Authors: R\^omulo Damasclin Chaves dos Santos, Jorge Henrique de Oliveira Sales
Abstract: In this work, we introduce a new class of neural network operators designed to handle problems where memory effects and randomness play a central role. In this work, we introduce a new class of neural network operators designed to handle problems where memory effects and randomness play a central role. These operators merge symmetrized activation functions, Caputo-type fractional derivatives, and stochastic perturbations introduced via It\^o type noise. The result is a powerful framework capable of approximating functions that evolve over time with both long-term memory and uncertain dynamics. We develop the mathematical foundations of these operators, proving three key theorems of Voronovskaya type. These results describe the asymptotic behavior of the operators, their convergence in the mean-square sense, and their consistency under fractional regularity assumptions. All estimates explicitly account for the influence of the memory parameter $\alpha$ and the noise level $\sigma$. As a practical application, we apply the proposed theory to the fractional Navier-Stokes equations with stochastic forcing, a model often used to describe turbulence in fluid flows with memory. Our approach provides theoretical guarantees for the approximation quality and suggests that these neural operators can serve as effective tools in the analysis and simulation of complex systems. By blending ideas from neural networks, fractional calculus, and stochastic analysis, this research opens new perspectives for modeling turbulent phenomena and other multiscale processes where memory and randomness are fundamental. The results lay the groundwork for hybrid learning-based methods with strong analytical backing.
Authors: Mohammad Rubyet Islam
Abstract: The pursuit of alpha returns that exceed market benchmarks has undergone a profound transformation, evolving from intuition-driven investing to autonomous, AI powered systems. This paper introduces a comprehensive five stage taxonomy that traces this progression across manual strategies, statistical models, classical machine learning, deep learning, and agentic architectures powered by large language models (LLMs). Unlike prior surveys focused narrowly on modeling techniques, this review adopts a system level lens, integrating advances in representation learning, multimodal data fusion, and tool augmented LLM agents. The strategic shift from static predictors to contextