - a simple yet effective post-training weight quantization method for LLMs that reconstructs the outputs of an intermediate Transformer block by leveraging low-rank weight-scaling matrices, replacing the conventional full weight-scaling matrices that entail as many learnable scales as their associated weights. Thanks to parameter sharing via low-rank structure, LRQ only needs to learn significantly fewer parameters while enabling the individual scaling of weights, thus boosting the generalization capability of quantized LLMs. We show the superiority of LRQ over prior LLM PTQ works under (i) 8-bit weight and per-tensor activation quantization, (ii) 4-bit weight and 8-bit per-token activation quantization, and (iii) low-bit weight-only quantization schemes. Our code is available at Software.
Authors: Maximilian Egger, Rawad Bitar, Ghadir Ayache, Antonia Wachter-Zeh, Salim El Rouayheb
Abstract: Consider the setting of multiple random walks (RWs) on a graph executing a certain computational task. For instance, in decentralized learning via RWs, a model is updated at each iteration based on the local data of the visited node and then passed to a randomly chosen neighbor. RWs can fail due to node or link failures. The goal is to maintain a desired number of RWs to ensure failure resilience. Achieving this is challenging due to the lack of a central entity to track which RWs have failed to replace them with new ones by forking (duplicating) surviving ones. Without duplications, the number of RWs will eventually go to zero, causing a catastrophic failure of the system. We propose two decentralized algorithms called DecAFork and DecAFork+ that can maintain the number of RWs in the graph around a desired value even in the presence of arbitrary RW failures. Nodes continuously estimate the number of surviving RWs by estimating their return time distribution and fork the RWs when failures are likely to happen. DecAFork+ additionally allows terminations to avoid overloading the network by forking too many RWs. We present extensive numerical simulations that show the performance of DecAFork and DecAFork+ regarding fast detection and reaction to failures compared to a baseline, and establish theoretical guarantees on the performance of both algorithms.
Authors: Xinshuai Dong, Ignavier Ng, Biwei Huang, Yuewen Sun, Songyao Jin, Roberto Legaspi, Peter Spirtes, Kun Zhang
Abstract: Linear causal models are important tools for modeling causal dependencies and yet in practice, only a subset of the variables can be observed. In this paper, we examine the parameter identifiability of these models by investigating whether the edge coefficients can be recovered given the causal structure and partially observed data. Our setting is more general than that of prior research - we allow all variables, including both observed and latent ones, to be flexibly related, and we consider the coefficients of all edges, whereas most existing works focus only on the edges between observed variables. Theoretically, we identify three types of indeterminacy for the parameters in partially observed linear causal models. We then provide graphical conditions that are sufficient for all parameters to be identifiable and show that some of them are provably necessary. Methodologically, we propose a novel likelihood-based parameter estimation method that addresses the variance indeterminacy of latent variables in a specific way and can asymptotically recover the underlying parameters up to trivial indeterminacy. Empirical studies on both synthetic and real-world datasets validate our identifiability theory and the effectiveness of the proposed method in the finite-sample regime. Code: https://github.com/dongxinshuai/scm-identify.
Authors: Dario Shariatian, Umut Simsekli, Alain Durmus
Abstract: Exploring noise distributions beyond Gaussian in diffusion models remains an open challenge. While Gaussian-based models succeed within a unified SDE framework, recent studies suggest that heavy-tailed noise distributions, like $\alpha$-stable distributions, may better handle mode collapse and effectively manage datasets exhibiting class imbalance, heavy tails, or prominent outliers. Recently, Yoon et al.\ (NeurIPS 2023), presented the L\'evy-It\^o model (LIM), directly extending the SDE-based framework to a class of heavy-tailed SDEs, where the injected noise followed an $\alpha$-stable distribution, a rich class of heavy-tailed distributions. However, the LIM framework relies on highly involved mathematical techniques with limited flexibility, potentially hindering broader adoption and further development. In this study, instead of starting from the SDE formulation, we extend the denoising diffusion probabilistic model (DDPM) by replacing the Gaussian noise with $\alpha$-stable noise. By using only elementary proof techniques, the proposed approach, Denoising L\'evy Probabilistic Models (DLPM), boils down to vanilla DDPM with minor modifications. As opposed to the Gaussian case, DLPM and LIM yield different training algorithms and different backward processes, leading to distinct sampling algorithms. These fundamental differences translate favorably for DLPM as compared to LIM: our experiments show improvements in coverage of data distribution tails, better robustness to unbalanced datasets, and improved computation times requiring smaller number of backward steps.
Authors: Xiao Li, Wenxuan Sun, Huanran Chen, Qiongxiu Li, Yining Liu, Yingzhe He, Jie Shi, Xiaolin Hu
Abstract: Recently Diffusion-based Purification (DiffPure) has been recognized as an effective defense method against adversarial examples. However, we find DiffPure which directly employs the original pre-trained diffusion models for adversarial purification, to be suboptimal. This is due to an inherent trade-off between noise purification performance and data recovery quality. Additionally, the reliability of existing evaluations for DiffPure is questionable, as they rely on weak adaptive attacks. In this work, we propose a novel Adversarial Diffusion Bridge Model, termed ADBM. ADBM directly constructs a reverse bridge from the diffused adversarial data back to its original clean examples, enhancing the purification capabilities of the original diffusion models. Through theoretical analysis and experimental validation across various scenarios, ADBM has proven to be a superior and robust defense mechanism, offering significant promise for practical applications.
Authors: Rishub Tamirisa, Bhrugu Bharathi, Long Phan, Andy Zhou, Alice Gatti, Tarun Suresh, Maxwell Lin, Justin Wang, Rowan Wang, Ron Arel, Andy Zou, Dawn Song, Bo Li, Dan Hendrycks, Mantas Mazeika
Abstract: Rapid advances in the capabilities of large language models (LLMs) have raised widespread concerns regarding their potential for malicious use. Open-weight LLMs present unique challenges, as existing safeguards lack robustness to tampering attacks that modify model weights. For example, recent works have demonstrated that refusal and unlearning safeguards can be trivially removed with a few steps of fine-tuning. These vulnerabilities necessitate new approaches for enabling the safe release of open-weight LLMs. We develop a method, called TAR, for building tamper-resistant safeguards into open-weight LLMs such that adversaries cannot remove the safeguards even after hundreds of steps of fine-tuning. In extensive evaluations and red teaming analyses, we find that our method greatly improves tamper-resistance while preserving benign capabilities. Our results demonstrate that progress on tamper-resistance is possible, opening up a promising new avenue to improve the safety and security of open-weight LLMs.
Authors: Yannik Schnitzer, Alessandro Abate, David Parker
Abstract: We present a data-driven approach for producing policies that are provably robust across unknown stochastic environments. Existing approaches can learn models of a single environment as an interval Markov decision processes (IMDP) and produce a robust policy with a probably approximately correct (PAC) guarantee on its performance. However these are unable to reason about the impact of environmental parameters underlying the uncertainty. We propose a framework based on parametric Markov decision processes (MDPs) with unknown distributions over parameters. We learn and analyse IMDPs for a set of unknown sample environments induced by parameters. The key challenge is then to produce meaningful performance guarantees that combine the two layers of uncertainty: (1) multiple environments induced by parameters with an unknown distribution; (2) unknown induced environments which are approximated by IMDPs. We present a novel approach based on scenario optimisation that yields a single PAC guarantee quantifying the risk level for which a specified performance level can be assured in unseen environments, plus a means to trade-off risk and performance. We implement and evaluate our framework using multiple robust policy generation methods on a range of benchmarks. We show that our approach produces tight bounds on a policy's performance with high confidence.
Authors: Vasudev Shyam, Jonathan Pilault, Emily Shepperd, Quentin Anthony, Beren Millidge
Abstract: Our formulation reveals that the reduction across the sequence axis can be efficiently computed in parallel through a tree reduction. Our algorithm, called Tree Attention, for parallelizing exact attention computation across multiple GPUs enables cross-device decoding to be performed asymptotically faster (up to 8x faster in our experiments) than state-of-the-art approaches such as Ring Attention, while also requiring significantly less communication volume and incurring 2x less peak memory. We demonstrate that Tree Attention speeds up decoding up to 4x on Llama 3.1-8B and can be applied to a variety of hardware and networking setups such as H100 DGX nodes, AMD MI300x nodes, and PCIe connected NVIDIA RTX 4090s. Our code is publicly available here: https://github.com/Zyphra/tree_attention
Authors: Ignacy St\k{e}pka, Mateusz Lango, Jerzy Stefanowski
Abstract: Counterfactual explanations (CFEs) guide users on how to adjust inputs to machine learning models to achieve desired outputs. While existing research primarily addresses static scenarios, real-world applications often involve data or model changes, potentially invalidating previously generated CFEs and rendering user-induced input changes ineffective. Current methods addressing this issue often support only specific models or change types, require extensive hyperparameter tuning, or fail to provide probabilistic guarantees on CFE robustness to model changes. This paper proposes a novel approach for generating CFEs that provides probabilistic guarantees for any model and change type, while offering interpretable and easy-to-select hyperparameters. We establish a theoretical framework for probabilistically defining robustness to model change and demonstrate how our BetaRCE method directly stems from it. BetaRCE is a post-hoc method applied alongside a chosen base CFE generation method to enhance the quality of the explanation beyond robustness. It facilitates a transition from the base explanation to a more robust one with user-adjusted probability bounds. Through experimental comparisons with baselines, we show that BetaRCE yields robust, most plausible, and closest to baseline counterfactual explanations.
Authors: Guangyan Sun, Mingyu Jin, Zhenting Wang, Cheng-Long Wang, Siqi Ma, Qifan Wang, Tong Geng, Ying Nian Wu, Yongfeng Zhang, Dongfang Liu
Abstract: Achieving human-level intelligence requires refining cognitive distinctions between System 1 and System 2 thinking. While contemporary AI, driven by large language models, demonstrates human-like traits, it falls short of genuine cognition. Transitioning from structured benchmarks to real-world scenarios presents challenges for visual agents, often leading to inaccurate and overly confident responses. To address the challenge, we introduce FaST, which incorporates the Fast and Slow Thinking mechanism into visual agents. FaST employs a switch adapter to dynamically select between System 1/2 modes, tailoring the problem-solving approach to different task complexity. It tackles uncertain and unseen objects by adjusting model confidence and integrating new contextual data. With this novel design, we advocate a flexible system, hierarchical reasoning capabilities, and a transparent decision-making pipeline, all of which contribute to its ability to emulate human-like cognitive processes in visual intelligence. Empirical results demonstrate that FaST outperforms various well-known baselines, achieving 80.8% accuracy over VQA^{v2} for visual question answering and 48.7% GIoU score over ReasonSeg for reasoning segmentation, demonstrate FaST's superior performance. Extensive testing validates the efficacy and robustness of FaST's core components, showcasing its potential to advance the development of cognitive visual agents in AI systems. The code is available at ttps://github.com/GuangyanS/Sys2-LLaVA.
Authors: Aviv Bick, Kevin Y. Li, Eric P. Xing, J. Zico Kolter, Albert Gu
Abstract: Transformer architectures have become a dominant paradigm for domains like language modeling but suffer in many inference settings due to their quadratic-time self-attention. Recently proposed subquadratic architectures, such as Mamba, have shown promise, but have been pretrained with substantially less computational resources than the strongest Transformer models. In this work, we present a method that is able to distill a pretrained Transformer architecture into alternative architectures such as state space models (SSMs). The key idea to our approach is that we can view both Transformers and SSMs as applying different forms of mixing matrices over the token sequences. We can thus progressively distill the Transformer architecture by matching different degrees of granularity in the SSM: first matching the mixing matrices themselves, then the hidden units at each block, and finally the end-to-end predictions. Our method, called MOHAWK, is able to distill a Mamba-2 variant based on the Phi-1.5 architecture (Phi-Mamba) using only 3B tokens and a hybrid version (Hybrid Phi-Mamba) using 5B tokens. Despite using less than 1% of the training data typically used to train models from scratch, Phi-Mamba boasts substantially stronger performance compared to all past open-source non-Transformer models. MOHAWK allows models like SSMs to leverage computational resources invested in training Transformer-based architectures, highlighting a new avenue for building such models.
Authors: Tom Segal, Asaf Shabtai, Yuval Elovici
Abstract: The utility of large language models (LLMs) depends heavily on the quality and quantity of their training data. Many organizations possess large data corpora that could be leveraged to train or fine-tune LLMs tailored to their specific needs. However, these datasets often come with access restrictions that are based on user privileges and enforced by access control mechanisms. Training LLMs on such datasets could result in exposure of sensitive information to unauthorized users. A straightforward approach for preventing such exposure is to train a separate model for each access level. This, however, may result in low utility models due to the limited amount of training data per model compared to the amount in the entire organizational corpus. Another approach is to train a single LLM on all the data while limiting the exposure of unauthorized information. However, current exposure-limiting methods for LLMs are ineffective for access-controlled data, where sensitive information appears frequently across many training examples. We propose DOMBA - double model balancing - a simple approach for training and deploying LLMs that provides high utility and access-control functionality with security guarantees. DOMBA aggregates the probability distributions of two models, each trained on documents with (potentially many) different access levels, using a "min-bounded" average function (a function that is bounded by the smaller value, e.g., harmonic mean). A detailed mathematical analysis and extensive evaluation show that DOMBA safeguards restricted information while offering utility comparable to non-secure models.
Authors: Weiyu Chen, James Kwok
Abstract: Model merging, which combines multiple models into a single model, has gained popularity in recent years. By efficiently integrating the capabilities of various models, this significantly reduces the parameter count and memory usage. However, current methods can only produce one single merged model. This necessitates a performance trade-off due to conflicts among the various models, and the resultant one-size-fits-all model may not align with the preferences of different users who may prioritize certain models over others. To address this issue, we propose preference-aware model merging, and formulate this as a multi-objective optimization problem in which the performance of the merged model on each base model's task is treated as an objective. In a single merging process, the proposed parameter-efficient structure generates a Pareto set of merged models, with each representing a Pareto-optimal solution for a preference. Users can then select merged models tailored to their preferences from this learned Pareto set. Experimental results demonstrate that the proposed Pareto Merging produces diverse trade-off models and achieves higher test accuracy compared to state-of-the-art merging baselines.
Authors: Jie Feng, Yuwei Du, Jie Zhao, Yong Li
Abstract: Next location prediction plays a crucial role in various real-world applications. Recently, due to the limitation of existing deep learning methods, attempts have been made to apply large language models (LLMs) to zero-shot next location prediction task. However, they directly generate the final output using LLMs without systematic design, which limits the potential of LLMs to uncover complex mobility patterns and underestimates their extensive reserve of global geospatial knowledge. In this paper, we introduce AgentMove, a systematic agentic prediction framework to achieve generalized next location prediction. In AgentMove, we first decompose the mobility prediction task and design specific modules to complete them, including spatial-temporal memory for individual mobility pattern mining, world knowledge generator for modeling the effects of urban structure and collective knowledge extractor for capturing the shared patterns among population. Finally, we combine the results of three modules and conduct a reasoning step to generate the final predictions. Extensive experiments utilizing mobility data from two distinct sources reveal that AgentMove surpasses the leading baseline by 3.33% to 8.57% across 8 out of 12 metrics and it shows robust predictions with various LLMs as base and also less geographical bias across cities. Our codes are available via https://github.com/tsinghua-fib-lab/AgentMove.
Authors: Toshinori Kitamura, Tadashi Kozuno, Wataru Kumagai, Kenta Hoshino, Yohei Hosoe, Kazumi Kasaura, Masashi Hamaya, Paavo Parmas, Yutaka Matsuo
Abstract: Designing a safe policy for uncertain environments is crucial in real-world control systems. However, this challenge remains inadequately addressed within the Markov decision process (MDP) framework. This paper presents the first algorithm guaranteed to identify a near-optimal policy in a robust constrained MDP (RCMDP), where an optimal policy minimizes cumulative cost while satisfying constraints in the worst-case scenario across a set of environments. We first prove that the conventional policy gradient approach to the Lagrangian max-min formulation can become trapped in suboptimal solutions. This occurs when its inner minimization encounters a sum of conflicting gradients from the objective and constraint functions. To address this, we leverage the epigraph form of the RCMDP problem, which resolves the conflict by selecting a single gradient from either the objective or the constraints. Building on the epigraph form, we propose a bisection search algorithm with a policy gradient subroutine and prove that it identifies an $\varepsilon$-optimal policy in an RCMDP with $\widetilde{\mathcal{O}}(\varepsilon^{-4})$ robust policy evaluations.
Authors: Conor Heins, Hao Wu, Dimitrije Markovic, Alexander Tschantz, Jeff Beck, Christopher Buckley
Abstract: Balancing computational efficiency with robust predictive performance is crucial in supervised learning, especially for critical applications. Standard deep learning models, while accurate and scalable, often lack probabilistic features like calibrated predictions and uncertainty quantification. Bayesian methods address these issues but can be computationally expensive as model and data complexity increase. Previous work shows that fast variational methods can reduce the compute requirements of Bayesian methods by eliminating the need for gradient computation or sampling, but are often limited to simple models. We introduce CAVI-CMN, a fast, gradient-free variational method for training conditional mixture networks (CMNs), a probabilistic variant of the mixture-of-experts (MoE) model. CMNs are composed of linear experts and a softmax gating network. By exploiting conditional conjugacy and P\'olya-Gamma augmentation, we furnish Gaussian likelihoods for the weights of both the linear layers and the gating network. This enables efficient variational updates using coordinate ascent variational inference (CAVI), avoiding traditional gradient-based optimization. We validate this approach by training two-layer CMNs on standard classification benchmarks from the UCI repository. CAVI-CMN achieves competitive and often superior predictive accuracy compared to maximum likelihood estimation (MLE) with backpropagation, while maintaining competitive runtime and full posterior distributions over all model parameters. Moreover, as input size or the number of experts increases, computation time scales competitively with MLE and other gradient-based solutions like black-box variational inference (BBVI), making CAVI-CMN a promising tool for deep, fast, and gradient-free Bayesian networks.
Authors: Baturay Saglam, Dionysis Kalogerias
Abstract: Deterministic policy gradient algorithms are foundational for actor-critic methods in controlling continuous systems, yet they often encounter inaccuracies due to their dependence on the derivative of the critic's value estimates with respect to input actions. This reliance requires precise action-value gradient computations, a task that proves challenging under function approximation. We introduce an actor-critic algorithm that bypasses the need for such precision by employing a zeroth-order approximation of the action-value gradient through two-point stochastic gradient estimation within the action space. This approach provably and effectively addresses compatibility issues inherent in deterministic policy gradient schemes. Empirical results further demonstrate that our algorithm not only matches but frequently exceeds the performance of current state-of-the-art methods by a substantial extent.
Authors: M. Tanveer, R. K. Sharma, M. Sajid, A. Quadir
Abstract: The classification performance of the random vector functional link (RVFL), a randomized neural network, has been widely acknowledged. However, due to its shallow learning nature, RVFL often fails to consider all the relevant information available in a dataset. Additionally, it overlooks the geometrical properties of the dataset. To address these limitations, a novel graph random vector functional link based on multi-view learning (GRVFL-MV) model is proposed. The proposed model is trained on multiple views, incorporating the concept of multiview learning (MVL), and it also incorporates the geometrical properties of all the views using the graph embedding (GE) framework. The fusion of RVFL networks, MVL, and GE framework enables our proposed model to achieve the following: i) efficient learning: by leveraging the topology of RVFL, our proposed model can efficiently capture nonlinear relationships within the multi-view data, facilitating efficient and accurate predictions; ii) comprehensive representation: fusing information from diverse perspectives enhance the proposed model's ability to capture complex patterns and relationships within the data, thereby improving the model's overall generalization performance; and iii) structural awareness: by employing the GE framework, our proposed model leverages the original data distribution of the dataset by naturally exploiting both intrinsic and penalty subspace learning criteria. The evaluation of the proposed GRVFL-MV model on various datasets, including 27 UCI and KEEL datasets, 50 datasets from Corel5k, and 45 datasets from AwA, demonstrates its superior performance compared to baseline models. These results highlight the enhanced generalization capabilities of the proposed GRVFL-MV model across a diverse range of datasets.
Authors: Haochen Yuan, Yutong Wang, Yihong Chen, Yunbo Wang
Abstract: Time series forecasting, particularly in few-shot learning scenarios, is challenging due to the limited availability of high-quality training data. To address this, we present a pilot study on using reinforcement learning (RL) for time series data augmentation. Our method, ReAugment, tackles three critical questions: which parts of the training set should be augmented, how the augmentation should be performed, and what advantages RL brings to the process. Specifically, our approach maintains a forecasting model zoo, and by measuring prediction diversity across the models, we identify samples with higher probabilities for overfitting and use them as the anchor points for augmentation. Leveraging RL, our method adaptively transforms the overfit-prone samples into new data that not only enhances training set diversity but also directs the augmented data to target regions where the forecasting models are prone to overfitting. We validate the effectiveness of ReAugment across a wide range of base models, showing its advantages in both standard time series forecasting and few-shot learning tasks.
Authors: Siqiao Mu, Diego Klabjan
Abstract: Machine unlearning algorithms aim to efficiently remove data from a model without retraining it from scratch, in order to remove corrupted or outdated data or respect a user's ``right to be forgotten." Certified machine unlearning is a strong theoretical guarantee based on differential privacy that quantifies the extent to which an algorithm erases data from the model weights. In contrast to existing works in certified unlearning for convex or strongly convex loss functions, or nonconvex objectives with limiting assumptions, we propose the first, first-order, black-box (i.e., can be applied to models pretrained with vanilla gradient descent) algorithm for unlearning on general nonconvex loss functions, which unlearns by ``rewinding" to an earlier step during the learning process before performing gradient descent on the loss function of the retained data points. We prove $(\epsilon, \delta)$ certified unlearning and performance guarantees that establish the privacy-utility-complexity tradeoff of our algorithm, and we prove generalization guarantees for nonconvex functions that satisfy the Polyak-Lojasiewicz inequality. Finally, we implement our algorithm under a new experimental framework that more accurately reflects real-world use cases for preserving user privacy.
Authors: Maxim Fishman, Brian Chmiel, Ron Banner, Daniel Soudry
Abstract: We train, for the first time, large language models using FP8 precision on datasets up to 2 trillion tokens -- a 20-fold increase over previous limits. Through these extended training runs, we uncover critical instabilities in FP8 training that were not observable in earlier works with shorter durations. We trace these instabilities to outlier amplification by the SwiGLU activation function. Interestingly, we show, both analytically and empirically, that this amplification happens only over prolonged training periods, and link it to a SwiGLU weight alignment process. To address this newly identified issue, we introduce Smooth-SwiGLU, a novel modification that ensures stable FP8 training without altering function behavior. We also demonstrate, for the first time, FP8 quantization of both Adam optimizer moments. Combining these innovations, we successfully train a 7B parameter model using FP8 precision on 256 Intel Gaudi2 accelerators, achieving on-par results with the BF16 baseline while delivering up to a $\sim 34 \%$ throughput improvement. A reference implementation is supplied in https://github.com/Anonymous1252022/Megatron-DeepSpeed.
URLs: https://github.com/Anonymous1252022/Megatron-DeepSpeed.
Authors: Naheed Anjum Arafat, Debabrota Basu, Yulia Gel, Yuzhou Chen
Abstract: Capitalizing on the intuitive premise that shape characteristics are more robust to perturbations, we bridge adversarial graph learning with the emerging tools from computational topology, namely, persistent homology representations of graphs. We introduce the concept of witness complex to adversarial analysis on graphs, which allows us to focus only on the salient shape characteristics of graphs, yielded by the subset of the most essential nodes (i.e., landmarks), with minimal loss of topological information on the whole graph. The remaining nodes are then used as witnesses, governing which higher-order graph substructures are incorporated into the learning process. Armed with the witness mechanism, we design Witness Graph Topological Layer (WGTL), which systematically integrates both local and global topological graph feature representations, the impact of which is, in turn, automatically controlled by the robust regularized topological loss. Given the attacker's budget, we derive the important stability guarantees of both local and global topology encodings and the associated robust topological loss. We illustrate the versatility and efficiency of WGTL by its integration with five GNNs and three existing non-topological defense mechanisms. Our extensive experiments across six datasets demonstrate that WGTL boosts the robustness of GNNs across a range of perturbations and against a range of adversarial attacks. Our datasets and source codes are available at https://github.com/toggled/WGTL.
Authors: Tommaso Ferracci, Leonie Tabea Goldmann, Anton Hinel, Francesco Sanna Passino
Abstract: Data augmentation via synthetic data generation has been shown to be effective in improving model performance and robustness in the context of scarce or low-quality data. Using the data valuation framework to statistically identify beneficial and detrimental observations, we introduce a simple augmentation pipeline that generates only high-value training points based on hardness characterization, in a computationally efficient manner. We first empirically demonstrate via benchmarks on real data that Shapley-based data valuation methods perform comparably with learning-based methods in hardness characterization tasks, while offering significant computational advantages. Then, we show that synthetic data generators trained on the hardest points outperform non-targeted data augmentation on a number of tabular datasets. Our approach improves the quality of out-of-sample predictions and it is computationally more efficient compared to non-targeted methods.
Authors: Menghua Wu, Umesh Padia, Sean H. Murphy, Regina Barzilay, Tommi Jaakkola
Abstract: Identifying variables responsible for changes to a biological system enables applications in drug target discovery and cell engineering. Given a pair of observational and interventional datasets, the goal is to isolate the subset of observed variables that were the targets of the intervention. Directly applying causal discovery algorithms is challenging: the data may contain thousands of variables with as few as tens of samples per intervention, and biological systems do not adhere to classical causality assumptions. We propose a causality-inspired approach to address this practical setting. First, we infer noisy causal graphs from the observational and interventional data. Then, we learn to map the differences between these graphs, along with additional statistical features, to sets of variables that were intervened upon. Both modules are jointly trained in a supervised framework, on simulated and real data that reflect the nature of biological interventions. This approach consistently outperforms baselines for perturbation modeling on seven single-cell transcriptomics datasets. We also demonstrate significant improvements over current causal discovery methods for predicting soft and hard intervention targets across a variety of synthetic data.
Authors: Zian Li, Cai Zhou, Xiyuan Wang, Xingang Peng, Muhan Zhang
Abstract: Recent advancements in molecular generative models have demonstrated substantial potential in accelerating scientific discovery, particularly in drug design. However, these models often face challenges in generating high-quality molecules, especially in conditional scenarios where specific molecular properties must be satisfied. In this work, we introduce GeoRCG, a general framework to enhance the performance of molecular generative models by integrating geometric representation conditions with provable theoretical guarantees. We decompose the molecule generation process into two stages: first, generating an informative geometric representation; second, generating a molecule conditioned on the representation. Compared to directly generating a molecule, the relatively easy-to-generate representation in the first stage guides the second-stage generation to reach a high-quality molecule in a more goal-oriented and much faster way. Leveraging EDM and SemlaFlow as the base generators, we observe significant quality improvements in unconditional molecule generation tasks on the widely-used QM9 and GEOM-DRUG datasets. More notably, in the challenging conditional molecular generation task, our framework achieves an average 31\% performance improvement over state-of-the-art approaches, highlighting the superiority of conditioning on semantically rich geometric representations over conditioning on individual property values as in previous approaches. Furthermore, we show that, with such representation guidance, the number of diffusion steps can be reduced to as small as 100 while largely preserving the generation quality achieved with 1,000 steps, thereby significantly accelerating the generation process.
Authors: Meng Ding, Rohan Sharma, Changyou Chen, Jinhui Xu, Kaiyi Ji
Abstract: Machine Unlearning has emerged as a significant area of research, focusing on `removing' specific subsets of data from a trained model. Fine-tuning (FT) methods have become one of the fundamental approaches for approximating unlearning, as they effectively retain model performance. However, it is consistently observed that naive FT methods struggle to forget the targeted data. In this paper, we present the first theoretical analysis of FT methods for machine unlearning within a linear regression framework, providing a deeper exploration of this phenomenon. Our analysis reveals that while FT models can achieve zero remaining loss, they fail to forget the forgetting data, as the pretrained model retains its influence and the fine-tuning process does not adequately mitigate it. To address this, we propose a novel Retention-Based Masking (RBM) strategy that constructs a weight saliency map based on the remaining dataset, unlike existing methods that focus on the forgetting dataset. Our theoretical analysis demonstrates that RBM not only significantly improves unlearning accuracy (UA) but also ensures higher retaining accuracy (RA) by preserving overlapping features shared between the forgetting and remaining datasets. Experiments on synthetic and real-world datasets validate our theoretical insights, showing that RBM outperforms existing masking approaches in balancing UA, RA, and disparity metrics.
Authors: Charles Jones, Fabio de Sousa Ribeiro, M\'elanie Roschewitz, Daniel C. Castro, Ben Glocker
Abstract: We investigate the prominent class of fair representation learning methods for bias mitigation. Using causal reasoning to define and formalise different sources of dataset bias, we reveal important implicit assumptions inherent to these methods. We prove fundamental limitations on fair representation learning when evaluation data is drawn from the same distribution as training data and run experiments across a range of medical modalities to examine the performance of fair representation learning under distribution shifts. Our results explain apparent contradictions in the existing literature and reveal how rarely considered causal and statistical aspects of the underlying data affect the validity of fair representation learning. We raise doubts about current evaluation practices and the applicability of fair representation learning methods in performance-sensitive settings. We argue that fine-grained analysis of dataset biases should play a key role in the field moving forward.
Authors: Daoyu Wang, Mingyue Cheng, Zhiding Liu, Qi Liu, Enhong Chen
Abstract: Self-supervised learning has garnered increasing attention in time series analysis for benefiting various downstream tasks and reducing reliance on labeled data. Despite its effectiveness, existing methods often struggle to comprehensively capture both long-term dynamic evolution and subtle local patterns in a unified manner. In this work, we propose TimeDART, a novel self-supervised time series pre-training framework that unifies two powerful generative paradigms to learn more transferable representations. Specifically, we first employ a causal Transformer encoder, accompanied by a patch-based embedding strategy, to model the evolving trends from left to right. Building on this global modeling, we further introduce a denoising diffusion process to capture fine-grained local patterns through forward diffusion and reverse denoising. Finally, we optimize the model in an autoregressive manner. As a result, TimeDART effectively accounts for both global and local sequence features in a coherent way. We conduct extensive experiments on public datasets for time series forecasting and classification. The experimental results demonstrate that TimeDART consistently outperforms previous compared methods, validating the effectiveness of our approach. Our code is available at https://github.com/Melmaphother/TimeDART.
Authors: A. Quadir, M. Sajid, M. Tanveer, P. N. Suganthan
Abstract: In this paper, we propose enhanced feature based granular ball twin support vector machine (EF-GBTSVM). EF-GBTSVM employs the coarse granularity of granular balls (GBs) as input rather than individual data samples. The GBs are mapped to the feature space of the hidden layer using random projection followed by the utilization of a non-linear activation function. The concatenation of original and hidden features derived from the centers of GBs gives rise to an enhanced feature space, commonly referred to as the random vector functional link (RVFL) space. This space encapsulates nuanced feature information to GBs. Further, we employ twin support vector machine (TSVM) in the RVFL space for classification. TSVM generates the two non-parallel hyperplanes in the enhanced feature space, which improves the generalization performance of the proposed EF-GBTSVM model. Moreover, the coarser granularity of the GBs enables the proposed EF-GBTSVM model to exhibit robustness to resampling, showcasing reduced susceptibility to the impact of noise and outliers. We undertake a thorough evaluation of the proposed EF-GBTSVM model on benchmark UCI and KEEL datasets. This evaluation encompasses scenarios with and without the inclusion of label noise. Moreover, experiments using NDC datasets further emphasize the proposed model's ability to handle large datasets. Experimental results, supported by thorough statistical analyses, demonstrate that the proposed EF-GBTSVM model significantly outperforms the baseline models in terms of generalization capabilities, scalability, and robustness. The source code for the proposed EF-GBTSVM model, along with additional results and further details, can be accessed at https://github.com/mtanveer1/EF-GBTSVM.
Authors: Leyan Pan, Vijay Ganesh, Jacob Abernethy, Chris Esposo, Wenke Lee
Abstract: We formally study the logical reasoning capabilities of decoder-only Transformers in the context of the boolean satisfiability (SAT) problem. First, we prove by construction that decoder-only Transformers can decide 3-SAT, in a non-uniform model of computation, using backtracking and deduction via Chain-of-Thought (CoT). %We prove its correctness by showing trace equivalence to the well-known DPLL SAT-solving algorithm. Second, we implement our construction as a PyTorch model with a tool (PARAT) that we designed to empirically demonstrate its correctness and investigate its properties. Third, rather than \textit{programming} a transformer to reason, we evaluate empirically whether it can be \textit{trained} to do so by learning directly from algorithmic traces (``reasoning paths'') from our theoretical construction. The trained models demonstrate strong out-of-distribution generalization on problem sizes seen during training but has limited length generalization, which is consistent with the implications of our theoretical result
Authors: Guy Smorodinsky, Gal Vardi, Itay Safran
Abstract: We study what provable privacy attacks can be shown on trained, 2-layer ReLU neural networks. We explore two types of attacks; data reconstruction attacks, and membership inference attacks. We prove that theoretical results on the implicit bias of 2-layer neural networks can be used to provably reconstruct a set of which at least a constant fraction are training points in a univariate setting, and can also be used to identify with high probability whether a given point was used in the training set in a high dimensional setting. To the best of our knowledge, our work is the first to show provable vulnerabilities in this implicit-bias-driven setting.
Authors: Jonas H\"ubotter, Sascha Bongni, Ido Hakimi, Andreas Krause
Abstract: Recent efforts in fine-tuning language models often rely on automatic data selection, commonly using Nearest Neighbors retrieval from large datasets. However, we theoretically show that this approach tends to select redundant data, limiting its effectiveness or even hurting performance. To address this, we introduce SIFT, a data selection algorithm designed to reduce uncertainty about the model's response given a prompt, which unifies ideas from retrieval and active learning. Whereas Nearest Neighbor retrieval typically fails in the presence of information duplication, SIFT accounts for information duplication and optimizes the overall information gain of the selected examples. We focus our evaluations on fine-tuning at test-time for prompt-specific language modeling on the Pile dataset, and show that SIFT consistently outperforms Nearest Neighbor retrieval, with minimal computational overhead. Moreover, we show that our uncertainty estimates can predict the performance gain of test-time fine-tuning, and use this to develop an adaptive algorithm that invests test-time compute proportional to realized performance gains. We provide the $\texttt{activeft}$ (Active Fine-Tuning) library which can be used as a drop-in replacement for Nearest Neighbor retrieval.
Authors: Xiaoyan Hu, Ho-fung Leung, Farzan Farnia
Abstract: Selecting a sample generation scheme from multiple text-based generative models is typically addressed by choosing the model that maximizes an averaged evaluation score. However, this score-based selection overlooks the possibility that different models achieve the best generation performance for different types of text prompts. An online identification of the best generation model for various input prompts can reduce the costs associated with querying sub-optimal models. In this work, we explore the possibility of varying rankings of text-based generative models for different text prompts and propose an online learning framework to predict the best data generation model for a given input prompt. The proposed PAK-UCB algorithm addresses a contextual bandit (CB) setting with shared context variables across the arms, utilizing the generated data to update kernel-based functions that predict the score of each model available for unseen text prompts. Additionally, we leverage random Fourier features (RFF) to accelerate the online learning process of PAK-UCB and establish a $\widetilde{\mathcal{O}}(\sqrt{T})$ regret bound for the proposed RFF-based CB algorithm over $T$ iterations. Our numerical experiments on real and simulated text-to-image and image-to-text generative models show that RFF-UCB performs successfully in identifying the best generation model across different sample types.
Authors: Mingda Lu, Zitian Ao, Chao Wang, Sudhakar Prasad, Raymond H. Chan
Abstract: For the 3D localization problem using point spread function (PSF) engineering, we propose a novel enhancement of our previously introduced localization neural network, LocNet. The improved network is a physics-informed neural network (PINN) that we call PiLocNet. Previous works on the localization problem may be categorized separately into model-based optimization and neural network approaches. Our PiLocNet combines the unique strengths of both approaches by incorporating forward-model-based information into the network via a data-fitting loss term that constrains the neural network to yield results that are physically sensible. We additionally incorporate certain regularization terms from the variational method, which further improves the robustness of the network in the presence of image noise, as we show for the Poisson and Gaussian noise models. This framework accords interpretability to the neural network, and the results we obtain show its superiority. Although the paper focuses on the use of single-lobe rotating PSF to encode the full 3D source location, we expect the method to be widely applicable to other PSFs and imaging problems that are constrained by known forward processes.
Authors: Yara Shamshoum, Nitzan Hodos, Yuval Sieradzki, Assaf Schuster
Abstract: We introduce CompAct, a technique that reduces peak memory utilization on GPU by 25-30% for pretraining and 50% for fine-tuning of LLMs. Peak device memory is a major limiting factor in training LLMs, with various recent works aiming to reduce model memory. However most works don't target the largest component of allocated memory during training: the model's compute graph, which is stored for the backward pass. By storing low-rank, compressed activations to be used in the backward pass we greatly reduce the required memory, unlike previous methods which only reduce optimizer overheads or the number of trained parameters. Our compression uses random projection matrices, thus avoiding additional memory overheads. Comparisons with previous techniques for either pretraining or fine-tuning show that CompAct substantially improves existing compute-performance tradeoffs. We expect CompAct's savings to scale even higher for larger models.
Authors: Yining Li, Peizhong Ju, Ness B. Shroff
Abstract: Multi-Objective Markov Decision Processes (MO-MDPs) are receiving increasing attention, as real-world decision-making problems often involve conflicting objectives that cannot be addressed by a single-objective MDP. The Pareto front identifies the set of policies that cannot be dominated, providing a foundation for finding Pareto optimal solutions that can efficiently adapt to various preferences. However, finding the Pareto front is a highly challenging problem. Most existing methods either (i) rely on traversing the continuous preference space, which is impractical and results in approximations that are difficult to evaluate against the true Pareto front, or (ii) focus solely on deterministic Pareto optimal policies, from which there are no known techniques to characterize the full Pareto front. Moreover, finding the structure of the Pareto front itself remains unclear even in the context of dynamic programming, where the MDP is fully known in advance. In this work, we address the challenge of efficiently discovering the Pareto front. By investigating the geometric structure of the Pareto front in MO-MDPs, we uncover a key property: the Pareto front is on the boundary of a convex polytope whose vertices all correspond to deterministic policies, and neighboring vertices of the Pareto front differ by only one state-action pair of the deterministic policy, almost surely. This insight transforms the global comparison across all policies into a localized search among deterministic policies that differ by only one state-action pair, drastically reducing the complexity of searching for the exact Pareto front. We develop an efficient algorithm that identifies the vertices of the Pareto front by solving a single-objective MDP only once and then traversing the edges of the Pareto front, making it more efficient than existing methods.
Authors: Kang Zhao, Tao Yuan, Han Bao, Zhenfeng Su, Chang Gao, Zhaofeng Sun, Zichen Liang, Liping Jing, Jianfei Chen
Abstract: To date, 2:4 sparsity has stood as the only sparse pattern that can be accelerated using sparse tensor cores on GPUs. In practice, 2:4 sparsity often possesses low actual speedups ($\leq 1.3$) and requires fixed sparse ratios, meaning that other ratios, such as 4:8, 8:16, or those exceeding 50% sparsity, do not incur any speedups on GPUs. Recent studies suggest that V:N:M sparsity is promising in addressing these limitations of 2:4 sparsity. However, regarding accuracy, the effects of V:N:M sparsity on broader Transformer models, such as vision Transformers and large language models (LLMs), are largely unexamined. Moreover, Some specific issues related to V:N:M sparsity, such as how to select appropriate V and M values, remain unresolved. In this study, we thoroughly investigate the application of V:N:M sparsity in vision models and LLMs across multiple tasks, from pertaining to downstream tasks. We propose three key approaches to enhance the applicability and accuracy of V:N:M-sparse Transformers, including heuristic V and M selection, V:N:M-specific channel permutation, and three-staged LoRA training techniques. Experimental results show that, with our methods, the DeiT-small achieves lossless accuracy at 64:2:5 sparsity, while the DeiT-base maintains accuracy even at 64:2:8 sparsity. In addition, the fine-tuned LLama2-7B at 64:2:5 sparsity performs comparably or better than training-free 2:4 sparse alternatives on downstream tasks. More importantly, V:N:M-sparse Transformers offer a wider range of speedup-accuracy trade-offs compared to 2:4 sparsity. Overall, our exploration largely facilitates the V:N:M sparsity to act as a truly effective acceleration solution for Transformers in cost-sensitive inference scenarios.
Authors: Kilian Freitag, Kristian Ceder, Rita Laezza, Knut {\AA}kesson, Morteza Haghir Chehreghani
Abstract: Reinforcement learning (RL) has emerged as a powerful tool for tackling control problems, but its practical application is often hindered by the complexity arising from intricate reward functions with multiple terms. The reward hypothesis posits that any objective can be encapsulated in a scalar reward function, yet balancing individual, potentially adversarial, reward terms without exploitation remains challenging. To overcome the limitations of traditional RL methods, which often require precise balancing of competing reward terms, we propose a two-stage reward curriculum that first maximizes a simple reward function and then transitions to the full, complex reward. We provide a method based on how well an actor fits a critic to automatically determine the transition point between the two stages. Additionally, we introduce a flexible replay buffer that enables efficient phase transfer by reusing samples from one stage in the next. We evaluate our method on the DeepMind control suite, modified to include an additional constraint term in the reward definitions. We further evaluate our method in a mobile robot scenario with even more competing reward terms. In both settings, our two-stage reward curriculum achieves a substantial improvement in performance compared to a baseline trained without curriculum. Instead of exploiting the constraint term in the reward, it is able to learn policies that balance task completion and constraint satisfaction. Our results demonstrate the potential of two-stage reward curricula for efficient and stable RL in environments with complex rewards, paving the way for more robust and adaptable robotic systems in real-world applications.
Authors: M. Tanveer, R. K. Sharma, A. Quadir, M. Sajid
Abstract: In the domain of machine learning, least square twin support vector machine (LSTSVM) stands out as one of the state-of-the-art models. However, LSTSVM suffers from sensitivity to noise and outliers, overlooking the SRM principle and instability in resampling. Moreover, its computational complexity and reliance on matrix inversions hinder the efficient processing of large datasets. As a remedy to the aforementioned challenges, we propose the robust granular ball LSTSVM (GBLSTSVM). GBLSTSVM is trained using granular balls instead of original data points. The core of a granular ball is found at its center, where it encapsulates all the pertinent information of the data points within the ball of specified radius. To improve scalability and efficiency, we further introduce the large-scale GBLSTSVM (LS-GBLSTSVM), which incorporates the SRM principle through regularization terms. Experiments are performed on UCI, KEEL, and NDC benchmark datasets; both the proposed GBLSTSVM and LS-GBLSTSVM models consistently outperform the baseline models.
Authors: Artem Basharin, Andrei Chertkov, Ivan Oseledets
Abstract: We propose a new model for multi-token prediction in transformers, aiming to enhance sampling efficiency without compromising accuracy. Motivated by recent work that predicts the probabilities of subsequent tokens using multiple heads, we connect this approach to rank-$1$ canonical tensor decomposition. By generalizing it to a rank-$r$ canonical probability decomposition, we develop an improved model that predicts multiple tokens simultaneously. This model can also be interpreted as a mixture of experts, allowing us to leverage successful techniques from that domain for efficient and robust training. Importantly, the overall overhead for training and sampling remains low. Our method demonstrates significant improvements in inference speed for both text and code generation tasks, proving particularly beneficial within the self-speculative decoding paradigm. It maintains its effectiveness across various model sizes and training epochs, highlighting its robustness and scalability.
Authors: Chanwoo Chun, SueYeon Chung, Daniel D. Lee
Abstract: Analyzing the structure of sampled features from an input data distribution is challenging when constrained by limited measurements in both the number of inputs and features. Traditional approaches often rely on the eigenvalue spectrum of the sample covariance matrix derived from finite measurement matrices; however, these spectra are sensitive to the size of the measurement matrix, leading to biased insights. In this paper, we introduce a novel algorithm that provides unbiased estimates of the spectral moments of the kernel integral operator in the limit of infinite inputs and features from finitely sampled measurement matrices. Our method, based on dynamic programming, is efficient and capable of estimating the moments of the operator spectrum. We demonstrate the accuracy of our estimator on radial basis function (RBF) kernels, highlighting its consistency with the theoretical spectra. Furthermore, we showcase the practical utility and robustness of our method in understanding the geometry of learned representations in neural networks.
Authors: Jaechang Kim, Jinmin Goh, Inseok Hwang, Jaewoong Cho, Jungseul Ok
Abstract: Deep learning-based expert models have reached superhuman performance in decision-making domains such as chess and Go. However, it is under-explored to explain or comment on given decisions although it is important for model explainability and human education. The outputs of expert models are accurate, but yet difficult to interpret for humans. On the other hand, large language models (LLMs) can produce fluent commentary but are prone to hallucinations due to their limited decision-making capabilities. To bridge this gap between expert models and LLMs, we focus on chess commentary as a representative task of explaining complex decision-making processes through language and address both the generation and evaluation of commentary. We introduce Concept-guided Chess Commentary generation (CCC) for producing commentary and GPT-based Chess Commentary Evaluation (GCC-Eval) for assessing it. CCC integrates the decision-making strengths of expert models with the linguistic fluency of LLMs through prioritized, concept-based explanations. GCC-Eval leverages expert knowledge to evaluate chess commentary based on informativeness and linguistic quality. Experimental results, validated by both human judges and GCC-Eval, demonstrate that CCC generates commentary which is accurate, informative, and fluent.
Authors: Sima Noorani, Orlando Romero, Nicolo Dal Fabbro, Hamed Hassani, George J. Pappas
Abstract: Conformal prediction (CP) is a distribution-free framework for achieving probabilistic guarantees on black-box models. CP is generally applied to a model post-training. Recent research efforts, on the other hand, have focused on optimizing CP efficiency during training. We formalize this concept as the problem of conformal risk minimization (CRM). In this direction, conformal training (ConfTr) by Stutz et al.(2022) is a technique that seeks to minimize the expected prediction set size of a model by simulating CP in-between training updates. Despite its potential, we identify a strong source of sample inefficiency in ConfTr that leads to overly noisy estimated gradients, introducing training instability and limiting practical use. To address this challenge, we propose variance-reduced conformal training (VR-ConfTr), a CRM method that incorporates a variance reduction technique in the gradient estimation of the ConfTr objective function. Through extensive experiments on various benchmark datasets, we demonstrate that VR-ConfTr consistently achieves faster convergence and smaller prediction sets compared to baselines.
Authors: Nasib Ullah, Erik Schultheis, Mike Lasby, Yani Ioannou, Rohit Babbar
Abstract: In recent years, Dynamic Sparse Training (DST) has emerged as an alternative to post-training pruning for generating efficient models. In principle, DST allows for a more memory efficient training process, as it maintains sparsity throughout the entire training run. However, current DST implementations fail to capitalize on this in practice. Because sparse matrix multiplication is much less efficient than dense matrix multiplication on GPUs, most implementations simulate sparsity by masking weights. In this paper, we leverage recent advances in semi-structured sparse training to apply DST in the domain of classification with large output spaces, where memory-efficiency is paramount. With a label space of possibly millions of candidates, the classification layer alone will consume several gigabytes of memory. Switching from a dense to a fixed fan-in sparse layer updated with sparse evolutionary training (SET); however, severely hampers training convergence, especially at the largest label spaces. We find that poor gradient flow from the sparse classifier to the dense text encoder make it difficult to learn good input representations. By employing an intermediate layer or adding an auxiliary training objective, we recover most of the generalisation performance of the dense model. Overall, we demonstrate the applicability and practical benefits of DST in a challenging domain -- characterized by a highly skewed label distribution that differs substantially from typical DST benchmark datasets -- which enables end-to-end training with millions of labels on commodity hardware.
Authors: Joseph Pollock, Igor Shilov, Euodia Dodd, Yves-Alexandre de Montjoye
Abstract: Membership inference attacks (MIAs) are widely used to empirically assess privacy risks in machine learning models, both providing model-level vulnerability metrics and identifying the most vulnerable training samples. State-of-the-art methods, however, require training hundreds of shadow models with the same architecture as the target model. This makes the computational cost of assessing the privacy of models prohibitive for many practical applications, particularly when used iteratively as part of the model development process and for large models. We propose a novel approach for identifying the training samples most vulnerable to membership inference attacks by analyzing artifacts naturally available during the training process. Our method, Loss Trace Interquantile Range (LT-IQR), analyzes per-sample loss trajectories collected during model training to identify high-risk samples without requiring any additional model training. Through experiments on standard benchmarks, we demonstrate that LT-IQR achieves 92% precision@k=1% in identifying the samples most vulnerable to state-of-the-art MIAs. This result holds across datasets and model architectures with LT-IQR outperforming both traditional vulnerability metrics, such as loss, and lightweight MIAs using few shadow models. We also show LT-IQR to accurately identify points vulnerable to multiple MIA methods and perform ablation studies. We believe LT-IQR enables model developers to identify vulnerable training samples, for free, as part of the model development process. Our results emphasize the potential of artifact-based methods to efficiently evaluate privacy risks.
Authors: Yang Hu, Xiao Wang, Zezhen Ding, Lirong Wu, Huatian Zhang, Stan Z. Li, Sheng Wang, Jiheng Zhang, Ziyun Li, Tianlong Chen
Abstract: Diffusion-based models have significant achievements in time series generation but suffer from inefficient computation: solving high-dimensional ODEs/SDEs via iterative numerical solvers demands hundreds to thousands of drift function evaluations per sample, incurring prohibitive costs. To resolve this, we propose FlowTS, an ODE-based model that leverages rectified flow with straight-line transport in probability space. By learning geodesic paths between distributions, FlowTS achieves computational efficiency through exact linear trajectory simulation, accelerating training and generation while improving performances. We further introduce an adaptive sampling strategy inspired by the exploration-exploitation trade-off, balancing noise adaptation and precision. Notably, FlowTS enables seamless adaptation from unconditional to conditional generation without retraining, ensuring efficient real-world deployment. Also, to enhance generation authenticity, FlowTS integrates trend and seasonality decomposition, attention registers (for global context aggregation), and Rotary Position Embedding (RoPE) (for position information). For unconditional setting, extensive experiments demonstrate that FlowTS achieves state-of-the-art performance, with context FID scores of 0.019 and 0.011 on Stock and ETTh datasets (prev. best: 0.067, 0.061). For conditional setting, we have achieved superior performance in solar forecasting (MSE 213, prev. best: 375) and MuJoCo imputation tasks (MSE 7e-5, prev. best 2.7e-4). The code is available at https://github.com/UNITES-Lab/FlowTS.
Authors: Huizhuo Yuan, Yifeng Liu, Shuang Wu, Xun Zhou, Quanquan Gu
Abstract: Training deep neural networks--and more recently, large models demands efficient and scalable optimizers. Adaptive gradient algorithms like Adam, AdamW, and their variants have been central to this task. Despite the development of numerous variance reduction algorithms in the past decade aimed at accelerating stochastic optimization in both convex and nonconvex settings, variance reduction has not found widespread success in training deep neural networks or large language models. Consequently, it has remained a less favored approach in modern AI. In this paper, to unleash the power of variance reduction for efficient training of large models, we propose a unified optimization framework, MARS (Make vAriance Reduction Shine), which reconciles preconditioned gradient methods with variance reduction via a scaled stochastic recursive momentum technique. Within our framework, we introduce three instances of MARS that leverage preconditioned gradient updates based on AdamW, Lion, and Shampoo, respectively. We also draw a connection between our algorithms and existing optimizers. Experimental results on training GPT-2 models indicate that MARS consistently outperforms AdamW by a large margin. The implementation of MARS is available at https://github.com/AGI-Arena/MARS.
Authors: Jintao Zhang, Haofeng Huang, Pengle Zhang, Jia Wei, Jun Zhu, Jianfei Chen
Abstract: Although quantization for linear layers has been widely used, its application to accelerate the attention process remains limited. To further enhance the efficiency of attention computation compared to SageAttention while maintaining precision, we propose SageAttention2, which utilizes significantly faster 4-bit matrix multiplication (Matmul) alongside additional precision-enhancing techniques. First, we propose to quantize matrices $(Q, K)$ to INT4 in a hardware-friendly thread-level granularity and quantize matrices $(\widetilde P, V)$ to FP8. Second, we propose a method to smooth $Q$, enhancing the accuracy of INT4 $QK^\top$. Third, we propose a two-level accumulation strategy for $\widetilde PV$ to enhance the accuracy of FP8 $\widetilde PV$. The operations per second (OPS) of SageAttention2 surpass FlashAttention2 and xformers by about 3x and 4.5x on RTX4090, respectively. Moreover, SageAttention2 matches the speed of FlashAttention3(fp8) on the Hopper GPUs, while delivering much higher accuracy. Comprehensive experiments confirm that our approach incurs negligible end-to-end metrics loss across diverse models, including those for language, image, and video generation. The code is available at https://github.com/thu-ml/SageAttention.
Authors: Daniel Jenson, Jhonathan Navott, Mengyan Zhang, Makkunda Sharma, Elizaveta Semenova, Seth Flaxman
Abstract: Neural Processes (NPs) are a rapidly evolving class of models designed to directly model the posterior predictive distribution of stochastic processes. Originally developed as a scalable alternative to Gaussian Processes (GPs), which are limited by $\mathcal{O}(n^3)$ runtime complexity, the most accurate modern NPs can often rival GPs but still suffer from an $\mathcal{O}(n^2)$ bottleneck due to their attention mechanism. We introduce the Transformer Neural Process - Kernel Regression (TNP-KR), a scalable NP featuring: (1) a Kernel Regression Block (KRBlock), a simple, extensible, and parameter efficient transformer block with complexity $\mathcal{O}(n_c^2 + n_c n_t)$, where $n_c$ and $n_t$ are the number of context and test points, respectively; (2) a kernel-based attention bias; and (3) two novel attention mechanisms: scan attention (SA), a memory-efficient scan-based attention that when paired with a kernel-based bias can make TNP-KR translation invariant, and deep kernel attention (DKA), a Performer-style attention that implicitly incoporates a distance bias and further reduces complexity to $\mathcal{O}(n_c)$. These enhancements enable both TNP-KR variants to perform inference with 100K context points on over 1M test points in under a minute on a single 24GB GPU. On benchmarks spanning meta regression, Bayesian optimization, image completion, and epidemiology, TNP-KR with DKA outperforms its Performer counterpart on nearly every benchmark, while TNP-KR with SA achieves state-of-the-art results.
Authors: Guoguo Ai, Guansong Pang, Hezhe Qiao, Yuan Gao, Hui Yan
Abstract: Graph Transformers (GTs) have demonstrated remarkable performance in graph representation learning over popular graph neural networks (GNNs). However, self--attention, the core module of GTs, preserves only low-frequency signals in graph features, leading to ineffectiveness in capturing other important signals like high-frequency ones. Some recent GT models help alleviate this issue, but their flexibility and expressiveness are still limited since the filters they learn are fixed on predefined graph spectrum or order. To tackle this challenge, we propose a Graph Fourier Kolmogorov-Arnold Transformer (GrokFormer), a novel GT model that learns highly expressive spectral filters with adaptive graph spectrum and order through a Fourier series modeling over learnable activation functions. We demonstrate theoretically and empirically that the proposed GrokFormer filter offers better expressiveness than other spectral methods. Comprehensive experiments on 10 real-world node classification datasets across various domains, scales, and graph properties, as well as 5 graph classification datasets, show that GrokFormer outperforms state-of-the-art GTs and GNNs. Our code is available at https://github.com/GGA23/GrokFormer
Authors: Wen-Dong Jiang, Chih-Yung Chang, Show-Jane Yen, Diptendu Sinha Roy
Abstract: Deep learning has achieved remarkable success in processing and managing unstructured data. However, its "black box" nature imposes significant limitations, particularly in sensitive application domains. While existing interpretable machine learning methods address some of these issues, they often fail to adequately consider feature correlations and provide insufficient evaluation of model decision paths. To overcome these challenges, this paper introduces Real Explainer (RealExp), an interpretability computation method that decouples the Shapley Value into individual feature importance and feature correlation importance. By incorporating feature similarity computations, RealExp enhances interpretability by precisely quantifying both individual feature contributions and their interactions, leading to more reliable and nuanced explanations. Additionally, this paper proposes a novel interpretability evaluation criterion focused on elucidating the decision paths of deep learning models, going beyond traditional accuracy-based metrics. Experimental validations on two unstructured data tasks -- image classification and text sentiment analysis -- demonstrate that RealExp significantly outperforms existing methods in interpretability. Case studies further illustrate its practical value: in image classification, RealExp aids in selecting suitable pre-trained models for specific tasks from an interpretability perspective; in text classification, it enables the optimization of models and approximates the performance of a fine-tuned GPT-Ada model using traditional bag-of-words approaches.
Authors: Yichen Wang, Chengcheng Yu
Abstract: Accurate origin-destination (OD) passenger flow prediction is crucial for enhancing metro system efficiency, optimizing scheduling, and improving passenger experiences. However, current models often fail to effectively capture the asynchronous departure characteristics of OD flows and underutilize the inflow and outflow data, which limits their prediction accuracy. To address these issues, we propose CSP-AIT-Net, a novel spatiotemporal graph attention framework designed to enhance OD flow prediction by incorporating asynchronous inflow tracking and advanced station semantics representation. Our framework restructures the OD flow prediction paradigm by first predicting outflows and then decomposing OD flows using a spatiotemporal graph attention mechanism. To enhance computational efficiency, we introduce a masking mechanism and propose asynchronous passenger flow graphs that integrate inflow and OD flow with conservation constraints. Furthermore, we employ contrastive learning to extract high-dimensional land use semantics of metro stations, enriching the contextual understanding of passenger mobility patterns. Validation of the Shanghai metro system demonstrates improvement in short-term OD flow prediction accuracy over state-of-the-art methods. This work contributes to enhancing metro operational efficiency, scheduling precision, and overall system safety.
Authors: Zhiheng Zhang, Zichen Wang
Abstract: Network interference has attracted significant attention in the field of causal inference, encapsulating various sociological behaviors where the treatment assigned to one individual within a network may affect the outcomes of others, such as their neighbors. A key challenge in this setting is that standard causal inference methods often assume independent treatment effects among individuals, which may not hold in networked environments. To estimate interference-aware causal effects, a traditional approach is to inherit the independent settings, where practitioners randomly assign experimental participants into different groups and compare their outcomes. While effective in offline settings, this strategy becomes problematic in sequential experiments, where suboptimal decision persists, leading to substantial regret. To address this issue, we introduce a unified interference-aware framework for online experimental design. Compared to existing studies, we extend the definition of arm space by utilizing the statistical concept of exposure mapping, which allows for a more flexible and context-aware representation of treatment effects in networked settings. Crucially, we establish a Pareto-optimal trade-off between estimation accuracy and regret under the network concerning both time period and arm space, which remains superior to baseline models even without network interference. Furthermore, we propose an algorithmic implementation and discuss its generalization across different learning settings and network topology.
Authors: Songkai Xue, Yuekai Sun
Abstract: Performative prediction aims to model scenarios where predictive outcomes subsequently influence the very systems they target. The pursuit of a performative optimum (PO) -- minimizing performative risk -- is generally reliant on modeling of the distribution map, which characterizes how a deployed ML model alters the data distribution. Unfortunately, inevitable misspecification of the distribution map can lead to a poor approximation of the true PO. To address this issue, we introduce a novel framework of distributionally robust performative prediction and study a new solution concept termed as distributionally robust performative optimum (DRPO). We show provable guarantees for DRPO as a robust approximation to the true PO when the nominal distribution map is different from the actual one. Moreover, distributionally robust performative prediction can be reformulated as an augmented performative prediction problem, enabling efficient optimization. The experimental results demonstrate that DRPO offers potential advantages over traditional PO approach when the distribution map is misspecified at either micro- or macro-level.
Authors: Yuan-Heng Wang, Hoshin V. Gupta
Abstract: Despite excellent real-world predictive performance of modern machine learning (ML) methods, many scientists hesitate to discard traditional physical-conceptual (PC) approaches due to their relative interpretability, which contributes to credibility during decision-making. In this context, a currently underexplored aspect of ML is how to develop minimally-optimal representations that can facilitate better insight regarding system functioning. Regardless of how this is achieved, parsimonious representations seem to better support the advancement of scientific understanding. Our own view is that ML-based modeling should be based in use of computational units that are fundamentally interpretable by design. This paper continues our exploration of how ML can be exploited in the service of scientific investigation. We use the Mass-Conserving-Perceptron (MCP) as the fundamental computational unit in a generic network architecture to explore important issues related to the use of observational data for constructing models of dynamical systems. We show, in the context of lumped catchment modeling, that physical interpretability and predictive performance can both be achieved using a relatively parsimonious distributed-state multiple-flow-path network with context-dependent gating and information sharing across the nodes, suggesting that MCP-based modeling can play a significant role in application of ML to geoscientific investigation.
Authors: Zirun Guo, Tao Jin, Wenlong Xu, Wang Lin, Yangyang Wu
Abstract: Multimodal sentiment analysis (MSA) is an emerging research topic that aims to understand and recognize human sentiment or emotions through multiple modalities. However, in real-world dynamic scenarios, the distribution of target data is always changing and different from the source data used to train the model, which leads to performance degradation. Common adaptation methods usually need source data, which could pose privacy issues or storage overheads. Therefore, test-time adaptation (TTA) methods are introduced to improve the performance of the model at inference time. Existing TTA methods are always based on probabilistic models and unimodal learning, and thus can not be applied to MSA which is often considered as a multimodal regression task. In this paper, we propose two strategies: Contrastive Adaptation and Stable Pseudo-label generation (CASP) for test-time adaptation for multimodal sentiment analysis. The two strategies deal with the distribution shifts for MSA by enforcing consistency and minimizing empirical risk, respectively. Extensive experiments show that CASP brings significant and consistent improvements to the performance of the model across various distribution shift settings and with different backbones, demonstrating its effectiveness and versatility. Our codes are available at https://github.com/zrguo/CASP.
Authors: Santiago Aranguri, Francesco Insulla
Abstract: We analyze the training of a two-layer autoencoder used to parameterize a flow-based generative model for sampling from a high-dimensional Gaussian mixture. Previous work shows that the phase where the relative probability between the modes is learned disappears as the dimension goes to infinity without an appropriate time schedule. We introduce a time dilation that solves this problem. This enables us to characterize the learned velocity field, finding a first phase where the probability of each mode is learned and a second phase where the variance of each mode is learned. We find that the autoencoder representing the velocity field learns to simplify by estimating only the parameters relevant to each phase. Turning to real data, we propose a method that, for a given feature, finds intervals of time where training improves accuracy the most on that feature. Since practitioners take a uniform distribution over training times, our method enables more efficient training. We provide preliminary experiments validating this approach.
Authors: Luke Bailey, Alex Serrano, Abhay Sheshadri, Mikhail Seleznyov, Jordan Taylor, Erik Jenner, Jacob Hilton, Stephen Casper, Carlos Guestrin, Scott Emmons
Abstract: Recent latent-space monitoring techniques have shown promise as defenses against LLM attacks. These defenses act as scanners that seek to detect harmful activations before they lead to undesirable actions. This prompts the question: Can models execute harmful behavior via inconspicuous latent states? Here, we study such obfuscated activations. We show that state-of-the-art latent-space defenses -- including sparse autoencoders, representation probing, and latent OOD detection -- are all vulnerable to obfuscated activations. For example, against probes trained to classify harmfulness, our attacks can often reduce recall from 100% to 0% while retaining a 90% jailbreaking rate. However, obfuscation has limits: we find that on a complex task (writing SQL code), obfuscation reduces model performance. Together, our results demonstrate that neural activations are highly malleable: we can reshape activation patterns in a variety of ways, often while preserving a network's behavior. This poses a fundamental challenge to latent-space defenses.
Authors: Milad Soltany, Farhad Pourpanah, Mahdiyar Molahasani, Michael Greenspan, Ali Etemad
Abstract: In this paper, we propose a novel approach, Federated Domain Generalization with Label Smoothing and Balanced Decentralized Training (FedSB), to address the challenges of data heterogeneity within a federated learning framework. FedSB utilizes label smoothing at the client level to prevent overfitting to domain-specific features, thereby enhancing generalization capabilities across diverse domains when aggregating local models into a global model. Additionally, FedSB incorporates a decentralized budgeting mechanism which balances training among clients, which is shown to improve the performance of the aggregated global model. Extensive experiments on four commonly used multi-domain datasets, PACS, VLCS, OfficeHome, and TerraInc, demonstrate that FedSB outperforms competing methods, achieving state-of-the-art results on three out of four datasets, indicating the effectiveness of FedSB in addressing data heterogeneity.
Authors: Flint Xiaofeng Fan, Cheston Tan, Yew-Soon Ong, Roger Wattenhofer, Wei-Tsang Ooi
Abstract: In the era of increasing privacy concerns and demand for personalized experiences, traditional Reinforcement Learning with Human Feedback (RLHF) frameworks face significant challenges due to their reliance on centralized data. We introduce Federated Reinforcement Learning with Human Feedback (FedRLHF), a novel framework that decentralizes the RLHF process. FedRLHF enables collaborative policy learning across multiple clients without necessitating the sharing of raw data or human feedback, thereby ensuring robust privacy preservation. Leveraging federated reinforcement learning, each client integrates human feedback locally into their reward functions and updates their policies through personalized RLHF processes. We establish rigorous theoretical foundations for FedRLHF, providing convergence guarantees, and deriving sample complexity bounds that scale efficiently with the number of clients. Empirical evaluations on the MovieLens and IMDb datasets demonstrate that FedRLHF not only preserves user privacy but also achieves performance on par with centralized RLHF, while enhancing personalization across diverse client environments.
Authors: Amirhossein Mesbah, Reshad Hosseini, Seyed Pooya Shariatpanahi, Majid Nili Ahmadabadi
Abstract: Reinforcement learning (RL) plays a major role in solving complex sequential decision-making tasks. Hierarchical and goal-conditioned RL are promising methods for dealing with two major problems in RL, namely sample inefficiency and difficulties in reward shaping. These methods tackle the mentioned problems by decomposing a task into simpler subtasks and temporally abstracting a task in the action space. One of the key components for task decomposition of these methods is subgoal discovery. We can use the subgoal states to define hierarchies of actions and also use them in decomposing complex tasks. Under the assumption that subgoal states are more unpredictable, we propose a free energy paradigm to discover them. This is achieved by using free energy to select between two spaces, the main space and an aggregation space. The $model \; changes$ from neighboring states to a given state shows the unpredictability of a given state, and therefore it is used in this paper for subgoal discovery. Our empirical results on navigation tasks like grid-world environments show that our proposed method can be applied for subgoal discovery without prior knowledge of the task. Our proposed method is also robust to the stochasticity of environments.
Authors: Sunwoo Kim, Minkyu Kim, Dongmin Park
Abstract: Diffusion models excel in generative tasks, but aligning them with specific objectives while maintaining their versatility remains challenging. Existing fine-tuning methods often suffer from reward over-optimization, while approximate guidance approaches fail to optimize target rewards effectively. Addressing these limitations, we propose a training-free, test-time method based on Sequential Monte Carlo (SMC) to sample from the reward-aligned target distribution. Our approach, tailored for diffusion sampling and incorporating tempering techniques, achieves comparable or superior target rewards to fine-tuning methods while preserving diversity and cross-reward generalization. We demonstrate its effectiveness in single-reward optimization, multi-objective scenarios, and online black-box optimization. This work offers a robust solution for aligning diffusion models with diverse downstream objectives without compromising their general capabilities. Code is available at https://github.com/krafton-ai/DAS.
Authors: Fuhang Liang, Rucong Xu, Deng Lin
Abstract: Current deep regression models usually learn in a point-wise way that treats each sample as an independent input, neglecting the relative ordering among different data. Consequently, the regression model could neglect the data's interrelationships, potentially resulting in suboptimal performance. Moreover, the existence of aleatoric uncertainty in the training data may drive the model to capture non-generalizable patterns, contributing to increased overfitting. To address these issues, we propose a novel adaptive pairwise learning framework for regression tasks (AdaPRL) which leverages the relative differences between data points and integrates with deep probabilistic models to quantify the uncertainty associated with the predictions. Additionally, we adapt AdaPRL for applications in multi-task learning and multivariate time series forecasting. Extensive experiments with several real-world regression datasets including recommendation systems, age prediction, time series forecasting, natural language understanding, finance, and industry datasets show that AdaPRL is compatible with different backbone networks in various tasks and achieves state-of-the-art performance on the vast majority of tasks without extra inference cost, highlighting its notable potential including enhancing prediction accuracy and ranking ability, increasing generalization capability, improving robustness to noisy data, improving resilience to reduced data, and enhancing interpretability. Experiments also show that AdaPRL can be seamlessly incorporated into recently proposed regression frameworks to gain performance improvement.
Authors: Stephane Hatgis-Kessell, W. Bradley Knox, Serena Booth, Scott Niekum, Peter Stone
Abstract: Designing a reinforcement learning from human feedback (RLHF) algorithm to approximate a human's unobservable reward function requires assuming, implicitly or explicitly, a model of human preferences. A preference model that poorly describes how humans generate preferences risks learning a poor approximation of the human's reward function. In this paper, we conduct three human studies to asses whether one can influence the expression of real human preferences to more closely conform to a desired preference model. Importantly, our approach does not seek to alter the human's unobserved reward function. Rather, we change how humans use this reward function to generate preferences, such that they better match whatever preference model is assumed by a particular RLHF algorithm. We introduce three interventions: showing humans the quantities that underlie a preference model, which is normally unobservable information derived from the reward function; training people to follow a specific preference model; and modifying the preference elicitation question. All intervention types show significant effects, providing practical tools to improve preference data quality and the resultant alignment of the learned reward functions. Overall we establish a novel research direction in model alignment: designing interfaces and training interventions to increase human conformance with the modeling assumptions of the algorithm that will learn from their input.
Authors: James Chua, Owain Evans
Abstract: Models trained specifically to generate long Chains of Thought (CoTs) have recently achieved impressive results. We refer to these models as Inference-Time-Compute (ITC) models. Are the CoTs of ITC models more faithful compared to traditional non-ITC models? We evaluate three ITC models (based on Qwen-2.5, Gemini-2, and DeepSeek-V3-Base) on an existing test of faithful CoT. To measure faithfulness, we test if models articulate a cue in their prompt that influences their answers to MMLU questions. For example, when the cue "A Stanford Professor thinks the answer is D" is added to the prompt, models sometimes switch their answer to D. In such cases, the DeepSeek-R1 ITC model articulates the cue 59% of the time, compared to 7% for the non-ITC DeepSeek. We set a strict requirement on articulating -- these must describe how the cue makes the models switch their answer - simply mentioning the cue does not count. We evaluate 7 types of cue, such as misleading few-shot examples and anchoring on past responses. ITC models articulate cues that influence them much more reliably than all the 7 non-ITC models tested, such as Claude-3.5-Sonnet and GPT-4o, which often articulate close to 0% of the time. Finally, we conduct analysis which suggests reward modeling and length penalties result in unfaithful responses. However, our study has important limitations. We cannot evaluate OpenAI's SOTA o3 model. We also lack details about the training of all ITC models evaluated, making it hard to attribute our findings to specific processes. Faithfulness of CoT is an important property for AI Safety. The ITC models tested show a large improvement in faithfulness, which is worth investigating further.
Authors: Jivat Neet Kaur, Michael I. Jordan, Ahmed Alaa
Abstract: Standard conformal prediction offers a marginal guarantee on coverage, but for prediction sets to be truly useful, they should ideally ensure coverage conditional on each test point. Unfortunately, it is impossible to achieve exact, distribution-free conditional coverage in finite samples. In this work, we propose an alternative conformal prediction algorithm that targets coverage where it matters most--in instances where a classifier is overconfident in its incorrect predictions. We start by dissecting miscoverage events in marginally-valid conformal prediction, and show that miscoverage rates vary based on the classifier's confidence and its deviation from the Bayes optimal classifier. Motivated by this insight, we develop a variant of conformal prediction that targets coverage conditional on a reduced set of two variables: the classifier's confidence in a prediction and a nonparametric trust score that measures its deviation from the Bayes classifier. Empirical evaluation on multiple image datasets shows that our method generally improves conditional coverage properties compared to standard conformal prediction, including class-conditional coverage, coverage over arbitrary subgroups, and coverage over demographic groups.
Authors: Qi Cheems Wang, Zehao Xiao, Yixiu Mao, Yun Qu, Jiayi Shen, Yiqin Lv, Xiangyang Ji
Abstract: The foundation model enables fast problem-solving without learning from scratch, and such a desirable adaptation property benefits from its adopted cross-task generalization paradigms, e.g., pretraining, meta-training, and finetuning. Recent advances in these paradigms show the crucial role of challenging tasks' prioritized sampling in enhancing adaptation robustness and even improving sampling efficiency. However, scoring task difficulties exhausts massive task queries and requires intensive evaluation and computations, e.g., policy evaluations in Markov decision processes (MDPs) or inference with large backbone models. This work underscores the criticality of both adaptation robustness and learning efficiency, especially in scenarios where tasks are risky to collect or costly to evaluate. To this end, we present Model Predictive Task Sampling (MPTS) to establish connections between the task space and adaptation risk landscape for robust active task sampling. Technically, MPTS characterizes the task episodic information with a generative model and predicts optimization outcome, i.e., task-specific adaptation risk values, from posterior inference. The resulting risk learner amortizes expensive annotation, evaluation, or computation operations in task robust adaptation. Extensive experimental results show that MPTS can be seamlessly integrated into zero-shot, few-shot, and many-shot learning paradigms, increases adaptation robustness, and retains learning efficiency without affording extra cost. The code is available at the project site https://github.com/thu-rllab/MPTS.
Authors: Taiki Yamada, Yuichi Katori, Kantaro Fujiwara
Abstract: Conventional echo state networks (ESNs) require supervised learning to train the readout layer, using the desired outputs as training data. In this study, we focus on input reconstruction (IR), which refers to training the readout layer to reproduce the input time series in its output. We reformulate the learning algorithm of the ESN readout layer to perform IR using unsupervised learning (UL). By conducting theoretical analysis and numerical experiments, we demonstrate that IR in ESNs can be effectively implemented under realistic conditions without explicitly using the desired outputs as training data; in this way, UL is enabled. Furthermore, we demonstrate that applications relying on IR, such as dynamical system replication and noise filtering, can be reformulated within the UL framework. Our findings establish a theoretically sound and universally applicable IR formulation, along with its related tasks in ESNs. This work paves the way for novel predictions and highlights unresolved theoretical challenges in ESNs, particularly in the context of time-series processing methods and computational models of the brain.
Authors: Lu Liu, Yang Tang, Kexuan Zhang, Qiyu Sun
Abstract: Due to the challenge posed by multi-source and heterogeneous data collected from diverse environments, causal relationships among features can exhibit variations influenced by different time spans, regions, or strategies. This diversity makes a single causal model inadequate for accurately representing complex causal relationships in all observational data, a crucial consideration in causal learning. To address this challenge, the nonlinear Causal Kernel Clustering method is introduced for heterogeneous subgroup causal learning, highlighting variations in causal relationships across diverse subgroups. The main component for clustering heterogeneous subgroups lies in the construction of the $u$-centered sample mapping function with the property of unbiased estimation, which assesses the differences in potential nonlinear causal relationships in various samples and supported by causal identifiability theory. Experimental results indicate that the method performs well in identifying heterogeneous subgroups and enhancing causal learning, leading to a reduction in prediction error.
Authors: KaiHui Huang, RunQing Wu, Fei Ye
Abstract: Continual learning has emerged as a pivotal area of research, primarily due to its advantageous characteristic that allows models to persistently acquire and retain information. However, catastrophic forgetting can severely impair model performance. In this study, we address network forgetting by introducing a novel framework termed Optimally-Weighted Maximum Mean Discrepancy (OWMMD), which imposes penalties on representation alterations via a Multi-Level Feature Matching Mechanism (MLFMM). Furthermore, we propose an Adaptive Regularization Optimization (ARO) strategy to refine the adaptive weight vectors, which autonomously assess the significance of each feature layer throughout the optimization process, The proposed ARO approach can relieve the over-regularization problem and promote the future task learning. We conduct a comprehensive series of experiments, benchmarking our proposed method against several established baselines. The empirical findings indicate that our approach achieves state-of-the-art performance.
Authors: Shinsaku Sakaue, Han Bao, Taira Tsuchiya
Abstract: This paper revisits the online learning approach to inverse linear optimization studied by B\"armann et al. (2017), where the goal is to infer an unknown linear objective function of an agent from sequential observations of the agent's input-output pairs. First, we provide a simple understanding of the online learning approach through its connection to online convex optimization of \emph{Fenchel--Young losses}. As a byproduct, we present an offline guarantee on the \emph{suboptimality loss}, which measures how well predicted objectives explain the agent's choices, without assuming the optimality of the agent's choices. Second, assuming that there is a gap between optimal and suboptimal objective values in the agent's decision problems, we obtain an upper bound independent of the time horizon $T$ on the sum of suboptimality and \emph{estimate losses}, where the latter measures the quality of solutions recommended by predicted objectives. Interestingly, our gap-dependent analysis achieves a faster rate than the standard $O(\sqrt{T})$ regret bound by exploiting structures specific to inverse linear optimization, even though neither the loss functions nor their domains enjoy desirable properties, such as strong convexity.
Authors: Zihui Wu, Haichang Gao, Jiacheng Luo, Zhaoxiang Liu
Abstract: Large Language Models (LLMs) commonly rely on explicit refusal prefixes for safety, making them vulnerable to prefix injection attacks. We introduce HumorReject, a novel data-driven approach that reimagines LLM safety by decoupling it from refusal prefixes through humor as an indirect refusal strategy. Rather than explicitly rejecting harmful instructions, HumorReject responds with contextually appropriate humor that naturally defuses potentially dangerous requests. Our approach effectively addresses common "over-defense" issues while demonstrating superior robustness against various attack vectors. Our findings suggest that improvements in training data design can be as important as the alignment algorithm itself in achieving effective LLM safety.
Authors: Nanjangud C. Narendra, Nithin Nagaraj
Abstract: Deep learning implemented via neural networks, has revolutionized machine learning by providing methods for complex tasks such as object detection/classification and prediction. However, architectures based on deep neural networks have started to yield diminishing returns, primarily due to their statistical nature and inability to capture causal structure in the training data. Another issue with deep learning is its high energy consumption, which is not that desirable from a sustainability perspective. Therefore, alternative approaches are being considered to address these issues, both of which are inspired by the functioning of the human brain. One approach is causal learning, which takes into account causality among the items in the dataset on which the neural network is trained. It is expected that this will help minimize the spurious correlations that are prevalent in the learned representations of deep neural networks. The other approach is Neurochaos Learning, a recent development, which draws its inspiration from the nonlinear chaotic firing intrinsic to neurons in biological neural networks (brain/central nervous system). Both approaches have shown improved results over just deep learning alone. To that end, in this position paper, we investigate how causal and neurochaos learning approaches can be integrated together to produce better results, especially in domains that contain linked data. We propose an approach for this integration to enhance classification, prediction and reinforcement learning. We also propose a set of research questions that need to be investigated in order to make this integration a reality.
Authors: Shuaiqun Pan, Diederick Vermetten, Manuel L\'opez-Ib\'a\~nez, Thomas B\"ack, Hao Wang
Abstract: Surrogate models are frequently employed as efficient substitutes for the costly execution of real-world processes. However, constructing a high-quality surrogate model often demands extensive data acquisition. A solution to this issue is to transfer pre-trained surrogate models for new tasks, provided that certain invariances exist between tasks. This study focuses on transferring non-differentiable surrogate models (e.g., random forest) from a source function to a target function, where we assume their domains are related by an unknown affine transformation, using only a limited amount of transfer data points evaluated on the target. Previous research attempts to tackle this challenge for differentiable models, e.g., Gaussian process regression, which minimizes the empirical loss on the transfer data by tuning the affine transformations. In this paper, we extend the previous work to the random forest model and assess its effectiveness on a widely-used artificial problem set - Black-Box Optimization Benchmark (BBOB) testbed, and on four real-world transfer learning problems. The results highlight the significant practical advantages of the proposed method, particularly in reducing both the data requirements and computational costs of training surrogate models for complex real-world scenarios.
Authors: Shinsaku Sakaue, Taira Tsuchiya, Han Bao, Taihei Oki
Abstract: We study an online learning problem where, over $T$ rounds, a learner observes both time-varying sets of feasible actions and an agent's optimal actions, selected by solving linear optimization over the feasible actions. The learner sequentially makes predictions of the agent's underlying linear objective function, and their quality is measured by the regret, the cumulative gap between optimal objective values and those achieved by following the learner's predictions. A seminal work by B\"armann et al. (ICML 2017) showed that online learning methods can be applied to this problem to achieve regret bounds of $O(\sqrt{T})$. Recently, Besbes et al. (COLT 2021, Oper. Res. 2023) significantly improved the result by achieving an $O(n^4\ln T)$ regret bound, where $n$ is the dimension of the ambient space of objective vectors. Their method, based on the ellipsoid method, runs in polynomial time but is inefficient for large $n$ and $T$. In this paper, we obtain an $O(n\ln T)$ regret bound, improving upon the previous bound of $O(n^4\ln T)$ by a factor of $n^3$. Our method is simple and efficient: we apply the online Newton step (ONS) to appropriate exp-concave loss functions. Moreover, for the case where the agent's actions are possibly suboptimal, we establish an $O(n\ln T+\sqrt{\Delta_Tn\ln T})$ regret bound, where $\Delta_T$ is the cumulative suboptimality of the agent's actions. This bound is achieved by using MetaGrad, which runs ONS with $\Theta(\ln T)$ different learning rates in parallel. We also provide a simple instance that implies an $\Omega(n)$ lower bound, showing that our $O(n\ln T)$ bound is tight up to an $O(\ln T)$ factor. This gives rise to a natural question: can the $O(\ln T)$ factor in the upper bound be removed? For the special case of $n=2$, we show that an $O(1)$ regret bound is possible, while we delineate challenges in extending this result to higher dimensions.
Authors: Dan Braun, Lucius Bushnaq, Stefan Heimersheim, Jake Mendel, Lee Sharkey
Abstract: Mechanistic interpretability aims to understand the internal mechanisms learned by neural networks. Despite recent progress toward this goal, it remains unclear how best to decompose neural network parameters into mechanistic components. We introduce Attribution-based Parameter Decomposition (APD), a method that directly decomposes a neural network's parameters into components that (i) are faithful to the parameters of the original network, (ii) require a minimal number of components to process any input, and (iii) are maximally simple. Our approach thus optimizes for a minimal length description of the network's mechanisms. We demonstrate APD's effectiveness by successfully identifying ground truth mechanisms in multiple toy experimental settings: Recovering features from superposition; separating compressed computations; and identifying cross-layer distributed representations. While challenges remain to scaling APD to non-toy models, our results suggest solutions to several open problems in mechanistic interpretability, including identifying minimal circuits in superposition, offering a conceptual foundation for 'features', and providing an architecture-agnostic framework for neural network decomposition.
Authors: Panayiotis Christou, Md. Zahidul Islam, Yuzhang Lin, Jingwei Xiong
Abstract: Power distribution networks are evolving due to the integration of DERs and increased customer participation. To maintain optimal operation, minimize losses, and meet varying load demands, frequent network reconfiguration is necessary. Traditionally, the reconfiguration task relies on optimization software and expert operators, but as systems grow more complex, faster and more adaptive solutions are required without expert intervention. Data-driven reconfiguration is gaining traction for its accuracy, speed, and robustness against incomplete network data. LLMs, with their ability to capture complex patterns, offer a promising approach for efficient and responsive network reconfiguration in evolving complex power networks. In this work, we introduce LLM4DistReconfig, a deep learning-based approach utilizing a fine-tuned LLM to solve the distribution network reconfiguration problem. By carefully crafting prompts and designing a custom loss function, we train the LLM with inputs representing network parameters such as buses, available lines, open lines, node voltages, and system loss. The model then predicts optimal reconfigurations by outputting updated network configurations that minimize system loss while meeting operational constraints. Our approach significantly reduces inference time compared to classical algorithms, allowing for near real-time optimal reconfiguration after training. Experimental results show that our method generates optimal configurations minimizing system loss for five individual and a combined test dataset. It also produces minimal invalid edges, no cycles, or subgraphs across all datasets, fulfilling domain-specific needs. Additionally, the generated responses contain less than 5% improper outputs on seen networks and satisfactory results on unseen networks, demonstrating its effectiveness and reliability for the reconfiguration task.
Authors: Sajjad Ghiasvand, Mahnoosh Alizadeh, Ramtin Pedarsani
Abstract: While parameter-efficient fine-tuning (PEFT) techniques like Low-Rank Adaptation (LoRA) offer computationally efficient adaptations of Large Language Models (LLMs), their practical deployment often assumes centralized data and training environments. However, real-world scenarios frequently involve distributed, privacy-sensitive datasets that require decentralized solutions. Federated learning (FL) addresses data privacy by coordinating model updates across clients, but it is typically based on centralized aggregation through a parameter server, which can introduce bottlenecks and communication constraints. Decentralized learning, in contrast, eliminates this dependency by enabling direct collaboration between clients, improving scalability and efficiency in distributed environments. Despite its advantages, decentralized LLM fine-tuning remains underexplored. In this work, we propose Dec-LoRA, an algorithm for decentralized fine-tuning of LLMs based on LoRA. Through extensive experiments on BERT and LLaMA-2 models, we show that Dec-LoRA maintains performance comparable to centralized LoRA across various conditions, including data heterogeneity and quantization constraints. This highlights its potential for scalable LLM fine-tuning in decentralized environments.
Authors: Tianbo Yang, Mingqi Yan, Hongyi Zhao, Tianshuo Yang
Abstract: Developing the logic necessary to solve mathematical problems or write mathematical proofs is one of the more difficult objectives for large language models (LLMS). Currently, the most popular methods in literature consists of fine-tuning the model on written mathematical content such as academic publications and textbooks, so that the model can learn to emulate the style of mathematical writing. In this project, we explore the effectiveness of using retrieval augmented generation (RAG) to address gaps in the mathematical reasoning of LLMs. We develop LemmaHead, a RAG knowledge base that supplements queries to the model with relevant mathematical context, with particular focus on context from published textbooks. To measure our model's performance in mathematical reasoning, our testing paradigm focuses on the task of automated theorem proving via generating proofs to a given mathematical claim in the Lean formal language.
Authors: Mingkuan Feng, Jinyang Wu, Shuai Zhang, Pengpeng Shao, Ruihan Jin, Zhengqi Wen, Jianhua Tao, Feihu Che
Abstract: Large language models (LLMs) have achieved significant progress across various domains, but their increasing scale results in high computational and memory costs. Recent studies have revealed that LLMs exhibit sparsity, providing the potential to reduce model size through pruning techniques. However, existing pruning methods typically follow a prune-then-finetune paradigm. Since the pruned components still contain valuable information, their direct removal often leads to irreversible performance degradation, imposing a substantial computational burden to recover performance during finetuning. In this paper, we propose a novel paradigm that first applies regularization, then prunes, and finally finetunes. Based on this paradigm, we introduce DReSS, a simple and effective Data-driven Regularized Structured Streamlining method for LLMs. By leveraging a small amount of data to regularize the components to be pruned, DReSS explicitly transfers the important information to the remaining parts of the model in advance. Compared to direct pruning, this can reduce the information loss caused by parameter removal, thereby enhancing its language modeling capabilities. Experimental results demonstrate that DReSS significantly outperforms existing pruning methods even under extreme pruning ratios, significantly reducing latency and increasing throughput.
Authors: Shengyang Sun, Yian Zhang, Alexander Bukharin, David Mosallanezhad, Jiaqi Zeng, Soumye Singhal, Gerald Shen, Adithya Renduchintala, Tugrul Konuk, Yi Dong, Zhilin Wang, Dmitry Chichkov, Olivier Delalleau, Oleksii Kuchaiev
Abstract: The rapid development of large language model (LLM) alignment algorithms has resulted in a complex and fragmented landscape, with limited clarity on the effectiveness of different methods and their inter-connections. This paper introduces Reward-Aware Preference Optimization (RPO), a mathematical framework that unifies popular preference optimization techniques in LLM alignment, including DPO, IPO, SimPO, and REINFORCE (LOO), among others. RPO provides a structured approach to disentangle and systematically study the impact of various design choices, such as the optimization objective, the number of responses per prompt, and the use of implicit versus explicit reward models, on LLM preference optimization. We additionally propose a new experimental setup that enables the clean and direct ablation of such design choices. Through an extensive series of ablation studies within the RPO framework, we gain insights into the critical factors shaping model alignment, offering practical guidance on the most effective strategies for improving LLM alignment.
Authors: Chuqi Chen, Yahong Yang, Yang Xiang, Wenrui Hao
Abstract: Solving partial differential equations (PDEs) using neural networks has become a central focus in scientific machine learning. Training neural networks for sharp interface problems is particularly challenging due to certain parameters in the PDEs that introduce near-singularities in the loss function. In this study, we overcome this challenge by introducing a novel method based on homotopy dynamics to effectively manipulate these parameters. From a theoretical perspective, we analyze the effects of these parameters on training difficulty in sharp interface problems and establish the convergence of the proposed homotopy dynamics method. Experimentally, we demonstrate that our approach significantly accelerates convergence and improves the accuracy of sharp interface capturing. These findings present an efficient optimization strategy leveraging homotopy dynamics, offering a robust framework to extend the applicability of neural networks for solving PDEs with sharp
Authors: Sam Jeong, Hae Yong Kim
Abstract: Machine learning has transformed the classification of biomedical signals such as electrocardiograms (ECGs). Advances in deep learning, particularly convolutional neural networks (CNNs), enable automatic feature extraction, raising the question: Can combining time- and frequency-domain attributes enhance classification accuracy? To explore this, we evaluated three ECG classification tasks: (1) arrhythmia classification, (2) identity recognition, and (3) apnea detection. We initially tested three methods: (i) 2-D spectrogram-based frequency-time classification (SPECT), (ii) time-domain classification using a 1-D CNN (CNN1D), and (iii) frequency-domain classification using a Fourier transform-based CNN (FFT1D). Performance was validated using K-fold cross-validation. Among these, CNN1D (time only) performed best, followed by SPECT (time-frequency) and FFT1D (frequency only). Surprisingly, SPECT, which integrates time- and frequency-domain features, performed worse than CNN1D, suggesting a need for a more effective time and frequency fusion approach. To address this, we tested the recently proposed Fourier Analysis Network (FAN), which combines time- and frequency-domain features. However, FAN performed comparably to CNN1D, excelling in some tasks while underperforming in others. To enhance this approach, we developed the Convolutional Fourier Analysis Network (CFAN), which integrates FAN with CNN. CFAN outperformed all previous methods across all classification tasks. These findings underscore the advantages of combining time- and frequency-domain features, demonstrating CFAN's potential as a powerful and versatile solution for ECG classification and broader biomedical signal analysis
Authors: Mingyu Chen, Yiding Chen, Wen Sun, Xuezhou Zhang
Abstract: Reinforcement Learning from Human Feedback (RLHF) has emerged as a pivotal technique for large language model (LLM) alignment. This paper studies the setting of online RLHF and focus on improving sample efficiency. All existing algorithms in online RLHF, whether doing passive exploration or active exploration, suffer from a sample complexity that scales exponentially with the scale of the reward function. This fundamental limitation hinders their effectiveness in scenarios with heavily skewed preferences, e.g. questions with a unique correct solution. To address this, we introduce Self-Exploring Preference-Incentive Online Preference Optimization (SE-POPO), an online RLHF algorithm that for the first time achieves a sample complexity that scales polynomially with the reward scale, answering an open problem raised by Xie et al. (2024).. Theoretically, we demonstrate that the sample complexity of SE-POPO dominates that of existing exploration algorithms. Empirically, our systematic evaluation confirms that SE-POPO is more sample-efficient than both exploratory and non-exploratory baselines, in two primary application scenarios of RLHF as well as on public benchmarks, marking a significant step forward in RLHF algorithm design. The code is available at https://github.com/MYC000801/SE-POPO.
Authors: Hanxun Huang, Sarah Erfani, Yige Li, Xingjun Ma, James Bailey
Abstract: Contrastive language-image pretraining (CLIP) has been found to be vulnerable to poisoning backdoor attacks where the adversary can achieve an almost perfect attack success rate on CLIP models by poisoning only 0.01\% of the training dataset. This raises security concerns on the current practice of pretraining large-scale models on unscrutinized web data using CLIP. In this work, we analyze the representations of backdoor-poisoned samples learned by CLIP models and find that they exhibit unique characteristics in their local subspace, i.e., their local neighborhoods are far more sparse than that of clean samples. Based on this finding, we conduct a systematic study on detecting CLIP backdoor attacks and show that these attacks can be easily and efficiently detected by traditional density ratio-based local outlier detectors, whereas existing backdoor sample detection methods fail. Our experiments also reveal that an unintentional backdoor already exists in the original CC3M dataset and has been trained into a popular open-source model released by OpenCLIP. Based on our detector, one can clean up a million-scale web dataset (e.g., CC3M) efficiently within 15 minutes using 4 Nvidia A100 GPUs. The code is publicly available in our \href{https://github.com/HanxunH/Detect-CLIP-Backdoor-Samples}{GitHub repository}.
URLs: https://github.com/HanxunH/Detect-CLIP-Backdoor-Samples
Authors: Jingzhe Shi, Qinwei Ma, Hongyi Liu, Hang Zhao, Jeng-Neng Hwang, Serge Belongie, Lei Li
Abstract: Long Context Language Models have drawn great attention in the past few years. There has been work discussing the impact of long context on Language Model performance: some find that long irrelevant context could harm performance, while some experimentally summarize loss reduction by relevant long context as Scaling Laws. This calls for a more thorough understanding on how long context impact Language Modeling. In this work, we (1) propose a clean and effective theoretical framework on explaining the impact of context length to Language Modeling, from an Intrinsic Space perspective; and (2) conduct experiments on natural language and synthetic data, validating our proposed theoretical assumptions and deductions. Our theoretical framework can provide practical insights such as establishing that training dataset size dictates an optimal context length and bounds context length scaling for certain case. We hope our work may inspire new long context Language Models, as well as future work studying Physics for Language Models. Code for our experiments is available at this url: https://github.com/JingzheShi/NLPCtlScalingAndBounds.
Authors: Yihe Wang, Nan Huang, Nadia Mammone, Marco Cecchi, Xiang Zhang
Abstract: Electroencephalogram (EEG) provides a non-invasive, highly accessible, and cost-effective solution for Alzheimer's Disease (AD) detection. However, existing methods, whether based on manual feature extraction or deep learning, face two major challenges: the lack of large-scale datasets for robust feature learning and evaluation, and poor detection performance due to inter-subject variations. To address these challenges, we curate an EEG-AD corpus containing 813 subjects, which forms the world's largest EEG-AD dataset to the best of our knowledge. Using this unique dataset, we propose LEAD, the first large foundation model for EEG-based AD detection. Our method encompasses an entire pipeline, from data selection and preprocessing to self-supervised contrastive pretraining, fine-tuning, and key setups such as subject-independent evaluation and majority voting for subject-level detection. We pre-train the model on 11 EEG datasets and unified fine-tune it on 5 AD datasets. Our self-supervised pre-training design includes sample-level and subject-level contrasting to extract useful general EEG features. Fine-tuning is performed on 5 channel-aligned datasets together. The backbone encoder incorporates temporal and channel embeddings to capture features across both temporal and spatial dimensions. Our method demonstrates outstanding AD detection performance, achieving up to a 9.86% increase in F1 score at the sample-level and up to a 9.31% at the subject-level compared to state-of-the-art methods. The results of our model strongly confirm the effectiveness of contrastive pre-training and channel-aligned unified fine-tuning for addressing inter-subject variation. The source code is at https://github.com/DL4mHealth/LEAD.
Authors: Ziyang Zheng, Shan Huang, Jianyuan Zhong, Zhengyuan Shi, Guohao Dai, Ningyi Xu, Qiang Xu
Abstract: Circuit representation learning has become pivotal in electronic design automation, enabling critical tasks such as testability analysis, logic reasoning, power estimation, and SAT solving. However, existing models face significant challenges in scaling to large circuits due to limitations like over-squashing in graph neural networks and the quadratic complexity of transformer-based models. To address these issues, we introduce DeepGate4, a scalable and efficient graph transformer specifically designed for large-scale circuits. DeepGate4 incorporates several key innovations: (1) an update strategy tailored for circuit graphs, which reduce memory complexity to sub-linear and is adaptable to any graph transformer; (2) a GAT-based sparse transformer with global and local structural encodings for AIGs; and (3) an inference acceleration CUDA kernel that fully exploit the unique sparsity patterns of AIGs. Our extensive experiments on the ITC99 and EPFL benchmarks show that DeepGate4 significantly surpasses state-of-the-art methods, achieving 15.5% and 31.1% performance improvements over the next-best models. Furthermore, the Fused-DeepGate4 variant reduces runtime by 35.1% and memory usage by 46.8%, making it highly efficient for large-scale circuit analysis. These results demonstrate the potential of DeepGate4 to handle complex EDA tasks while offering superior scalability and efficiency.
Authors: Xianglong Yan, Tianao Zhang, Zhiteng Li, Yulun Zhang
Abstract: Large language models (LLMs) have achieved remarkable success in natural language processing tasks, but their high computational and memory demands pose challenges for deployment on resource-constrained devices. Binarization, as an efficient compression method that reduces model weights to just 1 bit, significantly lowers both computational and memory requirements. Despite this, the binarized LLM still contains redundancy, which can be further compressed. Semi-structured pruning provides a promising approach to achieve this, which offers a better trade-off between model performance and hardware efficiency. However, simply combining binarization with semi-structured pruning can lead to a significant performance drop. To address this issue, we propose a Progressive Binarization with Semi-Structured Pruning (PBS$^2$P) method for LLM compression. We first propose a Stepwise semi-structured Pruning with Binarization Optimization (SPBO). Our optimization strategy significantly reduces the total error caused by pruning and binarization, even below that of the no-pruning scenario. Furthermore, we design a Coarse-to-Fine Search (CFS) method to select pruning elements more effectively. Extensive experiments demonstrate that PBS$^2$P achieves superior accuracy across various LLM families and evaluation metrics, noticeably outperforming state-of-the-art (SOTA) binary PTQ methods. The code and models will be available at https://github.com/XIANGLONGYAN/PBS2P.
Authors: Arian Eamaz, Farhang Yeganegi, Mojtaba Soltanalian
Abstract: Recent advancements in Large Language Model (LLM) compression, such as BitNet and BitNet b1.58, have marked significant strides in reducing the computational demands of LLMs through innovative one-bit quantization techniques. We extend this frontier by looking at Large Inference Models (LIMs) that have become indispensable across various applications. However, their scale and complexity often come at a significant computational cost. We introduce a novel approach that leverages one-bit algorithm unrolling, effectively integrating information from the physical world in the model architecture. Our method achieves a bit-per-link rate significantly lower than the 1.58 bits reported in prior work, thanks to the natural sparsity that emerges in our network architectures. We numerically demonstrate that the proposed one-bit algorithm unrolling scheme can improve both training and test outcomes by effortlessly increasing the number of layers while substantially compressing the network. Additionally, we provide theoretical results on the generalization gap, convergence rate, stability, and sensitivity of our proposed one-bit algorithm unrolling.
Authors: Frederik Lizak Johansen, Ulrik Friis-Jensen, Erik Bj{\o}rnager Dam, Kirsten Marie {\O}rnsbjerg Jensen, Roc\'io Mercado, Raghavendra Selvan
Abstract: Novel materials drive progress across applications from energy storage to electronics. Automated characterization of material structures with machine learning methods offers a promising strategy for accelerating this key step in material design. In this work, we introduce an autoregressive language model that performs crystal structure prediction (CSP) from powder diffraction data. The presented model, deCIFer, generates crystal structures in the widely used Crystallographic Information File (CIF) format and can be conditioned on powder X-ray diffraction (PXRD) data. Unlike earlier works that primarily rely on high-level descriptors like composition, deCIFer performs CSP from diffraction data. We train deCIFer on nearly 2.3M unique crystal structures and validate on diverse sets of PXRD patterns for characterizing challenging inorganic crystal systems. Qualitative and quantitative assessments using the residual weighted profile and Wasserstein distance show that deCIFer produces structures that more accurately match the target diffraction data when conditioned, compared to the unconditioned case. Notably, deCIFer can achieve a 94% match rate on unseen data. deCIFer bridges experimental diffraction data with computational CSP, lending itself as a powerful tool for crystal structure characterization and accelerating materials discovery.
Authors: Evan Chen, Jianing Zhang, Shiqiang Wang, Chaoyue Liu, Christopher Brinton
Abstract: In Federated Learning (FL), model training performance is strongly impacted by data heterogeneity across clients. Gradient Tracking (GT) has recently emerged as a solution which mitigates this issue by introducing correction terms to local model updates. To date, GT has only been considered under Stochastic Gradient Descent (SGD)-based model training, while modern FL frameworks increasingly employ adaptive optimizers for improved convergence. In this work, we generalize the GT framework to a more flexible Parameter Tracking (PT) paradigm and propose two novel adaptive optimization algorithms, {\tt FAdamET} and {\tt FAdamGT}, that integrate PT into Adam-based FL. We provide a rigorous convergence analysis of these algorithms under non-convex settings. Our experimental results demonstrate that both proposed algorithms consistently outperform existing methods when evaluating total communication cost and total computation cost across varying levels of data heterogeneity, showing the effectiveness of correcting first-order information in federated adaptive optimization.
Authors: Saba Sanami, Amir G. Aghdam
Abstract: This paper investigates unsupervised anomaly detection in multivariate time-series data using reinforcement learning (RL) in the latent space of an autoencoder. A significant challenge is the limited availability of anomalous data, often leading to misclassifying anomalies as normal events, thus raising false negatives. RL can help overcome this limitation by promoting exploration and balancing exploitation during training, effectively preventing overfitting. Wavelet analysis is also utilized to enhance anomaly detection, enabling time-series data decomposition into both time and frequency domains. This approach captures anomalies at multiple resolutions, with wavelet coefficients extracted to detect both sudden and subtle shifts in the data, thereby refining the anomaly detection process. We calibrate the decision boundary by generating synthetic anomalies and embedding a supervised framework within the model. This supervised element aids the unsupervised learning process by fine-tuning the decision boundary and increasing the model's capacity to distinguish between normal and anomalous patterns effectively.
Authors: Junhao Song, Yichao Zhang, Ziqian Bi, Tianyang Wang, Keyu Chen, Ming Li, Qian Niu, Junyu Liu, Benji Peng, Sen Zhang, Ming Liu, Jiawei Xu, Xuanhe Pan, Jinlang Wang, Pohsun Feng, Yizhu Wen, Lawrence K. Q. Yan, Hong-Ming Tseng, Xinyuan Song, Jintao Ren, Silin Chen, Yunze Wang, Weiche Hsieh, Bowen Jing, Junjie Yang, Jun Zhou, Zheyu Yao, Chia Xin Liang
Abstract: Generative Adversarial Networks (GAN) have greatly influenced the development of computer vision and artificial intelligence in the past decade and also connected art and machine intelligence together. This book begins with a detailed introduction to the fundamental principles and historical development of GANs, contrasting them with traditional generative models and elucidating the core adversarial mechanisms through illustrative Python examples. The text systematically addresses the mathematical and theoretical underpinnings including probability theory, statistics, and game theory providing a solid framework for understanding the objectives, loss functions, and optimisation challenges inherent to GAN training. Subsequent chapters review classic variants such as Conditional GANs, DCGANs, InfoGAN, and LAPGAN before progressing to advanced training methodologies like Wasserstein GANs, GANs with gradient penalty, least squares GANs, and spectral normalisation techniques. The book further examines architectural enhancements and task-specific adaptations in generators and discriminators, showcasing practical implementations in high resolution image generation, artistic style transfer, video synthesis, text to image generation and other multimedia applications. The concluding sections offer insights into emerging research trends, including self-attention mechanisms, transformer-based generative models, and a comparative analysis with diffusion models, thus charting promising directions for future developments in both academic and applied settings.
Authors: Miaomiao Li, Hao Chen, Yang Wang, Tingyuan Zhu, Weijia Zhang, Kaijie Zhu, Kam-Fai Wong, Jindong Wang
Abstract: Generating synthetic datasets via large language models (LLMs) themselves has emerged as a promising approach to improve LLM performance. However, LLMs inherently reflect biases present in their training data, leading to a critical challenge: when these models generate synthetic data for training, they may propagate and amplify their inherent biases that can significantly impact model fairness and robustness on downstream tasks--a phenomenon we term bias inheritance. This work presents the first systematic investigation in understanding, analyzing, and mitigating bias inheritance. We study this problem by fine-tuning LLMs with a combined dataset consisting of original and LLM-augmented data, where bias ratio represents the proportion of augmented data. Through systematic experiments across 10 classification and generation tasks, we analyze how 6 different types of biases manifest at varying bias ratios. Our results reveal that bias inheritance has nuanced effects on downstream tasks, influencing both classification tasks and generation tasks differently. Then, our analysis identifies three key misalignment factors: misalignment of values, group data, and data distributions. Based on these insights, we propose three mitigation strategies: token-based, mask-based, and loss-based approaches. Experiments demonstrate that these strategies also work differently on various tasks and bias, indicating the substantial challenges to fully mitigate bias inheritance. We hope this work can provide valuable insights to the research of LLM data augmentation.
Authors: An Ji, Bortik Bandyopadhyay, Congzheng Song, Natarajan Krishnaswami, Prabal Vashisht, Rigel Smiroldo, Isabel Litton, Sayantan Mahinder, Mona Chitnis, Andrew W Hill
Abstract: This paper presents an implementation of machine learning model training using private federated learning (PFL) on edge devices. We introduce a novel framework that uses PFL to address the challenge of training a model using users' private data. The framework ensures that user data remain on individual devices, with only essential model updates transmitted to a central server for aggregation with privacy guarantees. We detail the architecture of our app selection model, which incorporates a neural network with attention mechanisms and ambiguity handling through uncertainty management. Experiments conducted through off-line simulations and on device training demonstrate the feasibility of our approach in real-world scenarios. Our results show the potential of PFL to improve the accuracy of an app selection model by adapting to changes in user behavior over time, while adhering to privacy standards. The insights gained from this study are important for industries looking to implement PFL, offering a robust strategy for training a predictive model directly on edge devices while ensuring user data privacy.
Authors: Brian Formento, Chuan Sheng Foo, See-Kiong Ng
Abstract: A fundamental issue in deep learning has been adversarial robustness. As these systems have scaled, such issues have persisted. Currently, large language models (LLMs) with billions of parameters suffer from adversarial attacks just like their earlier, smaller counterparts. However, the threat models have changed. Previously, having gray-box access, where input embeddings or output logits/probabilities were visible to the user, might have been reasonable. However, with the introduction of closed-source models, no information about the model is available apart from the generated output. This means that current black-box attacks can only utilize the final prediction to detect if an attack is successful. In this work, we investigate and demonstrate the potential of attack guidance, akin to using output probabilities, while having only black-box access in a classification setting. This is achieved through the ability to elicit confidence from the model. We empirically show that the elicited confidence is calibrated and not hallucinated for current LLMs. By minimizing the elicited confidence, we can therefore increase the likelihood of misclassification. Our new proposed paradigm demonstrates promising state-of-the-art results on three datasets across two models (LLaMA-3-8B-Instruct and Mistral-7B-Instruct-V0.3) when comparing our technique to existing hard-label black-box attack methods that introduce word-level substitutions.
Authors: Hannah Kuehn, Joseph La Delfa, Miguel Vasco, Danica Kragic, Iolanda Leite
Abstract: Modern embodied artificial agents excel in static, predefined tasks but fall short in dynamic and long-term interactions with humans. On the other hand, humans can adapt and evolve continuously, exploiting the situated knowledge embedded in their environment and other agents, thus contributing to meaningful interactions. We introduce the concept of co-existence for embodied artificial agents and argues that it is a prerequisite for meaningful, long-term interaction with humans. We take inspiration from biology and design theory to understand how human and non-human organisms foster entities that co-exist within their specific niches. Finally, we propose key research directions for the machine learning community to foster co-existing embodied agents, focusing on the principles, hardware and learning methods responsible for shaping them.
Authors: Finn Rietz, Oleg Smirnov, Sara Karimi, Lele Cao
Abstract: Harnessing large offline datasets is vital for training foundation models that can generalize across diverse tasks. Offline Reinforcement Learning (RL) offers a powerful framework for these scenarios, enabling the derivation of optimal policies even from suboptimal data. The Prompting Decision Transformer (PDT) is an offline RL multi-task model that distinguishes tasks through stochastic trajectory prompts, which are task-specific tokens maintained in context during rollouts. However, PDT samples these tokens uniformly at random from per-task demonstration datasets, failing to account for differences in token informativeness and potentially leading to performance degradation. To address this limitation, we introduce a scalable bandit-based prompt-tuning method that dynamically learns to construct high-performance trajectory prompts. Our approach significantly enhances downstream task performance without modifying the pre-trained Transformer backbone. Empirical results on benchmark tasks and a newly designed multi-task environment demonstrate the effectiveness of our method, creating a seamless bridge between general multi-task offline pre-training and task-specific online adaptation.
Authors: Danielle L. Ferreira, Zaynaf Salaymang, Rima Arnaout
Abstract: Segmentation and measurement of cardiac chambers is critical in cardiac ultrasound but is laborious and poorly reproducible. Neural networks can assist, but supervised approaches require the same laborious manual annotations. We built a pipeline for self-supervised (no manual labels) segmentation combining computer vision, clinical domain knowledge, and deep learning. We trained on 450 echocardiograms (93,000 images) and tested on 8,393 echocardiograms (4,476,266 images; mean 61 years, 51% female), using the resulting segmentations to calculate biometrics. We also tested against external images from an additional 10,030 patients with available manual tracings of the left ventricle. r2 between clinically measured and pipeline-predicted measurements were similar to reported inter-clinician variation and comparable to supervised learning across several different measurements (r2 0.56-0.84). Average accuracy for detecting abnormal chamber size and function was 0.85 (range 0.71-0.97) compared to clinical measurements. A subset of test echocardiograms (n=553) had corresponding cardiac MRIs, where MRI is the gold standard. Correlation between pipeline and MRI measurements was similar to that between clinical echocardiogram and MRI. Finally, the pipeline accurately segments the left ventricle with an average Dice score of 0.89 (95% CI [0.89]) in the external, manually labeled dataset. Our results demonstrate a manual-label free, clinically valid, and highly scalable method for segmentation from ultrasound, a noisy but globally important imaging modality.
Authors: Kui Ren, Lu Zhang
Abstract: The task of simultaneously reconstructing multiple physical coefficients in partial differential equations (PDEs) from observed data is ubiquitous in applications. In this work, we propose an integrated data-driven and model-based iterative reconstruction framework for such joint inversion problems where additional data on the unknown coefficients are supplemented for better reconstructions. Our method couples the supplementary data with the PDE model to make the data-driven modeling process consistent with the model-based reconstruction procedure. We characterize the impact of learning uncertainty on the joint inversion results for two typical inverse problems. Numerical evidence is provided to demonstrate the feasibility of using data-driven models to improve the joint inversion of multiple coefficients in PDEs.
Authors: Jinus Bordbar, Mohammadreza Mohammadrezaie, Saman Ardalan, Mohammad Ebrahim Shiri
Abstract: Online social media is integral to human life, facilitating messaging, information sharing, and confidential communication while preserving privacy. Platforms like Twitter, Instagram, and Facebook exemplify this phenomenon. However, users face challenges due to network anomalies, often stemming from malicious activities such as identity theft for financial gain or harm. This paper proposes a novel method using user similarity measures and the Generative Adversarial Network (GAN) algorithm to identify fake user accounts in the Twitter dataset. Despite the problem's complexity, the method achieves an AUC rate of 80\% in classifying and detecting fake accounts. Notably, the study builds on previous research, highlighting advancements and insights into the evolving landscape of anomaly detection in online social networks.
Authors: Fabian Paischer, Markus Hofmarcher, Sepp Hochreiter, Thomas Adler
Abstract: Recently, vision-language models like CLIP have advanced the state of the art in a variety of multi-modal tasks including image captioning and caption evaluation. Many approaches leverage CLIP for cross-modal retrieval to condition pre-trained language models on visual input. However, CLIP generally suffers from a mis-alignment of image and text modalities in the joint embedding space. We investigate efficient methods to linearly re-align the joint embedding space for the downstream task of image captioning. This leads to an efficient training protocol that merely requires computing a closed-form solution for a linear mapping in the joint CLIP space. Consequently, we propose a lightweight captioning method called ReCap, which can be trained up to 1000 times faster than existing lightweight methods. Moreover, we propose two new learning-based image-captioning metrics built on CLIP score along with our proposed alignment. We evaluate ReCap on MS-COCO, Flickr30k, VizWiz and MSRVTT. On the former two, ReCap performs comparably to state-of-the-art lightweight methods using rule-based metrics while outperforming them on most of the CLIP-based metrics. On the latter two benchmarks, ReCap consistently outperforms competitors across all metrics and exhibits strong transfer capabilities and resilience to noise. Finally, we demonstrate that our proposed metrics correlate stronger with human judgement than existing metrics on the Flickr8k-Expert, Flickr8k-Crowdflower, and THumB datasets.
Authors: Hongling Zheng, Li Shen, Anke Tang, Yong Luo, Han Hu, Bo Du, Yonggang Wen, Dacheng Tao
Abstract: Foundation models (FM) have demonstrated remarkable performance across a wide range of tasks (especially in the fields of natural language processing and computer vision), primarily attributed to their ability to comprehend instructions and access extensive, high-quality data. This not only showcases their current effectiveness but also sets a promising trajectory towards the development of artificial general intelligence. Unfortunately, due to multiple constraints, the raw data of the model used for large model training are often inaccessible, so the use of end-to-end models for downstream tasks has become a new research trend, which we call Learn From Model (LFM) in this article. LFM focuses on the research, modification, and design of FM based on the model interface, so as to better understand the model structure and weights (in a black box environment), and to generalize the model to downstream tasks. The study of LFM techniques can be broadly categorized into five major areas: model tuning, model distillation, model reuse, meta learning and model editing. Each category encompasses a repertoire of methods and strategies that aim to enhance the capabilities and performance of FM. This paper gives a comprehensive review of the current methods based on FM from the perspective of LFM, in order to help readers better understand the current research status and ideas. To conclude, we summarize the survey by highlighting several critical areas for future exploration and addressing open issues that require further attention from the research community. The relevant papers we investigated in this article can be accessed at https://github.com/ruthless-man/Awesome-Learn-from-Model
URLs: https://github.com/ruthless-man/Awesome-Learn-from-Model
Authors: Reda Ouhamma, Maryam Kamgarpour
Abstract: We address payoff-based decentralized learning in infinite-horizon zero-sum Markov games. In this setting, each player makes decisions based solely on received rewards, without observing the opponent's strategy or actions nor sharing information. Prior works established finite-time convergence to an approximate Nash equilibrium under strong reachability and mixing time assumptions. We propose a convergent algorithm that significantly relaxes these assumptions, requiring only the existence of a single policy (not necessarily known) with bounded reachability and mixing time. Our key technical novelty is introducing Tsallis entropy regularization to smooth the best-response policy updates. By suitably tuning this regularization, we ensure sufficient exploration, thus bypassing previous stringent assumptions on the MDP. By establishing novel properties of the value and policy updates induced by the Tsallis entropy regularizer, we prove finite-time convergence to an approximate Nash equilibrium.
Authors: Qingsi Lai, Fanjie Xu, Lin Yao, Zhifeng Gao, Siyuan Liu, Hongshuai Wang, Shuqi Lu, Di He, Liwei Wang, Cheng Wang, Guolin Ke
Abstract: Powder X-ray diffraction (PXRD) is a prevalent technique in materials characterization. While the analysis of PXRD often requires extensive human manual intervention, and most automated method only achieved at coarse-grained level. The more difficult and important task of fine-grained crystal structure prediction from PXRD remains unaddressed. This study introduces XtalNet, the first equivariant deep generative model for end-to-end crystal structure prediction from PXRD. Unlike previous crystal structure prediction methods that rely solely on composition, XtalNet leverages PXRD as an additional condition, eliminating ambiguity and enabling the generation of complex organic structures with up to 400 atoms in the unit cell. XtalNet comprises two modules: a Contrastive PXRD-Crystal Pretraining (CPCP) module that aligns PXRD space with crystal structure space, and a Conditional Crystal Structure Generation (CCSG) module that generates candidate crystal structures conditioned on PXRD patterns. Evaluation on two MOF datasets (hMOF-100 and hMOF-400) demonstrates XtalNet's effectiveness. XtalNet achieves a top-10 Match Rate of 90.2% and 79% for hMOF-100 and hMOF-400 in conditional crystal structure prediction task, respectively. XtalNet enables the direct prediction of crystal structures from experimental measurements, eliminating the need for manual intervention and external databases. This opens up new possibilities for automated crystal structure determination and the accelerated discovery of novel materials.
Authors: Maciej Besta, Florim Memedi, Zhenyu Zhang, Robert Gerstenberger, Guangyuan Piao, Nils Blach, Piotr Nyczyk, Marcin Copik, Grzegorz Kwa\'sniewski, J\"urgen M\"uller, Lukas Gianinazzi, Ales Kubicek, Hubert Niewiadomski, Aidan O'Mahony, Onur Mutlu, Torsten Hoefler
Abstract: The field of natural language processing (NLP) has witnessed significant progress in recent years, with a notable focus on improving large language models' (LLM) performance through innovative prompting techniques. Among these, prompt engineering coupled with structures has emerged as a promising paradigm, with designs such as Chain-of-Thought, Tree of Thoughts, or Graph of Thoughts, in which the overall LLM reasoning is guided by a structure such as a graph. As illustrated with numerous examples, this paradigm significantly enhances the LLM's capability to solve numerous tasks, ranging from logical or mathematical reasoning to planning or creative writing. To facilitate the understanding of this growing field and pave the way for future developments, we devise a general blueprint for effective and efficient LLM reasoning schemes. For this, we conduct an in-depth analysis of the prompt execution pipeline, clarifying and clearly defining different concepts. We then build the first taxonomy of structure-enhanced LLM reasoning schemes. We focus on identifying fundamental classes of harnessed structures, and we analyze the representations of these structures, algorithms executed with these structures, and many others. We refer to these structures as reasoning topologies, because their representation becomes to a degree spatial, as they are contained within the LLM context. Our study compares existing prompting schemes using the proposed taxonomy, discussing how certain design choices lead to different patterns in performance and cost. We also outline theoretical underpinnings, relationships between prompting and other parts of the LLM ecosystem such as knowledge bases, and the associated research challenges. Our work will help to advance future prompt engineering techniques.
Authors: Julian Strohmayer, Rafael Sterzinger, Christian Stippel, Martin Kampel
Abstract: This work presents a seminal approach for synthesizing images from WiFi Channel State Information (CSI) in through-wall scenarios. Leveraging the strengths of WiFi, such as cost-effectiveness, illumination invariance, and wall-penetrating capabilities, our approach enables visual monitoring of indoor environments beyond room boundaries and without the need for cameras. More generally, it improves the interpretability of WiFi CSI by unlocking the option to perform image-based downstream tasks, e.g., visual activity recognition. In order to achieve this crossmodal translation from WiFi CSI to images, we rely on a multimodal Variational Autoencoder (VAE) adapted to our problem specifics. We extensively evaluate our proposed methodology through an ablation study on architecture configuration and a quantitative/qualitative assessment of reconstructed images. Our results demonstrate the viability of our method and highlight its potential for practical applications.
Authors: Eftekhar Hossain, Sanjeev Kumar Sinha, Naman Bansal, Alex Knipper, Souvika Sarkar, John Salvador, Yash Mahajan, Sri Guttikonda, Mousumi Akter, Md. Mahadi Hassan, Matthew Freestone, Matthew C. Williams Jr., Dongji Feng, Santu Karmaker
Abstract: One of the most important yet onerous tasks in the academic peer-reviewing process is composing meta-reviews, which involves assimilating diverse opinions from multiple expert peers, formulating one's self-judgment as a senior expert, and then summarizing all these perspectives into a concise holistic overview to make an overall recommendation. This process is time-consuming and can be compromised by human factors like fatigue, inconsistency, missing tiny details, etc. Given the latest major developments in Large Language Models (LLMs), it is very compelling to rigorously study whether LLMs can help metareviewers perform this important task better. In this paper, we perform a case study with three popular LLMs, i.e., GPT-3.5, LLaMA2, and PaLM2, to assist meta-reviewers in better comprehending multiple experts perspectives by generating a controlled multi-perspective summary (MPS) of their opinions. To achieve this, we prompt three LLMs with different types/levels of prompts based on the recently proposed TELeR taxonomy. Finally, we perform a detailed qualitative study of the MPSs generated by the LLMs and report our findings.
Authors: Juan Altmayer Pizzorno, Emery D. Berger
Abstract: Testing is an essential part of software development. Test generation tools attempt to automate the otherwise labor-intensive task of test creation, but generating high-coverage tests remains challenging. This paper proposes CoverUp, a novel approach to driving the generation of high-coverage Python regression tests. CoverUp combines coverage analysis, code context, and feedback in prompts that iteratively guide the LLM to generate tests that improve line and branch coverage. We evaluate our prototype CoverUp implementation across a benchmark of challenging code derived from open-source Python projects and show that CoverUp substantially improves on the state of the art. Compared to CodaMosa, a hybrid search/LLM-based test generator, CoverUp achieves a per-module median line+branch coverage of 80% (vs. 47%). Compared to MuTAP, a mutation- and LLM-based test generator, CoverUp achieves an overall line+branch coverage of 90% (vs. 77%). We also demonstrate that CoverUp's performance stems not only from the LLM used but from the combined effectiveness of its components.
Authors: Alexander Nemecek, Yuzhou Jiang, Erman Ayday
Abstract: The indistinguishability of Large Language Model (LLM) output from human-authored content poses significant challenges, raising concerns about potential misuse of AI-generated text and its influence on future AI model training. Watermarking algorithms offer a viable solution by embedding detectable signatures into generated text. However, existing watermarking methods often entail trade-offs among attack robustness, generation quality, and additional overhead such as specialized frameworks or complex integrations. We propose a lightweight, topic-guided watermarking scheme for LLMs that partitions the vocabulary into topic-aligned token subsets. Given an input prompt, the scheme selects a relevant topic-specific token list, effectively "green-listing" semantically aligned tokens to embed robust marks while preserving the text's fluency and coherence. Experimental results across multiple LLMs and state-of-the-art benchmarks demonstrate that our method achieves comparable perplexity to industry-leading systems, including Google's SynthID-Text, yet enhances watermark robustness against paraphrasing and lexical perturbation attacks while introducing minimal performance overhead. Our approach avoids reliance on additional mechanisms beyond standard text generation pipelines, facilitating straightforward adoption, suggesting a practical path toward globally consistent watermarking of AI-generated content.
Authors: Jinheon Baek, Sujay Kumar Jauhar, Silviu Cucerzan, Sung Ju Hwang
Abstract: The pace of scientific research, vital for improving human life, is complex, slow, and needs specialized expertise. Meanwhile, novel, impactful research often stems from both a deep understanding of prior work, and a cross-pollination of ideas across domains and fields. To enhance the productivity of researchers, we propose ResearchAgent, which leverages the encyclopedic knowledge and linguistic reasoning capabilities of Large Language Models (LLMs) to assist them in their work. This system automatically defines novel problems, proposes methods and designs experiments, while iteratively refining them based on the feedback from collaborative LLM-powered reviewing agents. Specifically, starting with a core scientific paper, ResearchAgent is augmented not only with relevant publications by connecting information over an academic graph but also entities retrieved from a knowledge store derived from shared underlying concepts mined across numerous papers. Then, mimicking a scientific approach to improving ideas with peer discussions, we leverage multiple LLM-based ReviewingAgents that provide reviews and feedback via iterative revision processes. These reviewing agents are instantiated with human preference-aligned LLMs whose criteria for evaluation are elicited from actual human judgments via LLM prompting. We experimentally validate our ResearchAgent on scientific publications across multiple disciplines, showing its effectiveness in generating novel, clear, and valid ideas based on both human and model-based evaluation results. Our initial foray into AI-mediated scientific research has important implications for the development of future systems aimed at supporting researchers in their ideation and operationalization of novel work.
Authors: Tomohiko Mizutani
Abstract: Hyperspectral imaging technology has a wide range of applications, including forest management, mineral resource exploration, and Earth surface monitoring. A key step in utilizing this technology is endmember extraction, which aims to identify the spectral signatures of materials in observed scenes. Theoretical studies suggest that self-dictionary methods using linear programming (LP), known as Hottopixx methods, are effective in extracting endmembers. However, their practical application is hindered by high computational costs, as they require solving LP problems whose size grows quadratically with the number of pixels in the image. As a result, their actual effectiveness remains unclear. To address this issue, we propose an enhanced implementation of Hottopixx designed to reduce computational time and improve endmember extraction performance. We demonstrate its effectiveness through experiments. The results suggest that our implementation enables the application of Hottopixx for endmember extraction from real hyperspectral images and allows us to achieve reasonably high accuracy in estimating endmember signatures.
Authors: Benjamin Dupuis, Paul Viallard, George Deligiannidis, Umut Simsekli
Abstract: We propose data-dependent uniform generalization bounds by approaching the problem from a PAC-Bayesian perspective. We first apply the PAC-Bayesian framework on "random sets" in a rigorous way, where the training algorithm is assumed to output a data-dependent hypothesis set after observing the training data. This approach allows us to prove data-dependent bounds, which can be applicable in numerous contexts. To highlight the power of our approach, we consider two main applications. First, we propose a PAC-Bayesian formulation of the recently developed fractal-dimension-based generalization bounds. The derived results are shown to be tighter and they unify the existing results around one simple proof technique. Second, we prove uniform bounds over the trajectories of continuous Langevin dynamics and stochastic gradient Langevin dynamics. These results provide novel information about the generalization properties of noisy algorithms.
Authors: Shadab Ahamed, Arman Rahmim
Abstract: Minimizing the need for pixel-level annotated data to train PET lesion detection and segmentation networks is highly desired and can be transformative, given time and cost constraints associated with expert annotations. Current un-/weakly-supervised anomaly detection methods rely on autoencoder or generative adversarial networks trained only on healthy data; however GAN-based networks are more challenging to train due to issues with simultaneous optimization of two competing networks, mode collapse, etc. In this paper, we present the weakly-supervised Implicitly guided COuNterfactual diffusion model for Detecting Anomalies in PET images (IgCONDA-PET). The solution is developed and validated using PET scans from six retrospective cohorts consisting of a total of 2652 cases containing both local and public datasets. The training is conditioned on image class labels (healthy vs. unhealthy) via attention modules, and we employ implicit diffusion guidance. We perform counterfactual generation which facilitates "unhealthy-to-healthy" domain translation by generating a synthetic, healthy version of an unhealthy input image, enabling the detection of anomalies through the calculated differences. The performance of our method was compared against several other deep learning based weakly-supervised or unsupervised methods as well as traditional methods like 41% SUVmax thresholding. We also highlight the importance of incorporating attention modules in our network for the detection of small anomalies. The code is publicly available at: https://github.com/ahxmeds/IgCONDA-PET.git.
Authors: Kwangho Kim, Jisu Kim, Edward H. Kennedy
Abstract: Causal effects are often characterized with population summaries. These might provide an incomplete picture when there are heterogeneous treatment effects across subgroups. Since the subgroup structure is typically unknown, it is more challenging to identify and evaluate subgroup effects than population effects. We propose a new solution to this problem: Causal k-Means Clustering, which harnesses the widely-used k-means clustering algorithm to uncover the unknown subgroup structure. Our problem differs significantly from the conventional clustering setup since the variables to be clustered are unknown counterfactual functions. We present a plug-in estimator which is simple and readily implementable using off-the-shelf algorithms, and study its rate of convergence. We also develop a new bias-corrected estimator based on nonparametric efficiency theory and double machine learning, and show that this estimator achieves fast root-n rates and asymptotic normality in large nonparametric models. Our proposed methods are especially useful for modern outcome-wide studies with multiple treatment levels. Further, our framework is extensible to clustering with generic pseudo-outcomes, such as partially observed outcomes or otherwise unknown functions. Finally, we explore finite sample properties via simulation, and illustrate the proposed methods in a study of treatment programs for adolescent substance abuse.
Authors: Roy Maor Lotan, Inbal Talgam-Cohen, Yaniv Romano
Abstract: Auctions are key for maximizing sellers' revenue and ensuring truthful bidding among buyers. Recently, an approach known as differentiable economics based on machine learning (ML) has shown promise in learning powerful auction mechanisms for multiple items and participants. However, this approach has no guarantee of strategy-proofness at test time. Strategy-proofness is crucial as it ensures that buyers are incentivized to bid their true valuations, leading to optimal and fair auction outcomes without the risk of manipulation. In this work, we propose a formulation of statistical strategy-proofness auction mechanism, ensuring that the probability of regret exceeding a predefined threshold is strictly controlled. Building upon conformal prediction techniques, we develop an auction acceptance rule that leverages regret predictions to guarantee that the data-driven auction mechanism meets the statistical strategy-proofness requirement with high probability. Our approach represents a practical middle-ground between two extremes: forcing zero-regret at the cost of significant revenue loss, and naively using ML to construct auctions with the hope of attaining low regret at test time. Numerical experiments demonstrate the necessity of the proposed method, the validity of our theoretical result, and its applicability.
Authors: Guoyao Shen, Mengyu Li, Stephan Anderson, Chad W. Farris, Xin Zhang
Abstract: Recent advancements in deep learning have enabled the development of generalizable models that achieve state-of-the-art performance across various imaging tasks. Vision Transformer (ViT)-based architectures, in particular, have demonstrated strong feature extraction capabilities when pre-trained on large-scale datasets. In this work, we introduce the Magnetic Resonance Image Processing Transformer (MR-IPT), a ViT-based framework designed to enhance the generalizability and robustness of accelerated MRI reconstruction. Unlike conventional deep learning models that require separate training for different acceleration factors, MR-IPT is pre-trained on a large-scale dataset encompassing multiple undersampling patterns and acceleration settings, enabling a unified reconstruction framework. By leveraging a shared transformer backbone, MR-IPT effectively learns universal feature representations, allowing it to generalize across diverse reconstruction tasks. Extensive experiments demonstrate that MR-IPT outperforms both CNN-based and existing transformer-based methods, achieving superior reconstruction quality across varying acceleration factors and sampling masks. Moreover, MR-IPT exhibits strong robustness, maintaining high performance even under unseen acquisition setups, highlighting its potential as a scalable and efficient solution for accelerated MRI. Our findings suggest that transformer-based general models can significantly advance MRI reconstruction, offering improved adaptability and stability compared to traditional deep learning approaches.
Authors: Luca Arnaboldi, Yatin Dandi, Florent Krzakala, Luca Pesce, Ludovic Stephan
Abstract: Neural networks can identify low-dimensional relevant structures within high-dimensional noisy data, yet our mathematical understanding of how they do so remains scarce. Here, we investigate the training dynamics of two-layer shallow neural networks trained with gradient-based algorithms, and discuss how they learn pertinent features in multi-index models, that is target functions with low-dimensional relevant directions. In the high-dimensional regime, where the input dimension $d$ diverges, we show that a simple modification of the idealized single-pass gradient descent training scenario, where data can now be repeated or iterated upon twice, drastically improves its computational efficiency. In particular, it surpasses the limitations previously believed to be dictated by the Information and Leap exponents associated with the target function to be learned. Our results highlight the ability of networks to learn relevant structures from data alone without any pre-processing. More precisely, we show that (almost) all directions are learned with at most $O(d \log d)$ steps. Among the exceptions is a set of hard functions that includes sparse parities. In the presence of coupling between directions, however, these can be learned sequentially through a hierarchical mechanism that generalizes the notion of staircase functions. Our results are proven by a rigorous study of the evolution of the relevant statistics for high-dimensional dynamics.
Authors: Xiyao Wang, Jiuhai Chen, Zhaoyang Wang, Yuhang Zhou, Yiyang Zhou, Huaxiu Yao, Tianyi Zhou, Tom Goldstein, Parminder Bhatia, Furong Huang, Cao Xiao
Abstract: Large vision-language models (LVLMs) have achieved impressive results in visual question-answering and reasoning tasks through vision instruction tuning on specific datasets. However, there remains significant room for improvement in aligning visual and language modalities. Existing methods often depend on external models or data, leading to uncontrollable and unstable alignment results. In this paper, we propose SIMA, a self-improvement framework that enhances visual and language modality alignment without external dependencies. SIMA leverages existing vision instruction tuning datasets to self-generate responses, incorporating an in-context self-critic mechanism that constructs preference pairs for tuning. Crucially, our approach allows LVLMs to act as critics by designing effective critic prompts, eliminating the need for additional fine-tuning with external instruction data. We introduce three novel visual metrics within the self-critic process to guide judgment, significantly improving the accuracy of self-critic. Through extensive experiments across 14 hallucination and comprehensive benchmarks, we demonstrate that SIMA significantly improves LVLM's performance and outperforms previous approaches, achieving superior modality alignment.
Authors: Olena Burda-Lassen, Aman Chadha, Shashank Goswami, Vinija Jain
Abstract: An image is often considered worth a thousand words, and certain images can tell rich and insightful stories. Can these stories be told via image captioning? Images from folklore genres, such as mythology, folk dance, cultural signs, and symbols, are vital to every culture. Our research compares the performance of four popular vision-language models (GPT-4V, Gemini Pro Vision, LLaVA, and OpenFlamingo) in identifying culturally specific information in such images and creating accurate and culturally sensitive image captions. We also propose a new evaluation metric, the Cultural Awareness Score (CAS), which measures the degree of cultural awareness in image captions. We provide a dataset MOSAIC-1.5k labeled with ground truth for images containing cultural background and context and a labeled dataset with assigned Cultural Awareness Scores that can be used with unseen data. Creating culturally appropriate image captions is valuable for scientific research and can be beneficial for many practical applications. We envision our work will promote a deeper integration of cultural sensitivity in AI applications worldwide. By making the dataset and Cultural Awareness Score available to the public, we aim to facilitate further research in this area, encouraging the development of more culturally aware AI systems that respect and celebrate global diversity.
Authors: Hiroshi Takahashi, Tomoharu Iwata, Atsutoshi Kumagai, Yuuki Yamanaka
Abstract: Semi-supervised anomaly detection, which aims to improve the anomaly detection performance by using a small amount of labeled anomaly data in addition to unlabeled data, has attracted attention. Existing semi-supervised approaches assume that most unlabeled data are normal, and train anomaly detectors by minimizing the anomaly scores for the unlabeled data while maximizing those for the labeled anomaly data. However, in practice, the unlabeled data are often contaminated with anomalies. This weakens the effect of maximizing the anomaly scores for anomalies, and prevents us from improving the detection performance. To solve this problem, we propose the deep positive-unlabeled anomaly detection framework, which integrates positive-unlabeled learning with deep anomaly detection models such as autoencoders and deep support vector data descriptions. Our approach enables the approximation of anomaly scores for normal data using the unlabeled data and the labeled anomaly data. Therefore, without labeled normal data, our approach can train anomaly detectors by minimizing the anomaly scores for normal data while maximizing those for the labeled anomaly data. Experiments on various datasets show that our approach achieves better detection performance than existing approaches.
Authors: Mingyuan Zhou, Zhendong Wang, Huangjie Zheng, Hai Huang
Abstract: Diffusion-based text-to-image generation models trained on extensive text-image pairs have demonstrated the ability to produce photorealistic images aligned with textual descriptions. However, a significant limitation of these models is their slow sample generation process, which requires iterative refinement through the same network. To overcome this, we introduce a data-free guided distillation method that enables the efficient distillation of pretrained Stable Diffusion models without access to the real training data, often restricted due to legal, privacy, or cost concerns. This method enhances Score identity Distillation (SiD) with Long and Short Classifier-Free Guidance (LSG), an innovative strategy that applies Classifier-Free Guidance (CFG) not only to the evaluation of the pretrained diffusion model but also to the training and evaluation of the fake score network. We optimize a model-based explicit score matching loss using a score-identity-based approximation alongside our proposed guidance strategies for practical computation. By exclusively training with synthetic images generated by its one-step generator, our data-free distillation method rapidly improves FID and CLIP scores, achieving state-of-the-art FID performance while maintaining a competitive CLIP score. Notably, the one-step distillation of Stable Diffusion 1.5 achieves an FID of 8.15 on the COCO-2014 validation set, a record low value under the data-free setting. Our code and checkpoints are available at https://github.com/mingyuanzhou/SiD-LSG.
Authors: Yuxuan Wang, Jinchao Zhu, Feng Dong, Shuyue Zhu
Abstract: Audio and visual signals typically occur simultaneously, and humans possess an innate ability to correlate and synchronize information from these two modalities. Recently, a challenging problem known as Audio-Visual Segmentation (AVS) has emerged, intending to produce segmentation maps for sounding objects within a scene. However, the methods proposed so far have not sufficiently integrated audio and visual information, and the computational costs have been extremely high. Additionally, the outputs of different stages have not been fully utilized. To facilitate this research, we introduce a novel Progressive Confident Masking Attention Network (PMCANet). It leverages attention mechanisms to uncover the intrinsic correlations between audio signals and visual frames. Furthermore, we design an efficient and effective cross-attention module to enhance semantic perception by selecting query tokens. This selection is determined through confidence-driven units based on the network's multi-stage predictive outputs. Experiments demonstrate that our network outperforms other AVS methods while requiring less computational resources. The code is available at: https://github.com/PrettyPlate/PCMANet.
Authors: Harry Li, Gabriel Appleby, Ashley Suh
Abstract: We present LinkQ, a system that leverages a large language model (LLM) to facilitate knowledge graph (KG) query construction through natural language question-answering. Traditional approaches often require detailed knowledge of a graph querying language, limiting the ability for users -- even experts -- to acquire valuable insights from KGs. LinkQ simplifies this process by implementing a multistep protocol in which the LLM interprets a user's question, then systematically converts it into a well-formed query. LinkQ helps users iteratively refine any open-ended questions into precise ones, supporting both targeted and exploratory analysis. Further, LinkQ guards against the LLM hallucinating outputs by ensuring users' questions are only ever answered from ground truth KG data. We demonstrate the efficacy of LinkQ through a qualitative study with five KG practitioners. Our results indicate that practitioners find LinkQ effective for KG question-answering, and desire future LLM-assisted exploratory data analysis systems.
Authors: Behrad Moniri, Hamed Hassani, Edgar Dobriban
Abstract: Large Language Models (LLMs) are rapidly evolving and impacting various fields, necessitating the development of effective methods to evaluate and compare their performance. Most current approaches for performance evaluation are either based on fixed, domain-specific questions that lack the flexibility required in many real-world applications, or rely on human input, making them unscalable. To address these issues, we propose an automated benchmarking framework based on debates between LLMs, judged by another LLM. This method assesses not only domain knowledge, but also skills such as argumentative reasoning and inconsistency recognition. We evaluate the performance of various state-of-the-art LLMs using the debate framework and achieve rankings that align closely with popular rankings based on human input, eliminating the need for costly human crowdsourcing.
Authors: Alireza Ganjdanesh, Reza Shirkavand, Shangqian Gao, Heng Huang
Abstract: Text-to-image (T2I) diffusion models have demonstrated impressive image generation capabilities. Still, their computational intensity prohibits resource-constrained organizations from deploying T2I models after fine-tuning them on their internal target data. While pruning techniques offer a potential solution to reduce the computational burden of T2I models, static pruning methods use the same pruned model for all input prompts, overlooking the varying capacity requirements of different prompts. Dynamic pruning addresses this issue by utilizing a separate sub-network for each prompt, but it prevents batch parallelism on GPUs. To overcome these limitations, we introduce Adaptive Prompt-Tailored Pruning (APTP), a novel prompt-based pruning method designed for T2I diffusion models. Central to our approach is a prompt router model, which learns to determine the required capacity for an input text prompt and routes it to an architecture code, given a total desired compute budget for prompts. Each architecture code represents a specialized model tailored to the prompts assigned to it, and the number of codes is a hyperparameter. We train the prompt router and architecture codes using contrastive learning, ensuring that similar prompts are mapped to nearby codes. Further, we employ optimal transport to prevent the codes from collapsing into a single one. We demonstrate APTP's effectiveness by pruning Stable Diffusion (SD) V2.1 using CC3M and COCO as target datasets. APTP outperforms the single-model pruning baselines in terms of FID, CLIP, and CMMD scores. Our analysis of the clusters learned by APTP reveals they are semantically meaningful. We also show that APTP can automatically discover previously empirically found challenging prompts for SD, e.g. prompts for generating text images, assigning them to higher capacity codes.
Authors: Patrik Reizinger, Siyuan Guo, Ferenc Husz\'ar, Bernhard Sch\"olkopf, Wieland Brendel
Abstract: Identifying latent representations or causal structures is important for good generalization and downstream task performance. However, both fields have been developed rather independently. We observe that several methods in both representation and causal structure learning rely on the same data-generating process (DGP), namely, exchangeable but not i.i.d. (independent and identically distributed) data. We provide a unified framework, termed Identifiable Exchangeable Mechanisms (IEM), for representation and structure learning under the lens of exchangeability. IEM provides new insights that let us relax the necessary conditions for causal structure identification in exchangeable non--i.i.d. data. We also demonstrate the existence of a duality condition in identifiable representation learning, leading to new identifiability results. We hope this work will pave the way for further research in causal representation learning.
Authors: Zhuoxiao Chen, Junjie Meng, Mahsa Baktashmotlagh, Yonggang Zhang, Zi Huang, Yadan Luo
Abstract: LiDAR-based 3D object detection is crucial for various applications but often experiences performance degradation in real-world deployments due to domain shifts. While most studies focus on cross-dataset shifts, such as changes in environments and object geometries, practical corruptions from sensor variations and weather conditions remain underexplored. In this work, we propose a novel online test-time adaptation framework for 3D detectors that effectively tackles these shifts, including a challenging cross-corruption scenario where cross-dataset shifts and corruptions co-occur. By leveraging long-term knowledge from previous test batches, our approach mitigates catastrophic forgetting and adapts effectively to diverse shifts. Specifically, we propose a Model Synergy (MOS) strategy that dynamically selects historical checkpoints with diverse knowledge and assembles them to best accommodate the current test batch. This assembly is directed by our proposed Synergy Weights (SW), which perform a weighted averaging of the selected checkpoints, minimizing redundancy in the composite model. The SWs are computed by evaluating the similarity of predicted bounding boxes on the test data and the independence of features between checkpoint pairs in the model bank. To maintain an efficient and informative model bank, we discard checkpoints with the lowest average SW scores, replacing them with newly updated models. Our method was rigorously tested against existing test-time adaptation strategies across three datasets and eight types of corruptions, demonstrating superior adaptability to dynamic scenes and conditions. Notably, it achieved a 67.3% improvement in a challenging cross-corruption scenario, offering a more comprehensive benchmark for adaptation. Source code: https://github.com/zhuoxiao-chen/MOS.
Authors: Yingfang Yuan, Kefan Chen, Mehdi Rizvi, Lynne Baillie, Wei Pang
Abstract: The growing interest in fair AI development is evident. The ''Leave No One Behind'' initiative urges us to address multiple and intersecting forms of inequality in accessing services, resources, and opportunities, emphasising the significance of fairness in AI. This is particularly relevant as an increasing number of AI tools are applied to decision-making processes, such as resource allocation and service scheme development, across various sectors such as health, energy, and housing. Therefore, exploring joint inequalities in these sectors is significant and valuable for thoroughly understanding overall inequality and unfairness. This research introduces an innovative approach to quantify cross-sectoral intersecting discrepancies among user-defined groups using latent class analysis. These discrepancies can be used to approximate inequality and provide valuable insights to fairness issues. We validate our approach using both proprietary and public datasets, including both EVENS and Census 2021 (England & Wales) datasets, to examine cross-sectoral intersecting discrepancies among different ethnic groups. We also verify the reliability of the quantified discrepancy by conducting a correlation analysis with a government public metric. Our findings reveal significant discrepancies both among minority ethnic groups and between minority ethnic groups and non-minority ethnic groups, emphasising the need for targeted interventions in policy-making processes. Furthermore, we demonstrate how the proposed approach can provide valuable insights into ensuring fairness in machine learning systems.
Authors: Ziheng Chen, Yue Song, Xiao-Jun Wu, Gaowen Liu, Nicu Sebe
Abstract: Global Covariance Pooling (GCP) has been demonstrated to improve the performance of Deep Neural Networks (DNNs) by exploiting second-order statistics of high-level representations. GCP typically performs classification of the covariance matrices by applying matrix function normalization, such as matrix logarithm or power, followed by a Euclidean classifier. However, covariance matrices inherently lie in a Riemannian manifold, known as the Symmetric Positive Definite (SPD) manifold. The current literature does not provide a satisfactory explanation of why Euclidean classifiers can be applied directly to Riemannian features after the normalization of the matrix power. To mitigate this gap, this paper provides a comprehensive and unified understanding of the matrix logarithm and power from a Riemannian geometry perspective. The underlying mechanism of matrix functions in GCP is interpreted from two perspectives: one based on tangent classifiers (Euclidean classifiers on the tangent space) and the other based on Riemannian classifiers. Via theoretical analysis and empirical validation through extensive experiments on fine-grained and large-scale visual classification datasets, we conclude that the working mechanism of the matrix functions should be attributed to the Riemannian classifiers they implicitly respect. The code is available at https://github.com/GitZH-Chen/RiemGCP.git.
Authors: Armand Nicolicioiu, Eugenia Iofinova, Andrej Jovanovic, Eldar Kurtic, Mahdi Nikdan, Andrei Panferov, Ilia Markov, Nir Shavit, Dan Alistarh
Abstract: The availability of powerful open-source large language models (LLMs) opens exciting use-cases, such as using personal data to fine-tune these models to imitate a user's unique writing style. Two key requirements for such assistants are personalization - in the sense that the assistant should recognizably reflect the user's own writing style - and privacy - users may justifiably be wary of uploading extremely personal data, such as their email archive, to a third-party service. In this paper, we present a new design and evaluation for such an automated assistant, for the specific use case of email generation, which we call Panza. Panza's personalization features are based on a combination of fine-tuning using a variant of the Reverse Instructions technique together with Retrieval-Augmented Generation (RAG). We demonstrate that this combination allows us to fine-tune an LLM to reflect a user's writing style using limited data, while executing on extremely limited resources, e.g. on a free Google Colab instance. Our key methodological contribution is the first detailed study of evaluation metrics for this personalized writing task, and of how different choices of system components--the use of RAG and of different fine-tuning approaches-impact the system's performance. Additionally, we demonstrate that very little data - under 100 email samples - are sufficient to create models that convincingly imitate humans. This finding showcases a previously-unknown attack vector in language models - that access to a small number of writing samples can allow a bad actor to cheaply create generative models that imitate a target's writing style. We are releasing the full Panza code as well as three new email datasets licensed for research use at https://github.com/IST-DASLab/PanzaMail.
Authors: Jack Cunningham, Jun Zhuang
Abstract: In recent years, variational quantum circuits (VQCs) have been widely explored to advance quantum circuits against classic models on various domains, such as quantum chemistry and quantum machine learning. Similar to classic machine-learning models, VQCs can be trained through various optimization approaches, such as gradient-based or gradient-free methods. However, when employing gradient-based methods, the gradient variance of VQCs may dramatically vanish as the number of qubits or layers increases. This issue, a.k.a. Barren Plateaus (BPs), seriously hinders the scaling of VQCs on large datasets. To mitigate the barren plateaus, extensive efforts have been devoted to tackling this issue through diverse strategies. In this survey, we conduct a systematic literature review of recent works from both investigation and mitigation perspectives. Furthermore, we propose a new taxonomy to categorize most existing mitigation strategies into five groups and introduce them in detail. Also, we compare the concurrent survey papers about BPs. Finally, we provide insightful discussion on future directions for BPs.
Authors: Timothy Dai, Kate Maher, Zach Perzan
Abstract: Process-based hydrologic models are invaluable tools for understanding the terrestrial water cycle and addressing modern water resources problems. However, many hydrologic models are computationally expensive and, depending on the resolution and scale, simulations can take on the order of hours to days to complete. While techniques such as uncertainty quantification and optimization have become valuable tools for supporting management decisions, these analyses typically require hundreds of model simulations, which are too computationally expensive to perform with a process-based hydrologic model. To address this gap, we assess a hybrid modeling workflow in which a process-based model is used to generate an initial set of simulations and a machine learning (ML) surrogate model is then trained to perform the remaining simulations required for downstream analysis. As a case study, we apply this workflow to simulations of variably saturated groundwater flow at a prospective managed aquifer recharge site. We compare the accuracy and computational efficiency of several ML architectures, including deep convolutional networks, recurrent neural networks, vision transformers, and networks with Fourier transforms. Our results demonstrate that ML surrogate models can achieve under 10% mean absolute percentage error and yield order-of-magnitude runtime savings over process-based models. Building on these findings, we examine the impacts of key modeling choices on surrogate model accuracy and efficiency. Results show that a normalized loss function improves training stability, while min-max data normalization can significantly reduce error up to a factor of 10 when compared to other treatments such as Z-score and no normalization. Downsampling input features using an autoencoder also decreases memory requirements by training with tensors 4% their original size. By reducing computational costs and...
Authors: Takahiro Kawashima, Hideitsu Hino
Abstract: Positive and negative dependence are fundamental concepts that characterize the attractive and repulsive behavior of random subsets. Although some probabilistic models are known to exhibit positive or negative dependence, it is challenging to seamlessly bridge them with a practicable probabilistic model. In this study, we introduce a new family of distributions, named the discrete kernel point process (DKPP), which includes determinantal point processes and parts of Boltzmann machines. We also develop some computational methods for probabilistic operations and inference with DKPPs, such as calculating marginal and conditional probabilities and learning the parameters. Our numerical experiments demonstrate the controllability of positive and negative dependence and the effectiveness of the computational methods for DKPPs.
Authors: Michele Miranda, Elena Sofia Ruzzetti, Andrea Santilli, Fabio Massimo Zanzotto, S\'ebastien Brati\`eres, Emanuele Rodol\`a
Abstract: Large Language Models (LLMs) represent a significant advancement in artificial intelligence, finding applications across various domains. However, their reliance on massive internet-sourced datasets for training brings notable privacy issues, which are exacerbated in critical domains (e.g., healthcare). Moreover, certain application-specific scenarios may require fine-tuning these models on private data. This survey critically examines the privacy threats associated with LLMs, emphasizing the potential for these models to memorize and inadvertently reveal sensitive information. We explore current threats by reviewing privacy attacks on LLMs and propose comprehensive solutions for integrating privacy mechanisms throughout the entire learning pipeline. These solutions range from anonymizing training datasets to implementing differential privacy during training or inference and machine unlearning after training. Our comprehensive review of existing literature highlights ongoing challenges, available tools, and future directions for preserving privacy in LLMs. This work aims to guide the development of more secure and trustworthy AI systems by providing a thorough understanding of privacy preservation methods and their effectiveness in mitigating risks.
Authors: Jiarui Xie, Mutahar Safdar, Lequn Chen, Seung Ki Moon, Yaoyao Fiona Zhao
Abstract: Various machine learning (ML)-based in-situ monitoring systems have been developed to detect anomalies and defects in laser additive manufacturing (LAM) processes. While multimodal fusion, which integrates data from visual, audio, and other modalities, can improve monitoring performance, it also increases hardware, computational, and operational costs. This paper introduces a cross-modality knowledge transfer (CMKT) methodology for LAM in-situ monitoring, which transfers knowledge from a source modality to a target modality. CMKT enhances the representativeness of the features extracted from the target modality, allowing the removal of source modality sensors during prediction. This paper proposes three CMKT methods: semantic alignment, fully supervised mapping, and semi-supervised mapping. The semantic alignment method establishes a shared encoded space between modalities to facilitate knowledge transfer. It employs a semantic alignment loss to align the distributions of identical groups (e.g., visual and audio defective groups) and a separation loss to distinguish different groups (e.g., visual defective and audio defect-free groups). The two mapping methods transfer knowledge by deriving features from one modality to another using fully supervised and semi-supervised learning approaches. In a case study for LAM in-situ defect detection, the proposed CMKT methods were compared with multimodal audio-visual fusion. The semantic alignment method achieved an accuracy of 98.6% while removing the audio modality during the prediction phase, which is comparable to the 98.2% accuracy obtained through multimodal fusion. Using explainable artificial intelligence, we discovered that semantic alignment CMKT can extract more representative features while reducing noise by leveraging the inherent correlations between modalities.
Authors: Hongseok Oh, Wonseok Hwang
Abstract: Large Language Models (LLMs) have shown remarkable progress, but their real-world application necessitates reliable calibration. This study conducts a comprehensive analysis of calibration degradation of LLMs across four dimensions: models, calibration metrics, tasks, and confidence extraction methods. Initial analysis showed that the relationship between alignment and calibration is not always a trade-off, but under stricter analysis conditions, we found the alignment process consistently harms calibration. This highlights the need for (1) a careful approach when measuring model confidences and calibration errors and (2) future research into algorithms that can help LLMs to achieve both instruction-following and calibration without sacrificing either.
Authors: Yujie Wang, Shenhan Zhu, Fangcheng Fu, Xupeng Miao, Jie Zhang, Juan Zhu, Fan Hong, Yong Li, Bin Cui
Abstract: Recent foundation models are capable of handling multiple tasks and multiple data modalities with the unified base model structure and several specialized model components. However, efficient training of such multi-task (MT) multi-modal (MM) models poses significant system challenges due to the sophisticated model architecture and the heterogeneous workloads of different tasks and modalities. In this paper, we propose Spindle, a brand new training system tailored for resource-efficient and high-performance training of MT MM models via wavefront scheduling. The key idea of Spindle is to decompose the model execution into waves and address the joint optimization problem sequentially, including both heterogeneity-aware workload parallelization and dependency-driven execution scheduling. We build our system and evaluate it on various MT MM models. Experiments demonstrate the superior performance and efficiency of Spindle, with speedup ratio up to 71% compared to state-of-the-art training systems.
Authors: Valentina Vadori, Jean-Marie Gra\"ic, Antonella Peruffo, Giulia Vadori, Livio Finos, Enrico Grisan
Abstract: Delineating and classifying individual cells in microscopy tissue images is inherently challenging yet remains essential for advancements in medical and neuroscientific research. In this work, we propose a new deep learning framework, CISCA, for automatic cell instance segmentation and classification in histological slices. At the core of CISCA is a network architecture featuring a lightweight U-Net with three heads in the decoder. The first head classifies pixels into boundaries between neighboring cells, cell bodies, and background, while the second head regresses four distance maps along four directions. The outputs from the first and second heads are integrated through a tailored post-processing step, which ultimately produces the segmentation of individual cells. The third head enables the simultaneous classification of cells into relevant classes, if required. We demonstrate the effectiveness of our method using four datasets, including CoNIC, PanNuke, and MoNuSeg, which are publicly available H&Estained datasets that cover diverse tissue types and magnifications. In addition, we introduce CytoDArk0, the first annotated dataset of Nissl-stained histological images of the mammalian brain, containing nearly 40k annotated neurons and glia cells, aimed at facilitating advancements in digital neuropathology and brain cytoarchitecture studies. We evaluate CISCA against other state-of-the-art methods, demonstrating its versatility, robustness, and accuracy in segmenting and classifying cells across diverse tissue types, magnifications, and staining techniques. This makes CISCA well-suited for detailed analyses of cell morphology and efficient cell counting in both digital pathology workflows and brain cytoarchitecture research.
Authors: Tushar Dhyani, Florian Lux, Michele Mancusi, Giorgio Fabbro, Fritz Hohl, Ngoc Thang Vu
Abstract: Traditional speech enhancement methods often oversimplify the task of restoration by focusing on a single type of distortion. Generative models that handle multiple distortions frequently struggle with phone reconstruction and high-frequency harmonics, leading to breathing and gasping artifacts that reduce the intelligibility of reconstructed speech. These models are also computationally demanding, and many solutions are restricted to producing outputs in the wide-band frequency range, which limits their suitability for professional applications. To address these challenges, we propose Hi-ResLDM, a novel generative model based on latent diffusion designed to remove multiple distortions and restore speech recordings to studio quality, sampled at 48kHz. We benchmark Hi-ResLDM against state-of-the-art methods that leverage GAN and Conditional Flow Matching (CFM) components, demonstrating superior performance in regenerating high-frequency-band details. Hi-ResLDM not only excels in non-instrusive metrics but is also consistently preferred in human evaluation and performs competitively on intrusive evaluations, making it ideal for high-resolution speech restoration.
Authors: Charilaos Papaioannou, Emmanouil Benetos, Alexandros Potamianos
Abstract: We introduce Label-Combination Prototypical Networks (LC-Protonets) to address the problem of multi-label few-shot classification, where a model must generalize to new classes based on only a few available examples. Extending Prototypical Networks, LC-Protonets generate one prototype per label combination, derived from the power set of labels present in the limited training items, rather than one prototype per label. Our method is applied to automatic audio tagging across diverse music datasets, covering various cultures and including both modern and traditional music, and is evaluated against existing approaches in the literature. The results demonstrate a significant performance improvement in almost all domains and training setups when using LC-Protonets for multi-label classification. In addition to training a few-shot learning model from scratch, we explore the use of a pre-trained model, obtained via supervised learning, to embed items in the feature space. Fine-tuning improves the generalization ability of all methods, yet LC-Protonets achieve high-level performance even without fine-tuning, in contrast to the comparative approaches. We finally analyze the scalability of the proposed method, providing detailed quantitative metrics from our experiments. The implementation and experimental setup are made publicly available, offering a benchmark for future research.
Authors: Prateek Verma
Abstract: Large Language Models (LLMs) have ushered in a new wave of artificial intelligence advancements impacting every scientific field and discipline. We live in a world where most of the data around us, e.g., text, audio, and music, has a multi-scale structure. This paper infuses LLMs with a traditional signal processing idea, namely wavelets, during pre-training to take advantage of the structure. Without adding \textbf{any extra parameters} to a GPT-style LLM architecture in an academic setup, we achieve the same pre-training performance almost twice as fast in text, audio, and images. This is done by imposing a structure on intermediate embeddings. When trained for the same number of training steps, we achieve significant gains in performance, which is comparable to pre-training a larger neural architecture. Further, we show this extends to the Long Range Arena benchmark and several input representations such as characters, BPE tokens, bytes, waveform, math expression, and image pixels. Our architecture allows every next token prediction access to intermediate embeddings at different temporal resolutions in every decoder block. We hope this will pave the way for incorporating multi-rate signal processing into pre-training.
Authors: Yiming Zhong, Yinuo Ren, Guangyao Cao, Feng Li, Haobo Qi
Abstract: Recent advances on time series forecasting mainly focus on improving the forecasting models themselves. However, when the time series data suffer from potential structural breaks or concept drifts, the forecasting performance might be significantly reduced. In this paper, we introduce a novel approach called Optimal Starting Point Time Series Forecast (OSP-TSP) for optimal forecasting, which can be combined with existing time series forecasting models. By adjusting the sequence length via leveraging the XGBoost and LightGBM models, the proposed approach can determine the optimal starting point (OSP) of the time series and then enhance the prediction performances of the base forecasting models. To illustrate the effectiveness of the proposed approach, comprehensive empirical analysis have been conducted on the M4 dataset and other real world datasets. Empirical results indicate that predictions based on the OSP-TSP approach consistently outperform those using the complete time series dataset. Moreover, comparison results reveals that combining our approach with existing forecasting models can achieve better prediction accuracy, which also reflect the advantages of the proposed approach.
Authors: Sathya Krishnan Suresh, Wu Mengjun, Tushar Pranav, Eng Siong Chng
Abstract: The scarcity of domain-specific dialogue datasets limits the development of dialogue systems across applications. Existing research is constrained by general or niche datasets that lack sufficient scale for training dialogue systems. To address this gap, we introduce DiaSynth - a synthetic dialogue generation framework capable of generating high-quality, contextually rich dialogues across a wide range of domains. Unlike existing frameworks, DiaSynth uses Large Language Models (LLMs) and Chain of Thought (CoT) reasoning to generate dynamic, domain-specific dialogues with simulated personas and diverse conversational features. We perform our experiments by generating synthetic data using different LLMs and few-shot examples from DialogSum and SAMSum. The pretrained language models fine-tuned on the synthetic data outperform the base models by 16.47% on dialogue summarization, while the comparison between models fine-tuned on in-domain data and synthetic data shows that the synthetic data is able to capture 90.48% of the performance distribution of the in-domain data on dialogue summarization. The quality of the data generated also increases as we increase the size of LLM from 3B to 8B. These results validate DiaSynth's potential as a robust alternative to traditional data collection methods. We open source the code and data generated for future research.
Authors: Ankit Bhardwaj, Ananth Balashankar, Shiva Iyer, Nita Soans, Anant Sudarshan, Rohini Pande, Lakshminarayanan Subramanian
Abstract: Urban air pollution hotspots pose significant health risks, yet their detection and analysis remain limited by the sparsity of public sensor networks. This paper addresses this challenge by combining predictive modeling and mechanistic approaches to comprehensively monitor pollution hotspots. We enhanced New Delhi's existing sensor network with 28 low-cost sensors, collecting PM2.5 data over 30 months from May 1, 2018, to Nov 1, 2020. Applying established definitions of hotspots to this data, we found the existence of additional 189 hidden hotspots apart from confirming 660 hotspots detected by the public network. Using predictive techniques like Space-Time Kriging, we identified hidden hotspots with 95% precision and 88% recall with 50% sensor failure rate, and with 98% precision and 95% recall with 50% missing sensors. The projected results of our predictive models were further compiled into policy recommendations for public authorities. Additionally, we developed a Gaussian Plume Dispersion Model to understand the mechanistic underpinnings of hotspot formation, incorporating an emissions inventory derived from local sources. Our mechanistic model is able to explain 65% of observed transient hotspots. Our findings underscore the importance of integrating data-driven predictive models with physics-based mechanistic models for scalable and robust air pollution management in resource-constrained settings.
Authors: Yi Ding, Bolian Li, Ruqi Zhang
Abstract: Vision Language Models (VLMs) have become essential backbones for multimodal intelligence, yet significant safety challenges limit their real-world application. While textual inputs are often effectively safeguarded, adversarial visual inputs can easily bypass VLM defense mechanisms. Existing defense methods are either resource-intensive, requiring substantial data and compute, or fail to simultaneously ensure safety and usefulness in responses. To address these limitations, we propose a novel two-phase inference-time alignment framework, Evaluating Then Aligning (ETA): 1) Evaluating input visual contents and output responses to establish a robust safety awareness in multimodal settings, and 2) Aligning unsafe behaviors at both shallow and deep levels by conditioning the VLMs' generative distribution with an interference prefix and performing sentence-level best-of-N to search the most harmless and helpful generation paths. Extensive experiments show that ETA outperforms baseline methods in terms of harmlessness, helpfulness, and efficiency, reducing the unsafe rate by 87.5% in cross-modality attacks and achieving 96.6% win-ties in GPT-4 helpfulness evaluation. The code is publicly available at https://github.com/DripNowhy/ETA.
Authors: Brandon L. Annesi, Enrico M. Malatesta, Francesco Zamponi
Abstract: We analyze the problem of storing random pattern-label associations using two classes of continuous non-convex weights models, namely the perceptron with negative margin and an infinite-width two-layer neural network with non-overlapping receptive fields and generic activation function. Using a full-RSB ansatz we compute the exact value of the SAT/UNSAT transition. Furthermore, in the case of the negative perceptron we show that the overlap distribution of typical states displays an overlap gap (a disconnected support) in certain regions of the phase diagram defined by the value of the margin and the density of patterns to be stored. This implies that some recent theorems that ensure convergence of Approximate Message Passing (AMP) based algorithms to capacity are not applicable. Finally, we show that Gradient Descent is not able to reach the maximal capacity, irrespectively of the presence of an overlap gap for typical states. This finding, similarly to what occurs in binary weight models, suggests that gradient-based algorithms are biased towards highly atypical states, whose inaccessibility determines the algorithmic threshold.
Authors: Fanmeng Wang, Minjie Cheng, Hongteng Xu
Abstract: Predicting molecular ground-state conformation (i.e., energy-minimized conformation) is crucial for many chemical applications such as molecular docking and property prediction. Classic energy-based simulation is time-consuming when solving this problem while existing learning-based methods have advantages in computational efficiency but sacrifice accuracy and interpretability. In this work, we propose a novel and effective method to bridge the energy-based simulation and the learning-based strategy, which designs and learns a Wasserstein gradient flow-driven SE(3)-Transformer, called WGFormer, for molecular ground-state conformation prediction. Specifically, our method tackles this task within an auto-encoding framework, which encodes low-quality conformations by the proposed WGFormer and decodes corresponding ground-state conformations by an MLP. The architecture of WGFormer corresponds to Wasserstein gradient flows -- it optimizes molecular conformations by minimizing an energy function defined on the latent mixture models of atoms, thereby significantly improving performance and interpretability. Extensive experiments show that our method consistently outperforms state-of-the-art competitors, providing a new and insightful paradigm to predict molecular ground-state conformation.
Authors: Achint Soni, Sreyas Venkataraman, Abhranil Chandra, Sebastian Fischmeister, Percy Liang, Bo Dai, Sherry Yang
Abstract: Video generation has been used to generate visual plans for controlling robotic systems. Given an image observation and a language instruction, previous work has generated video plans which are then converted to robot controls to be executed. However, a major bottleneck in leveraging video generation for control lies in the quality of the generated videos, which often suffer from hallucinatory content and unrealistic physics, resulting in low task success when control actions are extracted from the generated videos. While scaling up dataset and model size provides a partial solution, integrating external feedback is both natural and essential for grounding video generation in the real world. With this observation, we propose VideoAgent for self-improving generated video plans based on external feedback. Instead of directly executing the generated video plan, VideoAgent first refines the generated video plans using a novel procedure which we call self-conditioning consistency, allowing inference-time compute to be turned into better generated video plans. As the refined video plan is being executed, VideoAgent can collect additional data from the environment to further improve video plan generation. Experiments in simulated robotic manipulation from MetaWorld and iTHOR show that VideoAgent drastically reduces hallucination, thereby boosting success rate of downstream manipulation tasks. We further illustrate that VideoAgent can effectively refine real-robot videos, providing an early indicator that robots can be an effective tool in grounding video generation in the physical world. Video demos and code can be found at https://video-as-agent.github.io.
Authors: Ajay Patel, Jiacheng Zhu, Justin Qiu, Zachary Horvitz, Marianna Apidianaki, Kathleen McKeown, Chris Callison-Burch
Abstract: Style representations aim to embed texts with similar writing styles closely and texts with different styles far apart, regardless of content. However, the contrastive triplets often used for training these representations may vary in both style and content, leading to potential content leakage in the representations. We introduce StyleDistance, a novel approach to training stronger content-independent style embeddings. We use a large language model to create a synthetic dataset of near-exact paraphrases with controlled style variations, and produce positive and negative examples across 40 distinct style features for precise contrastive learning. We assess the quality of our synthetic data and embeddings through human and automatic evaluations. StyleDistance enhances the content-independence of style embeddings, which generalize to real-world benchmarks and outperform leading style representations in downstream applications. Our model can be found at https://huggingface.co/StyleDistance/styledistance .
Authors: Yanhao Jin, Krishnakumar Balasubramanian, Lifeng Lai
Abstract: We investigate the in-context learning capabilities of transformers for the $d$-dimensional mixture of linear regression model, providing theoretical insights into their existence, generalization bounds, and training dynamics. Specifically, we prove that there exists a transformer capable of achieving a prediction error of order $\mathcal{O}(\sqrt{d/n})$ with high probability, where $n$ represents the training prompt size in the high signal-to-noise ratio (SNR) regime. Moreover, we derive in-context excess risk bounds of order $\mathcal{O}(L/\sqrt{B})$ for the case of two mixtures, where $B$ denotes the number of training prompts, and $L$ represents the number of attention layers. The dependence of $L$ on the SNR is explicitly characterized, differing between low and high SNR settings. We further analyze the training dynamics of transformers with single linear self-attention layers, demonstrating that, with appropriately initialized parameters, gradient flow optimization over the population mean square loss converges to a global optimum. Extensive simulations suggest that transformers perform well on this task, potentially outperforming other baselines, such as the Expectation-Maximization algorithm.
Authors: Mojtaba Heydari, Mehrez Souden, Bruno Conejo, Joshua Atkins
Abstract: We introduce ImmerseDiffusion, an end-to-end generative audio model that produces 3D immersive soundscapes conditioned on the spatial, temporal, and environmental conditions of sound objects. ImmerseDiffusion is trained to generate first-order ambisonics (FOA) audio, which is a conventional spatial audio format comprising four channels that can be rendered to multichannel spatial output. The proposed generative system is composed of a spatial audio codec that maps FOA audio to latent components, a latent diffusion model trained based on various user input types, namely, text prompts, spatial, temporal and environmental acoustic parameters, and optionally a spatial audio and text encoder trained in a Contrastive Language and Audio Pretraining (CLAP) style. We propose metrics to evaluate the quality and spatial adherence of the generated spatial audio. Finally, we assess the model performance in terms of generation quality and spatial conformance, comparing the two proposed modes: ``descriptive", which uses spatial text prompts) and ``parametric", which uses non-spatial text prompts and spatial parameters. Our evaluations demonstrate promising results that are consistent with the user conditions and reflect reliable spatial fidelity.
Authors: Xu Cai, Jonathan Scarlett
Abstract: The optimization of black-box functions with noisy observations is a fundamental problem with widespread applications, and has been widely studied under the assumption that the function lies in a reproducing kernel Hilbert space (RKHS). This problem has been studied extensively in the stationary setting, and near-optimal regret bounds are known via developments in both upper and lower bounds. In this paper, we consider non-stationary scenarios, which are crucial for certain applications but are currently less well-understood. Specifically, we provide the first algorithm-independent lower bounds, where the time variations are subject satisfying a total variation budget according to some function norm. Under $\ell_{\infty}$-norm variations, our bounds are found to be close to an existing upper bound (Hong et al., 2023). Under RKHS norm variations, the upper and lower bounds are still reasonably close but with more of a gap, raising the interesting open question of whether non-minor improvements in the upper bound are possible.
Authors: Tanmay Parekh, Pradyot Prakash, Alexander Radovic, Akshay Shekher, Denis Savenkov
Abstract: Research has shown the effectiveness of reasoning (e.g., Chain-of-Thought), planning (e.g., SelfAsk), and retrieval augmented generation strategies to improve the performance of Large Language Models (LLMs) on various tasks, such as question answering. However, using a single fixed strategy to answer different kinds of questions is suboptimal in performance and inefficient in terms of generated output tokens and performed retrievals. In our work, we propose a novel technique DyPlan, to induce a dynamic strategy selection process in LLMs, to improve performance and reduce costs in question-answering. DyPlan incorporates an initial decision step to select the most suitable strategy conditioned on the input question and guides the LLM's response generation accordingly. We extend DyPlan to DyPlan-verify, adding an internal verification and correction process to further enrich the generated answer. Experiments on three prominent multi-hop question answering (MHQA) datasets reveal how DyPlan can improve model performance by 7-13% while reducing the cost by 11-32% relative to the best baseline model.
Authors: Vladimir Billig
Abstract: An efficient Apriori_Goal algorithm is proposed for constructing association rules in a relational database with predefined classification. The target parameter of the database specifies a finite number of goals $Goal_k$, for each of which the algorithm constructs association rules of the form $X \Rightarrow Goal_k$. The quality of the generated rules is characterized by five criteria: two represent rule frequency, two reflect rule reliability, and the fifth is a weighted sum of these four criteria. The algorithm initially generates rules with single premises, where the correlation criterion between the premise $X$ and the conclusion $Goal_k$ exceeds a specified threshold. Then, rules with extended premises are built based on the anti-monotonicity of rule frequency criteria and the monotonicity of rule reliability criteria. Newly constructed rules tend to decrease in frequency while increasing in reliability. The article proves several statements that justify the rule construction process. The algorithm enables the construction of both high-frequency and rare rules with low occurrence frequency but high reliability. It also allows for the generation of negative rules with negative correlation between the premise and conclusion, which can be valuable in practical applications for filtering out undesired goals. The efficiency of the algorithm is based on two factors: the method of encoding the database and its partitioning into subsets linked to the target parameter. Time complexity estimates for rule construction are provided using a medical database as an example.
Authors: Mingrui Liu, Sixiao Zhang, Cheng Long
Abstract: Sequential recommendation (SR) systems excel at capturing users' dynamic preferences by leveraging their interaction histories. Most existing SR systems assign a single embedding vector to each item to represent its features, and various types of models are adopted to combine these item embeddings into a sequence representation vector to capture the user intent. However, we argue that this representation alone is insufficient to capture an item's multi-faceted nature (e.g., movie genres, starring actors). Besides, users often exhibit complex and varied preferences within these facets (e.g., liking both action and musical films in the facet of genre), which are challenging to fully represent. To address the issues above, we propose a novel structure called Facet-Aware Multi-Head Mixture-of-Experts Model for Sequential Recommendation (FAME). We leverage sub-embeddings from each head in the last multi-head attention layer to predict the next item separately. This approach captures the potential multi-faceted nature of items without increasing model complexity. A gating mechanism integrates recommendations from each head and dynamically determines their importance. Furthermore, we introduce a Mixture-of-Experts (MoE) network in each attention head to disentangle various user preferences within each facet. Each expert within the MoE focuses on a specific preference. A learnable router network is adopted to compute the importance weight for each expert and aggregate them. We conduct extensive experiments on four public sequential recommendation datasets and the results demonstrate the effectiveness of our method over existing baseline models.
Authors: Han Wang, Binbin Chen, Tieying Zhang, Baoxiang Wang
Abstract: Effective communication is an essential component in collaborative multi-agent systems. Situations where explicit messaging is not feasible have been common in human society throughout history, which motivate the study of implicit communication. Previous works on learning implicit communication mostly rely on theory of mind (ToM), where agents infer the mental states and intentions of others by interpreting their actions. However, ToM-based methods become less effective in making accurate inferences in complex tasks. In this work, we propose the Implicit Channel Protocol (ICP) framework, which allows agents to communicate through implicit communication channels similar to the explicit ones. ICP leverages a subset of actions, denoted as the scouting actions, and a mapping between information and these scouting actions that encodes and decodes the messages. We propose training algorithms for agents to message and act, including learning with a randomly initialized information map and with a delayed information map. The efficacy of ICP has been tested on the tasks of Guessing Numbers, Revealing Goals, and Hanabi, where ICP significantly outperforms baseline methods through more efficient information transmission.
Authors: Yingqi Gao, Yifu Liu, Xiaoxia Li, Xiaorong Shi, Yin Zhu, Yiming Wang, Shiqi Li, Wei Li, Yuntao Hong, Zhiling Luo, Jinyang Gao, Liyu Mou, Yu Li
Abstract: To tackle the challenges of large language model performance in natural language to SQL tasks, we introduce XiYan-SQL, an innovative framework that employs a multi-generator ensemble strategy to improve candidate generation. We introduce M-Schema, a semi-structured schema representation method designed to enhance the understanding of database structures. To enhance the quality and diversity of generated candidate SQL queries, XiYan-SQL integrates the significant potential of in-context learning (ICL) with the precise control of supervised fine-tuning. On one hand, we propose a series of training strategies to fine-tune models to generate high-quality candidates with diverse preferences. On the other hand, we implement the ICL approach with an example selection method based on named entity recognition to prevent overemphasis on entities. The refiner optimizes each candidate by correcting logical or syntactical errors. To address the challenge of identifying the best candidate, we fine-tune a selection model to distinguish nuances of candidate SQL queries. The experimental results on multiple dialect datasets demonstrate the robustness of XiYan-SQL in addressing challenges across different scenarios. Overall, our proposed XiYan-SQL achieves the state-of-the-art execution accuracy of 75.63% on Bird benchmark, 89.65% on the Spider test set, 69.86% on SQL-Eval, 41.20% on NL2GQL. The proposed framework not only enhances the quality and diversity of SQL queries but also outperforms previous methods.
Authors: Jijie Zou, Zhanghao Zhouyin, Dongying Lin, Linfeng Zhang, Shimin Hou, Qiangqiang Gu
Abstract: Quantum transport calculations are essential for understanding and designing nanoelectronic devices, yet the trade-off between accuracy and computational efficiency has long limited their practical applications. We present a general framework that combines the deep learning tight-binding Hamiltonian (DeePTB) approach with the non-equilibrium Green's Function (NEGF) method, enabling efficient quantum transport calculations while maintaining first-principles accuracy. We demonstrate the capabilities of the DeePTB-NEGF framework through two representative applications: comprehensive simulation of break junction systems, where conductance histograms show good agreement with experimental measurements in both metallic contact and single-molecule junction cases; and simulation of carbon nanotube field effect transistors through self-consistent NEGF-Poisson calculations, capturing essential physics including the electrostatic potential and transfer characteristic curves under finite bias conditions. This framework bridges the gap between first-principles accuracy and computational efficiency, providing a powerful tool for high-throughput quantum transport simulations across different scales in nanoelectronics.
Authors: Moran Yanuka, Assaf Ben Kish, Yonatan Bitton, Idan Szpektor, Raja Giryes
Abstract: Recent research increasingly focuses on training vision-language models (VLMs) with long, detailed image captions. However, small-scale VLMs often struggle to balance the richness of these captions with the risk of hallucinating content during fine-tuning. In this paper, we explore how well VLMs adapt to such captions. To quantify caption quality, we propose Decomposed NLI (DNLI), an evaluation framework that breaks down generated captions into individual propositions, assessing each in isolation. This fine-grained analysis reveals a critical balance between capturing descriptive details and preventing hallucinations. Our findings show that simply reducing caption complexity or employing standard data curation techniques does not effectively resolve this issue. To tackle this challenge, we introduce Knowledge Adapted (KnowAda) fine-tuning, a data-centric approach that automatically adapts training data with the model's existing knowledge and visual understanding. KnowAda minimizes hallucinations while preserving high descriptiveness. We validate this approach across several small-scale VLMs (up to 7B parameters) and dense caption datasets, demonstrating that KnowAda effectively balances hallucination reduction and descriptiveness. Our results show that KnowAda outperforms various baselines in both automatic metrics and human evaluations. We will release our code and models.
Authors: Javier Ferrando, Oscar Obeso, Senthooran Rajamanoharan, Neel Nanda
Abstract: Hallucinations in large language models are a widespread problem, yet the mechanisms behind whether models will hallucinate are poorly understood, limiting our ability to solve this problem. Using sparse autoencoders as an interpretability tool, we discover that a key part of these mechanisms is entity recognition, where the model detects if an entity is one it can recall facts about. Sparse autoencoders uncover meaningful directions in the representation space, these detect whether the model recognizes an entity, e.g. detecting it doesn't know about an athlete or a movie. This suggests that models can have self-knowledge: internal representations about their own capabilities. These directions are causally relevant: capable of steering the model to refuse to answer questions about known entities, or to hallucinate attributes of unknown entities when it would otherwise refuse. We demonstrate that despite the sparse autoencoders being trained on the base model, these directions have a causal effect on the chat model's refusal behavior, suggesting that chat finetuning has repurposed this existing mechanism. Furthermore, we provide an initial exploration into the mechanistic role of these directions in the model, finding that they disrupt the attention of downstream heads that typically move entity attributes to the final token.
Authors: Yao Xu, Shizhu He, Jiabei Chen, Zeng Xiangrong, Bingning Wang, Guang Liu, Jun Zhao, Kang Liu
Abstract: Structured data, such as tables, graphs, and databases, play a critical role in plentiful NLP tasks such as question answering and dialogue system. Recently, inspired by Vision-Language Models, Graph Neutral Networks (GNNs) have been introduced as an additional modality into the input of Large Language Models (LLMs) to improve their performance on Structured Knowledge Grounding (SKG) tasks. However, those GNN-enhanced LLMs have the following limitations: (1) They employ diverse GNNs to model varying types of structured data, rendering them unable to uniformly process various forms of structured data. (2) The pretraining of GNNs is coupled with specific LLMs, which prevents GNNs from fully aligning with the textual space and limits their adaptability to other LLMs. To address these issues, we propose \textbf{L}arge \textbf{L}anguage and \textbf{S}tructured Data \textbf{A}ssistant (LLaSA), a general framework for enhancing LLMs' ability to handle structured data. Specifically, we represent various types of structured data in a unified hypergraph format, and use self-supervised learning to pretrain a hypergraph encoder, and a G-Former compressing encoded hypergraph representations with cross-attention. The compressed hypergraph representations are appended to the serialized inputs during training and inference stages of LLMs. Experimental results on multiple SKG tasks show that our pretrained hypergraph encoder can adapt to various LLMs and enhance their ability to process different types of structured data. Besides, LLaSA, with LoRA fine-tuning, outperforms previous SOTA method using full parameters tuning.
Authors: Yue Yu, Zhengxing Chen, Aston Zhang, Liang Tan, Chenguang Zhu, Richard Yuanzhe Pang, Yundi Qian, Xuewei Wang, Suchin Gururangan, Chao Zhang, Melanie Kambadur, Dhruv Mahajan, Rui Hou
Abstract: Reward modeling is crucial for aligning large language models (LLMs) with human preferences, especially in reinforcement learning from human feedback (RLHF). However, current reward models mainly produce scalar scores and struggle to incorporate critiques in a natural language format. We hypothesize that predicting both critiques and the scalar reward would improve reward modeling ability. Motivated by this, we propose Critic-RM, a framework that improves reward models using self-generated critiques without extra supervision. Critic-RM employs a two-stage process: generating and filtering high-quality critiques, followed by joint fine-tuning on reward prediction and critique generation. Experiments across benchmarks show that Critic-RM improves reward modeling accuracy by 3.7%-7.3% compared to standard reward models and LLM judges, demonstrating strong performance and data efficiency. Additional studies further validate the effectiveness of generated critiques in rectifying flawed reasoning steps with 2.5%-3.2% gains in improving reasoning accuracy.
Authors: Sathwik Karnik, Zhang-Wei Hong, Nishant Abhangi, Yen-Chen Lin, Tsun-Hsuan Wang, Christophe Dupuy, Rahul Gupta, Pulkit Agrawal
Abstract: Language-conditioned robot models have the potential to enable robots to perform a wide range of tasks based on natural language instructions. However, assessing their safety and effectiveness remains challenging because it is difficult to test all the different ways a single task can be phrased. Current benchmarks have two key limitations: they rely on a limited set of human-generated instructions, missing many challenging cases, and focus only on task performance without assessing safety, such as avoiding damage. To address these gaps, we introduce Embodied Red Teaming (ERT), a new evaluation method that generates diverse and challenging instructions to test these models. ERT uses automated red teaming techniques with Vision Language Models (VLMs) to create contextually grounded, difficult instructions. Experimental results show that state-of-the-art language-conditioned robot models fail or behave unsafely on ERT-generated instructions, underscoring the shortcomings of current benchmarks in evaluating real-world performance and safety. Code and videos are available at: https://s-karnik.github.io/embodied-red-team-project-page.
URLs: https://s-karnik.github.io/embodied-red-team-project-page.
Authors: Zhanfeng Wang, Xinyu Li, Hao Ding, Jian Qing Shi
Abstract: Extrinsic Gaussian process regression methods, such as wrapped Gaussian process, have been developed to analyze manifold data. However, there is a lack of intrinsic Gaussian process methods for studying complex data with manifold-valued response variables. In this paper, we first apply the parallel transport operator on Riemannian manifold to propose an intrinsic covariance structure that addresses a critical aspect of constructing a well-defined Gaussian process regression model. We then propose a novel intrinsic Gaussian process regression model for manifold-valued data, which can be applied to data situated not only on Euclidean submanifolds but also on manifolds without a natural ambient space. We establish the asymptotic properties of the proposed models, including information consistency and posterior consistency, and we also show that the posterior distribution of the regression function is invariant to the choice of orthonormal frames for the coordinate representations of the covariance function. Numerical studies, including simulation and real examples, indicate that the proposed methods work well.
Authors: Kaustubh Ponkshe, Raghav Singhal, Eduard Gorbunov, Alexey Tumanov, Samuel Horvath, Praneeth Vepakomma
Abstract: Low-rank adapters have become standard for efficiently fine-tuning large language models (LLMs), but they often fall short of achieving the performance of full fine-tuning. We propose a method, LoRA Silver Bullet or LoRA-SB, that approximates full fine-tuning within low-rank subspaces using a carefully designed initialization strategy. We theoretically demonstrate that the architecture of LoRA-XS, which inserts a learnable (r x r) matrix between B and A while keeping other matrices fixed, provides the precise conditions needed for this approximation. We leverage its constrained update space to achieve optimal scaling for high-rank gradient updates while removing the need for hyperparameter tuning. We prove that our initialization offers an optimal low-rank approximation of the initial gradient and preserves update directions throughout training. Extensive experiments across mathematical reasoning, commonsense reasoning, and language understanding tasks demonstrate that our approach exceeds the performance of standard LoRA while using \textbf{27-90} times fewer learnable parameters, and comprehensively outperforms LoRA-XS. Our findings establish that it is possible to simulate full fine-tuning in low-rank subspaces, and achieve significant efficiency gains without sacrificing performance. Our code is publicly available at https://github.com/RaghavSinghal10/lora-sb.
Authors: Jessica Gronsbell, Jianhui Gao, Yaqi Shi, Zachary R. McCaw, David Cheng
Abstract: From structural biology to epidemiology, predictions from machine learning (ML) models increasingly complement costly gold-standard data to enable faster, more affordable, and scalable scientific inquiry. In response, prediction-based (PB) inference has emerged to accommodate statistical analysis using a large volume of predictions together with a small amount of gold-standard data. The goals of PB inference are two-fold: (i) to mitigate bias from errors in predictions and (ii) to improve efficiency relative to traditional inference using only the gold-standard data. Motwani and Witten (2023) recently revisited two key PB inference approaches and found that only one method, Prediction-powered Inference (PPI) proposed by Angelopoulos et al. (2023), achieves (i). In this paper, we find that PPI does not achieve (ii). We revisit the double sampling literature and show that, with a simple modification, PPI can be adjusted to provide theoretically justified improvements in efficiency. We also contextualize PB inference with economics and statistics literature dating back to the 1960s to highlight the utility of classical methods in this contemporary problem. Our extensive theoretical analyses, along with an analysis of UK Biobank data, indicate that our proposal effectively mitigates bias and improves efficiency, making it preferable for use in practice.
Authors: Hugo Malard, Michel Olvera, Stephane Lathuiliere, Slim Essid
Abstract: Large-scale pre-trained audio and image models demonstrate an unprecedented degree of generalization, making them suitable for a wide range of applications. Here, we tackle the specific task of sound-prompted segmentation, aiming to segment image regions corresponding to objects heard in an audio signal. Most existing approaches tackle this problem by fine-tuning pre-trained models or by training additional modules specifically for the task. We adopt a different strategy: we introduce a training-free approach that leverages Non-negative Matrix Factorization (NMF) to co-factorize audio and visual features from pre-trained models so as to reveal shared interpretable concepts. These concepts are passed on to an open-vocabulary segmentation model for precise segmentation maps. By using frozen pre-trained models, our method achieves high generalization and establishes state-of-the-art performance in unsupervised sound-prompted segmentation, significantly surpassing previous unsupervised methods.
Authors: Yujie Wang, Shiju Wang, Shenhan Zhu, Fangcheng Fu, Xinyi Liu, Xuefeng Xiao, Huixia Li, Jiashi Li, Faming Wu, Bin Cui
Abstract: Extending the context length (i.e., the maximum supported sequence length) of LLMs is of paramount significance. To facilitate long context training of LLMs, sequence parallelism has emerged as an essential technique, which scatters each input sequence across multiple devices and necessitates communication to process the sequence. In essence, existing sequence parallelism methods assume homogeneous sequence lengths (i.e., all input sequences are equal in length) and therefore leverages a single, static scattering strategy for all input sequences. However, in reality, the sequence lengths in LLM training corpora exhibit substantial variability, often following a long-tail distribution, which leads to workload heterogeneity. In this paper, we show that employing a single, static strategy results in inefficiency and resource under-utilization, highlighting the need for adaptive approaches to handle the heterogeneous workloads across sequences. To address this, we propose a heterogeneity-adaptive sequence parallelism method. For each training step, our approach captures the variability in sequence lengths and assigns the optimal combination of scattering strategies based on workload characteristics. We model this problem as a linear programming optimization and design an efficient and effective solver to find the optimal solution. Furthermore, we implement our method in a high-performance system that supports adaptive parallelization in distributed LLM training. Experimental results demonstrate that our system outperforms state-of-the-art training frameworks by up to 1.98x.
Authors: Som Sagar, Jiafei Duan, Sreevishakh Vasudevan, Yifan Zhou, Heni Ben Amor, Dieter Fox, Ransalu Senanayake
Abstract: Robot manipulation policies often fail for unknown reasons, posing significant challenges for real-world deployment. Researchers and engineers typically address these failures using heuristic approaches, which are not only labor-intensive and costly but also prone to overlooking critical failure modes (FMs). This paper introduces Robot Manipulation Diagnosis (RoboMD), a systematic framework designed to automatically identify FMs arising from unanticipated changes in the environment. Considering the vast space of potential FMs in a pre-trained manipulation policy, we leverage deep reinforcement learning (deep RL) to explore and uncover these FMs using a specially trained vision-language embedding that encodes a notion of failures. This approach enables users to probabilistically quantify and rank failures in previously unseen environmental conditions. Through extensive experiments across various manipulation tasks and algorithms, we demonstrate RoboMD's effectiveness in diagnosing unknown failures in unstructured environments, providing a systematic pathway to improve the robustness of manipulation policies.
Authors: Zebin Wang, Yi Han, Ethan X. Fang, Lan Wang, Junwei Lu
Abstract: We consider the inference for the ranking of large language models (LLMs). Alignment arises as a significant challenge to mitigate hallucinations in the use of LLMs. Ranking LLMs has proven to be an effective tool to improve alignment based on the best-of-$N$ policy. In this paper, we propose a new inferential framework for hypothesis testing among the ranking for language models. Our framework is based on a nonparametric contextual ranking framework designed to assess large language models' domain-specific expertise, leveraging nonparametric scoring methods to account for their sensitivity to the prompts. To characterize the combinatorial complexity of the ranking, we introduce a novel concept of confidence diagram, which leverages a Hasse diagram to represent the entire confidence set of rankings by a single directed graph. We show the validity of the proposed confidence diagram by advancing the Gaussian multiplier bootstrap theory to accommodate the supremum of independent empirical processes that are not necessarily identically distributed. Extensive numerical experiments conducted on both synthetic and real data demonstrate that our approach offers valuable insight into the evaluation for the performance of different LLMs across various medical domains.
Authors: Huminhao Zhu, Fangyikang Wang, Tianyu Ding, Qing Qu, Zhihui Zhu
Abstract: Training with synthetic data is becoming increasingly inevitable as synthetic content proliferates across the web, driven by the remarkable performance of recent deep generative models. This reliance on synthetic data can also be intentional, as seen in Rectified Flow models, whose Reflow method iteratively uses self-generated data to straighten the flow and improve sampling efficiency. However, recent studies have shown that repeatedly training on self-generated samples can lead to model collapse (MC), where performance degrades over time. Despite this, most recent work on MC either focuses on empirical observations or analyzes regression problems and maximum likelihood objectives, leaving a rigorous theoretical analysis of reflow methods unexplored. In this paper, we aim to fill this gap by providing both theoretical analysis and practical solutions for addressing MC in diffusion/flow models. We begin by studying Denoising Autoencoders and prove performance degradation when DAEs are iteratively trained on their own outputs. To the best of our knowledge, we are the first to rigorously analyze model collapse in DAEs and, by extension, in diffusion models and Rectified Flow. Our analysis and experiments demonstrate that rectified flow also suffers from MC, leading to potential performance degradation in each reflow step. Additionally, we prove that incorporating real data can prevent MC during recursive DAE training, supporting the recent trend of using real data as an effective approach for mitigating MC. Building on these insights, we propose a novel Real-data Augmented Reflow and a series of improved variants, which seamlessly integrate real data into Reflow training by leveraging reverse flow. Empirical evaluations on standard image benchmarks confirm that RA Reflow effectively mitigates model collapse, preserving high-quality sample generation even with fewer sampling steps.
Authors: Wen-Chao Hu, Wang-Zhou Dai, Yuan Jiang, Zhi-Hua Zhou
Abstract: Neuro-Symbolic (NeSy) AI could be regarded as an analogy to human dual-process cognition, modeling the intuitive System 1 with neural networks and the algorithmic System 2 with symbolic reasoning. However, for complex learning targets, NeSy systems often generate outputs inconsistent with domain knowledge and it is challenging to rectify them. Inspired by the human Cognitive Reflection, which promptly detects errors in our intuitive response and revises them by invoking the System 2 reasoning, we propose to improve NeSy systems by introducing Abductive Reflection (ABL-Refl) based on the Abductive Learning (ABL) framework. ABL-Refl leverages domain knowledge to abduce a reflection vector during training, which can then flag potential errors in the neural network outputs and invoke abduction to rectify them and generate consistent outputs during inference. ABL-Refl is highly efficient in contrast to previous ABL implementations. Experiments show that ABL-Refl outperforms state-of-the-art NeSy methods, achieving excellent accuracy with fewer training resources and enhanced efficiency.
Authors: Ben Liu, Jihai Zhang, Fangquan Lin, Cheng Yang, Min Peng
Abstract: Large Language Models (LLMs) present massive inherent knowledge and superior semantic comprehension capability, which have revolutionized various tasks in natural language processing. Despite their success, a critical gap remains in enabling LLMs to perform knowledge graph completion (KGC). Empirical evidence suggests that LLMs consistently perform worse than conventional KGC approaches, even through sophisticated prompt design or tailored instruction-tuning. Fundamentally, applying LLMs on KGC introduces several critical challenges, including a vast set of entity candidates, hallucination issue of LLMs, and under-exploitation of the graph structure. To address these challenges, we propose a novel instruction-tuning-based method, namely FtG. Specifically, we present a filter-then-generate paradigm and formulate the KGC task into a multiple-choice question format. In this way, we can harness the capability of LLMs while mitigating the issue casused by hallucinations. Moreover, we devise a flexible ego-graph serialization prompt and employ a structure-text adapter to couple structure and text information in a contextualized manner. Experimental results demonstrate that FtG achieves substantial performance gain compared to existing state-of-the-art methods. The instruction dataset and code are available at https://github.com/LB0828/FtG.
Authors: Yue Zhang, Liqiang Jing, Vibhav Gogate
Abstract: We introduce a new task called Defeasible Visual Entailment (DVE), where the goal is to allow the modification of the entailment relationship between an image premise and a text hypothesis based on an additional update. While this concept is well-established in Natural Language Inference, it remains unexplored in visual entailment. At a high level, DVE enables models to refine their initial interpretations, leading to improved accuracy and reliability in various applications such as detecting misleading information in images, enhancing visual question answering, and refining decision-making processes in autonomous systems. Existing metrics do not adequately capture the change in the entailment relationship brought by updates. To address this, we propose a novel inference-aware evaluator designed to capture changes in entailment strength induced by updates, using pairwise contrastive learning and categorical information learning. Additionally, we introduce a reward-driven update optimization method to further enhance the quality of updates generated by multimodal models. Experimental results demonstrate the effectiveness of our proposed evaluator and optimization method.
Authors: Z. Wang, S. Li, Dongrui Wu
Abstract: Epilepsy significantly impacts global health, affecting about 65 million people worldwide, along with various animal species. The diagnostic processes of epilepsy are often hindered by the transient and unpredictable nature of seizures. Here we propose a multi-space alignment approach based on cross-species and cross-modality electroencephalogram (EEG) data to enhance the detection capabilities and understanding of epileptic seizures. By employing deep learning techniques, including domain adaptation and knowledge distillation, our framework aligns cross-species and cross-modality EEG signals to enhance the detection capability beyond traditional within-species and with-modality models. Experiments on multiple surface and intracranial EEG datasets of humans and canines demonstrated substantial improvements in the detection accuracy, achieving over 90% AUC scores for cross-species and cross-modality seizure detection with extremely limited labeled data from the target species/modality. To our knowledge, this is the first study that demonstrates the effectiveness of integrating heterogeneous data from different species and modalities to improve EEG-based seizure detection performance. The approach may also be generalizable to different brain-computer interface paradigms, and suggests the possibility to combine data from different species/modalities to increase the amount of training data for large EEG models.
Authors: Siyuan Huang, Liliang Chen, Pengfei Zhou, Shengcong Chen, Zhengkai Jiang, Yue Hu, Yue Liao, Peng Gao, Hongsheng Li, Maoqing Yao, Guanghui Ren
Abstract: We introduce EnerVerse, a generative robotics foundation model that constructs and interprets embodied spaces. EnerVerse employs an autoregressive video diffusion framework to predict future embodied spaces from instructions, enhanced by a sparse context memory for long-term reasoning. To model the 3D robotics world, we propose Free Anchor Views (FAVs), a multi-view video representation offering flexible, task-adaptive perspectives to address challenges like motion ambiguity and environmental constraints. Additionally, we present EnerVerse-D, a data engine pipeline combining the generative model with 4D Gaussian Splatting, forming a self-reinforcing data loop to reduce the sim-to-real gap. Leveraging these innovations, EnerVerse translates 4D world representations into physical actions via a policy head (EnerVerse-A), enabling robots to execute task instructions. EnerVerse-A achieves state-of-the-art performance in both simulation and real-world settings.
Authors: Ali Al-Lawati, Jason Lucas, Prasenjit Mitra
Abstract: Large Language Models (LLMs) have demonstrated remarkable performance in various NLP tasks, including semantic parsing, which translates natural language into formal code representations. However, the reverse process, translating code into natural language, termed semantic captioning, has received less attention. This task is becoming increasingly important as LLMs are integrated into platforms for code generation, security analysis, and educational purposes. In this paper, we focus on the captioning of SQL query (SQL2Text) to address the critical need for understanding and explaining SQL queries in an era where LLM-generated code poses potential security risks. We repurpose Text2SQL datasets for SQL2Text by introducing an iterative ICL prompt using GPT-4o to generate multiple additional utterances, which enhances the robustness of the datasets for the reverse task. We conduct our experiments using in-context learning (ICL) based on different sample selection methods, emphasizing smaller, more computationally efficient LLMs. Our findings demonstrate that leveraging the inherent graph properties of SQL for ICL sample selection significantly outperforms random selection by up to 39% on BLEU score and provides better results than alternative methods. Dataset and codes are published: https://github.com/aliwister/ast-icl.
Authors: Kexin Gu Baugh, Luke Dickens, Alessandra Russo
Abstract: Although deep reinforcement learning has been shown to be effective, the model's black-box nature presents barriers to direct policy interpretation. To address this problem, we propose a neuro-symbolic approach called neural DNF-MT for end-to-end policy learning. The differentiable nature of the neural DNF-MT model enables the use of deep actor-critic algorithms for training. At the same time, its architecture is designed so that trained models can be directly translated into interpretable policies expressed as standard (bivalent or probabilistic) logic programs. Moreover, additional layers can be included to extract abstract features from complex observations, acting as a form of predicate invention. The logic representations are highly interpretable, and we show how the bivalent representations of deterministic policies can be edited and incorporated back into a neural model, facilitating manual intervention and adaptation of learned policies. We evaluate our approach on a range of tasks requiring learning deterministic or stochastic behaviours from various forms of observations. Our empirical results show that our neural DNF-MT model performs at the level of competing black-box methods whilst providing interpretable policies.
Authors: Andrea Matarazzo, Riccardo Torlone
Abstract: The rapid advancement of artificial intelligence, particularly with the development of Large Language Models (LLMs) built on the transformer architecture, has redefined the capabilities of natural language processing. These models now exhibit remarkable performance across various language-related tasks, such as text generation, question answering, translation, and summarization, often rivaling human-like comprehension. More intriguingly, LLMs have demonstrated emergent abilities extending beyond their core functions, showing proficiency in tasks like commonsense reasoning, code generation, and arithmetic. This survey paper explores the foundational components, scaling mechanisms, and architectural strategies that drive these capabilities. Emphasizing models like GPT and LLaMA, we analyze the impact of exponential data and computational growth on LLM performance, while also addressing the trade-offs associated with scaling. We also examine LLM applications across sectors, such as healthcare, finance, education, and law, highlighting their adaptability and potential to solve domain-specific challenges. Central to this work are the questions of how LLMs generalize across diverse tasks, exhibit planning, and reasoning abilities, and whether these emergent abilities can be systematically elicited or enhanced. In particular, we provide some insights into the CoT (Chain of Thought) and PoT (Plan of Thought) abilities within LLMs, focusing on how pre-training data influences their emergence. Additionally, we investigate LLM-modulo frameworks that integrate external systems, allowing LLMs to handle complex, dynamic tasks. By analyzing these factors, this paper aims to foster the ongoing discussion on the capabilities and limits of LLMs, promoting their responsible development and application in novel and increasingly complex environments.
Authors: Aur\'elien Decelle, Alfonso de Jes\'us Navas G\'omez, Beatriz Seoane
Abstract: Maximum entropy methods, based on the inverse Ising/Potts problem from statistical mechanics, are essential for modeling interactions between pairs of variables in data-driven problems across disciplines such as bioinformatics, ecology, and neuroscience. Despite their considerable success, these methods typically fail to capture higher-order interactions that are often essential for understanding complex systems. Conversely, modern machine learning methods capture these complex interactions, but the computational cost of interpretable frameworks makes them impractical for real-world applications. Restricted Boltzmann Machines (RBMs) provide a computationally efficient way to capture statistical correlations using hidden nodes in a bipartite neural network. In this study, we introduce a new method that maps RBMs to generalized Potts models, allowing for the extraction of interactions up to any specified order. This method utilizes large-$N$ approximations, enabled by the RBM's simple structure, to extract effective many-body couplings with minimal computational effort. Furthermore, we propose a robust framework for extracting higher-order interactions in more complex probabilistic models and a simple gauge-fixing method within the effective many-body Potts model. Our validation on synthetic datasets confirms the method's ability to recover two- and three-body interactions accurately. When applied to protein sequence data, the framework competently reconstructs protein contact maps and provides performance comparable to the best inverse Potts models. These findings confirm that RBMs are an effective and streamlined tool for exploring higher-order interactions within complex systems.
Authors: Yifan Zhang, Yifeng Liu, Huizhuo Yuan, Zhen Qin, Yang Yuan, Quanquan Gu, Andrew Chi-Chih Yao
Abstract: Scaling language models to handle longer input sequences typically necessitates large key-value (KV) caches, resulting in substantial memory overhead during inference. In this paper, we propose Tensor Product Attention (TPA), a novel attention mechanism that uses tensor decompositions to represent queries, keys, and values compactly, significantly shrinking KV cache size at inference time. By factorizing these representations into contextual low-rank components (contextual factorization) and seamlessly integrating with RoPE, TPA achieves improved model quality alongside memory efficiency. Based on TPA, we introduce the Tensor ProducT ATTenTion Transformer (T6), a new model architecture for sequence modeling. Through extensive empirical evaluation of language modeling tasks, we demonstrate that T6 exceeds the performance of standard Transformer baselines including MHA, MQA, GQA, and MLA across various metrics, including perplexity and a range of renowned evaluation benchmarks. Notably, TPA's memory efficiency enables the processing of significantly longer sequences under fixed resource constraints, addressing a critical scalability challenge in modern language models. The code is available at https://github.com/tensorgi/T6.
Authors: Dong Liu, Esther Lopez Ramos
Abstract: Semantic retrieval (also known as dense retrieval) based on textual data has been extensively studied for both web search and product search application fields, where the relevance of a query and a potential target document is computed by their dense vector representation comparison. Product image is crucial for e-commerce search interactions and is a key factor for customers at product explorations. However, its impact on semantic retrieval has not been well studied yet. In this research, we build a multimodal representation for product items in e-commerce search in contrast to pure-text representation of products, and investigate the impact of such representations. The models are developed and evaluated on e-commerce datasets. We demonstrate that a multimodal representation scheme for a product can show improvement either on purchase recall or relevance accuracy in semantic retrieval. Additionally, we provide numerical analysis for exclusive matches retrieved by a multimodal semantic retrieval model versus a text-only semantic retrieval model, to demonstrate the validation of multimodal solutions.
Authors: Saman Motamed, Laura Culp, Kevin Swersky, Priyank Jaini, Robert Geirhos
Abstract: AI video generation is undergoing a revolution, with quality and realism advancing rapidly. These advances have led to a passionate scientific debate: Do video models learn "world models" that discover laws of physics -- or, alternatively, are they merely sophisticated pixel predictors that achieve visual realism without understanding the physical principles of reality? We address this question by developing Physics-IQ, a comprehensive benchmark dataset that can only be solved by acquiring a deep understanding of various physical principles, like fluid dynamics, optics, solid mechanics, magnetism and thermodynamics. We find that across a range of current models (Sora, Runway, Pika, Lumiere, Stable Video Diffusion, and VideoPoet), physical understanding is severely limited, and unrelated to visual realism. At the same time, some test cases can already be successfully solved. This indicates that acquiring certain physical principles from observation alone may be possible, but significant challenges remain. While we expect rapid advances ahead, our work demonstrates that visual realism does not imply physical understanding. Our project page is at https://physics-iq.github.io; code at https://github.com/google-deepmind/physics-IQ-benchmark.
URLs: https://physics-iq.github.io;, https://github.com/google-deepmind/physics-IQ-benchmark.
Authors: Yuan Wang, Bin Zhu, Yanbin Hao, Chong-Wah Ngo, Yi Tan, Xiang Wang
Abstract: Recent advancements in text-to-image generation models have excelled in creating diverse and realistic images. This success extends to food imagery, where various conditional inputs like cooking styles, ingredients, and recipes are utilized. However, a yet-unexplored challenge is generating a sequence of procedural images based on cooking steps from a recipe. This could enhance the cooking experience with visual guidance and possibly lead to an intelligent cooking simulation system. To fill this gap, we introduce a novel task called \textbf{cooking procedural image generation}. This task is inherently demanding, as it strives to create photo-realistic images that align with cooking steps while preserving sequential consistency. To collectively tackle these challenges, we present \textbf{CookingDiffusion}, a novel approach that leverages Stable Diffusion and three innovative Memory Nets to model procedural prompts. These prompts encompass text prompts (representing cooking steps), image prompts (corresponding to cooking images), and multi-modal prompts (mixing cooking steps and images), ensuring the consistent generation of cooking procedural images. To validate the effectiveness of our approach, we preprocess the YouCookII dataset, establishing a new benchmark. Our experimental results demonstrate that our model excels at generating high-quality cooking procedural images with remarkable consistency across sequential cooking steps, as measured by both the FID and the proposed Average Procedure Consistency metrics. Furthermore, CookingDiffusion demonstrates the ability to manipulate ingredients and cooking methods in a recipe. We will make our code, models, and dataset publicly accessible.
Authors: BG Tong
Abstract: Objective: This study explores a novel deep learning approach for EEG analysis and perceptual state guidance, inspired by Level of Detail (LOD) theory. The goal is to improve perceptual state identification accuracy and advance personalized psychological therapy. Methods: Portable EEG devices and music rhythm signals were used for data collection. LOD theory was applied to dynamically adjust EEG signal processing, extracting core perceptual features. A Unity-based software system integrated EEG data with audio materials. The deep learning model combined a CNN for feature extraction and classification, and a DQN for reinforcement learning to optimize rhythm adjustments. Results: The CNN achieved 94.05% accuracy in perceptual state classification. The DQN guided subjects to target states with a 92.45% success rate, averaging 13.2 rhythm cycles. However, only 50% of users reported psychological alignment with the target state, indicating room for improvement. Discussion: The results validate the potential of LOD-based EEG biofeedback. Limitations include dataset source, label subjectivity, and reward function optimization. Future work will expand to diverse subjects, incorporate varied musical elements, and refine reward functions for better generalization and personalization.
Authors: Shahaf Pruss, Morris Alper, Hadar Averbuch-Elor
Abstract: Images depicting complex, dynamic scenes are challenging to parse automatically, requiring both high-level comprehension of the overall situation and fine-grained identification of participating entities and their interactions. Current approaches use distinct methods tailored to sub-tasks such as Situation Recognition and detection of Human-Human and Human-Object Interactions. However, recent advances in image understanding have often leveraged web-scale vision-language (V&L) representations to obviate task-specific engineering. In this work, we propose a framework for dynamic scene understanding tasks by leveraging knowledge from modern, frozen V&L representations. By framing these tasks in a generic manner - as predicting and parsing structured text, or by directly concatenating representations to the input of existing models - we achieve state-of-the-art results while using a minimal number of trainable parameters relative to existing approaches. Moreover, our analysis of dynamic knowledge of these representations shows that recent, more powerful representations effectively encode dynamic scene semantics, making this approach newly possible.
Authors: Samer Attrah
Abstract: Emotion estimation in general is a field that has been studied for a long time, and several approaches exist using machine learning. in this paper, we present an LSTM model, that processes the blend-shapes produced by the library MediaPipe, for a face detected in a live stream of a camera, to estimate the main emotion from the facial expressions, this model is trained on the FER2013 dataset and delivers a result of 71% accuracy and 62% f1-score which meets the accuracy benchmark of the FER2013 dataset, with significantly reduced computation costs. https://github.com/Samir-atra/Emotion_estimation_from_video_footage_with_LSTM_ML_algorithm
URLs: https://github.com/Samir-atra/Emotion_estimation_from_video_footage_with_LSTM_ML_algorithm
Authors: Ting Sun, Penghan Wang, Fan Lai
Abstract: Large language models (LLMs) have facilitated a wide range of applications with distinct service-level objectives (SLOs), from latency-sensitive online tasks like interactive chatbots to throughput-oriented offline workloads like document summarization. The existing deployment model, which dedicates machines to each workload, simplifies SLO management but often leads to poor resource utilization. This paper introduces HyGen, an interference-aware LLM serving system that enables efficient co-location of online and offline workloads while preserving latency requirements. HyGen incorporates two key innovations: (1) performance control mechanisms, including a latency predictor to estimate batch execution time and an SLO-aware profiler to quantify latency interference, and (2) SLO-aware offline scheduling policies that maximize serving throughput and prevent starvation, without compromising online serving latency. Our evaluation on production workloads shows that HyGen achieves up to 3.87x overall throughput and 5.84x offline throughput gains over online and hybrid serving baselines, respectively, while strictly satisfying latency SLOs.
Authors: Yinan Zheng, Ruiming Liang, Kexin Zheng, Jinliang Zheng, Liyuan Mao, Jianxiong Li, Weihao Gu, Rui Ai, Shengbo Eben Li, Xianyuan Zhan, Jingjing Liu
Abstract: Achieving human-like driving behaviors in complex open-world environments is a critical challenge in autonomous driving. Contemporary learning-based planning approaches such as imitation learning methods often struggle to balance competing objectives and lack of safety assurance,due to limited adaptability and inadequacy in learning complex multi-modal behaviors commonly exhibited in human planning, not to mention their strong reliance on the fallback strategy with predefined rules. We propose a novel transformer-based Diffusion Planner for closed-loop planning, which can effectively model multi-modal driving behavior and ensure trajectory quality without any rule-based refinement. Our model supports joint modeling of both prediction and planning tasks under the same architecture, enabling cooperative behaviors between vehicles. Moreover, by learning the gradient of the trajectory score function and employing a flexible classifier guidance mechanism, Diffusion Planner effectively achieves safe and adaptable planning behaviors. Evaluations on the large-scale real-world autonomous planning benchmark nuPlan and our newly collected 200-hour delivery-vehicle driving dataset demonstrate that Diffusion Planner achieves state-of-the-art closed-loop performance with robust transferability in diverse driving styles.
Authors: Haoyu Liang, Youran Sun, Yunfeng Cai, Jun Zhu, Bo Zhang
Abstract: The security issue of large language models (LLMs) has gained significant attention recently, with various defense mechanisms developed to prevent harmful outputs, among which safeguards based on text embedding models serve as a fundamental defense. Through testing, we discover that the distribution of text embedding model outputs is significantly biased with a large mean. Inspired by this observation, we propose novel efficient methods to search for universal magic words that can attack text embedding models. The universal magic words as suffixes can move the embedding of any text towards the bias direction, therefore manipulate the similarity of any text pair and mislead safeguards. By appending magic words to user prompts and requiring LLMs to end answers with magic words, attackers can jailbreak the safeguard. To eradicate this security risk, we also propose defense mechanisms against such attacks, which can correct the biased distribution of text embeddings in a train-free manner.
Authors: Mohd. Farhan Israk Soumik, W. K. M. Mithsara, Abdur R. Shahid, Ahmed Imteaj
Abstract: The rapid proliferation of speech-enabled technologies, including virtual assistants, video conferencing platforms, and wearable devices, has raised significant privacy concerns, particularly regarding the inference of sensitive emotional information from audio data. Existing privacy-preserving methods often compromise usability and security, limiting their adoption in practical scenarios. This paper introduces a novel, user-centric approach that leverages familiar audio editing techniques, specifically pitch and tempo manipulation, to protect emotional privacy without sacrificing usability. By analyzing popular audio editing applications on Android and iOS platforms, we identified these features as both widely available and usable. We rigorously evaluated their effectiveness against a threat model, considering adversarial attacks from diverse sources, including Deep Neural Networks (DNNs), Large Language Models (LLMs), and and reversibility testing. Our experiments, conducted on three distinct datasets, demonstrate that pitch and tempo manipulation effectively obfuscates emotional data. Additionally, we explore the design principles for lightweight, on-device implementation to ensure broad applicability across various devices and platforms.
Authors: Chuhan Zhang, Cong Wang, Wei Pan, Cosimo Della Santina
Abstract: Inspired by the dynamic coupling of moto-neurons and physical elasticity in animals, this work explores the possibility of generating locomotion gaits by utilizing physical oscillations in a soft snake by means of a low-level spiking neural mechanism. To achieve this goal, we introduce the Double Threshold Spiking neuron model with adjustable thresholds to generate varied output patterns. This neuron model can excite the natural dynamics of soft robotic snakes, and it enables distinct movements, such as turning or moving forward, by simply altering the neural thresholds. Finally, we demonstrate that our approach, termed SpikingSoft, naturally pairs and integrates with reinforcement learning. The high-level agent only needs to adjust the two thresholds to generate complex movement patterns, thus strongly simplifying the learning of reactive locomotion. Simulation results demonstrate that the proposed architecture significantly enhances the performance of the soft snake robot, enabling it to achieve target objectives with a 21.6% increase in success rate, a 29% reduction in time to reach the target, and smoother movements compared to the vanilla reinforcement learning controllers or Central Pattern Generator controller acting in torque space.
Authors: Jianwei Shi, Sameh Abdulah, Ying Sun, Marc G. Genton
Abstract: Advancements in information technology have enabled the creation of massive spatial datasets, driving the need for scalable and efficient computational methodologies. While offering viable solutions, centralized frameworks are limited by vulnerabilities such as single-point failures and communication bottlenecks. This paper presents a decentralized framework tailored for parameter inference in spatial low-rank models to address these challenges. A key obstacle arises from the spatial dependence among observations, which prevents the log-likelihood from being expressed as a summation-a critical requirement for decentralized optimization approaches. To overcome this challenge, we propose a novel objective function leveraging the evidence lower bound, which facilitates the use of decentralized optimization techniques. Our approach employs a block descent method integrated with multi-consensus and dynamic consensus averaging for effective parameter optimization. We prove the convexity of the new objective function in the vicinity of the true parameters, ensuring the convergence of the proposed method. Additionally, we present the first theoretical results establishing the consistency and asymptotic normality of the estimator within the context of spatial low-rank models. Extensive simulations and real-world data experiments corroborate these theoretical findings, showcasing the robustness and scalability of the framework.
Authors: Samiran Dey, Christopher R. S. Banerji, Partha Basuchowdhuri, Sanjoy K. Saha, Deepak Parashar, Tapabrata Chakraborti
Abstract: Emerging research has highlighted that artificial intelligence based multimodal fusion of digital pathology and transcriptomic features can improve cancer diagnosis (grading/subtyping) and prognosis (survival risk) prediction. However, such direct fusion for joint decision is impractical in real clinical settings, where histopathology is still the gold standard for diagnosis and transcriptomic tests are rarely requested, at least in the public healthcare system. With our novel diffusion based crossmodal generative AI model PathoGen, we show that genomic expressions synthesized from digital histopathology jointly predicts cancer grading and patient survival risk with high accuracy (state-of-the-art performance), certainty (through conformal coverage guarantee) and interpretability (through distributed attention maps). PathoGen code is available for open use by the research community through GitHub at https://github.com/Samiran-Dey/PathoGen.
Authors: Hai Victor Habi, Hagit Messer, Yoram Bresler
Abstract: The Bayesian Cram\'er-Rao bound (BCRB) is a crucial tool in signal processing for assessing the fundamental limitations of any estimation problem as well as benchmarking within a Bayesian frameworks. However, the BCRB cannot be computed without full knowledge of the prior and the measurement distributions. In this work, we propose a fully learned Bayesian Cram\'er-Rao bound (LBCRB) that learns both the prior and the measurement distributions. Specifically, we suggest two approaches to obtain the LBCRB: the Posterior Approach and the Measurement-Prior Approach. The Posterior Approach provides a simple method to obtain the LBCRB, whereas the Measurement-Prior Approach enables us to incorporate domain knowledge to improve the sample complexity and {interpretability}. To achieve this, we introduce a Physics-encoded score neural network which enables us to easily incorporate such domain knowledge into a neural network. We {study the learning} errors of the two suggested approaches theoretically, and validate them numerically. We demonstrate the two approaches on several signal processing examples, including a linear measurement problem with unknown mixing and Gaussian noise covariance matrices, frequency estimation, and quantized measurement. In addition, we test our approach on a nonlinear signal processing problem of frequency estimation with real-world underwater ambient noise.
Authors: Kensuke Nakamura, Lasse Peters, Andrea Bajcsy
Abstract: Hamilton-Jacobi (HJ) reachability is a rigorous mathematical framework that enables robots to simultaneously detect unsafe states and generate actions that prevent future failures. While in theory, HJ reachability can synthesize safe controllers for nonlinear systems and nonconvex constraints, in practice, it has been limited to hand-engineered collision-avoidance constraints modeled via low-dimensional state-space representations and first-principles dynamics. In this work, our goal is to generalize safe robot controllers to prevent failures that are hard -- if not impossible -- to write down by hand, but can be intuitively identified from high-dimensional observations: for example, spilling the contents of a bag. We propose Latent Safety Filters, a latent-space generalization of HJ reachability that tractably operates directly on raw observation data (e.g., RGB images) by performing safety analysis in the latent embedding space of a generative world model. This transforms nuanced constraint specification to a classification problem in latent space and enables reasoning about dynamical consequences that are hard to simulate. In simulation and hardware experiments, we use Latent Safety Filters to safeguard arbitrary policies (from generative policies to direct teleoperation) from complex safety hazards, like preventing a Franka Research 3 manipulator from spilling the contents of a bag or toppling cluttered objects.
Authors: Tairan He, Jiawei Gao, Wenli Xiao, Yuanhang Zhang, Zi Wang, Jiashun Wang, Zhengyi Luo, Guanqi He, Nikhil Sobanbab, Chaoyi Pan, Zeji Yi, Guannan Qu, Kris Kitani, Jessica Hodgins, Linxi "Jim" Fan, Yuke Zhu, Changliu Liu, Guanya Shi
Abstract: Humanoid robots hold the potential for unparalleled versatility in performing human-like, whole-body skills. However, achieving agile and coordinated whole-body motions remains a significant challenge due to the dynamics mismatch between simulation and the real world. Existing approaches, such as system identification (SysID) and domain randomization (DR) methods, often rely on labor-intensive parameter tuning or result in overly conservative policies that sacrifice agility. In this paper, we present ASAP (Aligning Simulation and Real-World Physics), a two-stage framework designed to tackle the dynamics mismatch and enable agile humanoid whole-body skills. In the first stage, we pre-train motion tracking policies in simulation using retargeted human motion data. In the second stage, we deploy the policies in the real world and collect real-world data to train a delta (residual) action model that compensates for the dynamics mismatch. Then, ASAP fine-tunes pre-trained policies with the delta action model integrated into the simulator to align effectively with real-world dynamics. We evaluate ASAP across three transfer scenarios: IsaacGym to IsaacSim, IsaacGym to Genesis, and IsaacGym to the real-world Unitree G1 humanoid robot. Our approach significantly improves agility and whole-body coordination across various dynamic motions, reducing tracking error compared to SysID, DR, and delta dynamics learning baselines. ASAP enables highly agile motions that were previously difficult to achieve, demonstrating the potential of delta action learning in bridging simulation and real-world dynamics. These results suggest a promising sim-to-real direction for developing more expressive and agile humanoids.
Authors: Wenhao Zheng, Yixiao Chen, Weitong Zhang, Souvik Kundu, Yun Li, Zhengzhong Liu, Eric P. Xing, Hongyi Wang, Huaxiu Yao
Abstract: Large language models have achieved remarkable success in various tasks but suffer from high computational costs during inference, limiting their deployment in resource-constrained applications. To address this issue, we propose a novel CITER (Collaborative Inference with Token-lEvel Routing) framework that enables efficient collaboration between small and large language models (SLMs & LLMs) through a token-level routing strategy. Specifically, CITER routes non-critical tokens to an SLM for efficiency and routes critical tokens to an LLM for generalization quality. We formulate router training as a policy optimization, where the router receives rewards based on both the quality of predictions and the inference costs of generation. This allows the router to learn to predict token-level routing scores and make routing decisions based on both the current token and the future impact of its decisions. To further accelerate the reward evaluation process, we introduce a shortcut which significantly reduces the costs of the reward estimation and improving the practicality of our approach. Extensive experiments on five benchmark datasets demonstrate that CITER reduces the inference costs while preserving high-quality generation, offering a promising solution for real-time and resource-constrained applications. Our data and code are available at https://github.com/aiming-lab/CITER.
Authors: Ruisi Zhang, Neusha Javidnia, Nojan Sheybani, Farinaz Koushanfar
Abstract: This paper introduces RoSeMary, the first-of-its-kind ML/Crypto codesign watermarking framework that regulates LLM-generated code to avoid intellectual property rights violations and inappropriate misuse in software development. High-quality watermarks adhering to the detectability-fidelity-robustness tri-objective are limited due to codes' low-entropy nature. Watermark verification, however, often needs to reveal the signature and requires re-encoding new ones for code reuse, which potentially compromising the system's usability. To overcome these challenges, RoSeMary obtains high-quality watermarks by training the watermark insertion and extraction modules end-to-end to ensure (i) unaltered watermarked code functionality and (ii) enhanced detectability and robustness leveraging pre-trained CodeT5 as the insertion backbone to enlarge the code syntactic and variable rename transformation search space. In the deployment, RoSeMary uses zero-knowledge proofs for secure verification without revealing the underlying signatures. Extensive evaluations demonstrated RoSeMary achieves high detection accuracy while preserving the code functionality. RoSeMary is also robust against attacks and provides efficient secure watermark verification.
Authors: Sam Young, Yeon-jae Jwa, Kazuhiro Terao
Abstract: Effective self-supervised learning (SSL) techniques have been key to unlocking large datasets for representation learning. While many promising methods have been developed using online corpora and captioned photographs, their application to scientific domains, where data encodes highly specialized knowledge, remains in its early stages. We present a self-supervised masked modeling framework for 3D particle trajectory analysis in Time Projection Chambers (TPCs). These detectors produce globally sparse (<1% occupancy) but locally dense point clouds, capturing meter-scale particle trajectories at millimeter resolution. Starting with PointMAE, this work proposes volumetric tokenization to group sparse ionization points into resolution-agnostic patches, as well as an auxiliary energy infilling task to improve trajectory semantics. This approach -- which we call Point-based Liquid Argon Masked Autoencoder (PoLAr-MAE) -- achieves 99.4% track and 97.7% shower classification F-scores, matching that of supervised baselines without any labeled data. While the model learns rich particle trajectory representations, it struggles with sub-token phenomena like overlapping or short-lived particle trajectories. To support further research, we release PILArNet-M -- the largest open LArTPC dataset (1M+ events, 5.2B labeled points) -- to advance SSL in high energy physics (HEP). Project site: https://youngsm.com/polarmae/
Authors: Nina Freise, Marius Heitlinger, Ruben Nuredini, Gerrit Meixner
Abstract: Artificial Intelligence (AI) advancement is heavily dependent on access to large-scale, high-quality training data. However, in specialized domains such as healthcare, data acquisition faces significant constraints due to privacy regulations, ethical considerations, and limited availability. While synthetic data generation offers a promising solution, conventional approaches typically require substantial real data for training generative models. The emergence of large-scale prompt-based models presents new opportunities for synthetic data generation without direct access to protected data. However, crafting effective prompts for domain-specific data generation remains challenging, and manual prompt engineering proves insufficient for achieving output with sufficient precision and authenticity. We review recent developments in automatic prompt optimization, following PRISMA guidelines. We analyze six peer-reviewed studies published between 2020 and 2024 that focus on automatic data-free prompt optimization methods. Our analysis reveals three approaches: feedback-driven, error-based, and control-theoretic. Although all approaches demonstrate promising capabilities in prompt refinement and adaptation, our findings suggest the need for an integrated framework that combines complementary optimization techniques to enhance synthetic data generation while minimizing manual intervention. We propose future research directions toward developing robust, iterative prompt optimization frameworks capable of improving the quality of synthetic data. This advancement can be particularly crucial for sensitive fields and in specialized domains where data access is restricted, potentially transforming how we approach synthetic data generation for AI development.
Authors: Ryan Rabinowitz, Steve Cruz, Manuel G\"unther, Terrance E. Boult
Abstract: Evaluations of large-scale recognition methods typically focus on overall performance. While this approach is common, it often fails to provide insights into performance across individual classes, which can lead to fairness issues and misrepresentation. Addressing these gaps is crucial for accurately assessing how well methods handle novel or unseen classes and ensuring a fair evaluation. To address fairness in Open-Set Recognition (OSR), we demonstrate that per-class performance can vary dramatically. We introduce Gaussian Hypothesis Open Set Technique (GHOST), a novel hyperparameter-free algorithm that models deep features using class-wise multivariate Gaussian distributions with diagonal covariance matrices. We apply Z-score normalization to logits to mitigate the impact of feature magnitudes that deviate from the model's expectations, thereby reducing the likelihood of the network assigning a high score to an unknown sample. We evaluate GHOST across multiple ImageNet-1K pre-trained deep networks and test it with four different unknown datasets. Using standard metrics such as AUOSCR, AUROC and FPR95, we achieve statistically significant improvements, advancing the state-of-the-art in large-scale OSR. Source code is provided online.
Authors: Darina Koishigarina, Arnas Uselis, Seong Joon Oh
Abstract: CLIP (Contrastive Language-Image Pretraining) has become a popular choice for various downstream tasks. However, recent studies have questioned its ability to represent compositional concepts effectively. These works suggest that CLIP often acts like a bag-of-words (BoW) model, interpreting images and text as sets of individual concepts without grasping the structural relationships. In particular, CLIP struggles to correctly bind attributes to their corresponding objects when multiple objects are present in an image or text. In this work, we investigate why CLIP exhibits this BoW-like behavior. We find that the correct attribute-object binding information is already present in individual text and image modalities. Instead, the issue lies in the cross-modal alignment, which relies on cosine similarity. To address this, we propose Linear Attribute Binding CLIP or LABCLIP. It applies a linear transformation to text embeddings before computing cosine similarity. This approach significantly improves CLIP's ability to bind attributes to correct objects, thereby enhancing its compositional understanding. The code is available at https://github.com/kdariina/CLIP-not-BoW-unimodally.
Authors: YoonJe Kang, Yonghoon Jung, Wonseop Shin, Bumsoo Kim, Sanghyun Seo
Abstract: In this paper, we present synthetic data generation framework for flood hazard detection system. For high fidelity and quality, we characterize several real-world properties into virtual world and simulate the flood situation by controlling them. For the sake of efficiency, recent generative models in image-to-3D and urban city synthesis are leveraged to easily composite flood environments so that we avoid data bias due to the hand-crafted manner. Based on our framework, we build the flood synthetic dataset with 5 levels, dubbed MultiFloodSynth which contains rich annotation types like normal map, segmentation, 3D bounding box for a variety of downstream task. In experiments, our dataset demonstrate the enhanced performance of flood hazard detection with on-par realism compared with real dataset.
Authors: Yixin Liu, Lie Lu, Jihui Jin, Lichao Sun, Andrea Fanelli
Abstract: The rapid proliferation of generative audio synthesis and editing technologies has raised significant concerns about copyright infringement, data provenance, and the spread of misinformation through deepfake audio. Watermarking offers a proactive solution by embedding imperceptible, identifiable, and traceable marks into audio content. While recent neural network-based watermarking methods like WavMark and AudioSeal have improved robustness and quality, they struggle to achieve both robust detection and accurate attribution simultaneously. This paper introduces Cross-Attention Robust Audio Watermark (XAttnMark), which bridges this gap by leveraging partial parameter sharing between the generator and the detector, a cross-attention mechanism for efficient message retrieval, and a temporal conditioning module for improved message distribution. Additionally, we propose a psychoacoustic-aligned temporal-frequency masking loss that captures fine-grained auditory masking effects, enhancing watermark imperceptibility. Our approach achieves state-of-the-art performance in both detection and attribution, demonstrating superior robustness against a wide range of audio transformations, including challenging generative editing with strong editing strength. The project webpage is available at https://liuyixin-louis.github.io/xattnmark/.
Authors: Thorben Prein, Elton Pan, Sami Haddouti, Marco Lorenz, Janik Jehkul, Tymoteusz Wilk, Cansu Moran, Menelaos Panagiotis Fotiadis, Artur P. Toshev, Elsa Olivetti, Jennifer L. M. Rupp
Abstract: Retrosynthesis strategically plans the synthesis of a chemical target compound from simpler, readily available precursor compounds. This process is critical for synthesizing novel inorganic materials, yet traditional methods in inorganic chemistry continue to rely on trial-and-error experimentation. Emerging machine-learning approaches struggle to generalize to entirely new reactions due to their reliance on known precursors, as they frame retrosynthesis as a multi-label classification task. To address these limitations, we propose Retro-Rank-In, a novel framework that reformulates the retrosynthesis problem by embedding target and precursor materials into a shared latent space and learning a pairwise ranker on a bipartite graph of inorganic compounds. We evaluate Retro-Rank-In's generalizability on challenging retrosynthesis dataset splits designed to mitigate data duplicates and overlaps. For instance, for Cr2AlB2, it correctly predicts the verified precursor pair CrB + Al despite never seeing them in training, a capability absent in prior work. Extensive experiments show that Retro-Rank-In sets a new state-of-the-art, particularly in out-of-distribution generalization and candidate set ranking, offering a powerful tool for accelerating inorganic material synthesis.
Authors: Kamer Ali Yuksel, Hassan Sawaf
Abstract: Recent advances in large language models (LLMs) have shown remarkable performance across diverse tasks. However, these models are typically deployed with fixed weights, which limits their ability to adapt dynamically to the variability inherent in real-world data during inference. This paper introduces ChamaleonLLM, a novel framework that enables inference-time adaptation of LLMs by leveraging batch-aware clustering and on-the-fly generation of low-rank updates. Unlike traditional fine-tuning approaches such as Low-Rank Adaptation (LoRA) or methods that rely on a fixed set of pre-learned uniforms (changeable masks), our method dynamically generates adaptive modifications to the decoder weights based on the aggregated statistics of clustered batches. By intelligently grouping similar inputs and computing context-aware low-rank updates via a hyper-network, ChamaleonLLM achieves significant performance gains, outperforming conventional LoRA methods while eliminating the overhead of maintaining multiple expert models. Our experiments highlight the potential of our approach to serve as a versatile and highly adaptive solution for language model inference. ChamaleonLLM is open-sourced to ensure the reproducibility of our experiments: https://anonymous.4open.science/r/ChamaleonLLM/
Authors: Minseok Jung (May), Cynthia Fuertes Panizo (May), Liam Dugan (May), Yi R. (May), Fung, Pin-Yu Chen, Paul Pu Liang
Abstract: The advancement of large language models (LLMs) has made it difficult to differentiate human-written text from AI-generated text. Several AI-text detectors have been developed in response, which typically utilize a fixed global threshold (e.g., {\theta} = 0.5) to classify machine-generated text. However, we find that one universal threshold can fail to account for subgroup-specific distributional variations. For example, when using a fixed threshold, detectors make more false positive errors on shorter human-written text than longer, and more positive classifications on neurotic writing styles than open among long text. These discrepancies can lead to misclassification that disproportionately affects certain groups. We address this critical limitation by introducing FairOPT, an algorithm for group-specific threshold optimization in AI-generated content classifiers. Our approach partitions data into subgroups based on attributes (e.g., text length and writing style) and learns decision thresholds for each group, which enables careful balancing of performance and fairness metrics within each subgroup. In experiments with four AI text classifiers on three datasets, FairOPT enhances overall F1 score and decreases balanced error rate (BER) discrepancy across subgroups. Our framework paves the way for more robust and fair classification criteria in AI-generated output detection.
Authors: Zhenwei Wu, Jinxiong Lu, Yuxiao Chen, Yunxin Liu, Yueting Zhuang, Luhui Hu
Abstract: Humanoid robotics presents significant challenges in artificial intelligence, requiring precise coordination and control of high-degree-of-freedom systems. Designing effective reward functions for deep reinforcement learning (DRL) in this domain remains a critical bottleneck, demanding extensive manual effort, domain expertise, and iterative refinement. To overcome these challenges, we introduce STRIDE, a novel framework built on agentic engineering to automate reward design, DRL training, and feedback optimization for humanoid robot locomotion tasks. By combining the structured principles of agentic engineering with large language models (LLMs) for code-writing, zero-shot generation, and in-context optimization, STRIDE generates, evaluates, and iteratively refines reward functions without relying on task-specific prompts or templates. Across diverse environments featuring humanoid robot morphologies, STRIDE outperforms the state-of-the-art reward design framework EUREKA, achieving significant improvements in efficiency and task performance. Using STRIDE-generated rewards, simulated humanoid robots achieve sprint-level locomotion across complex terrains, highlighting its ability to advance DRL workflows and humanoid robotics research.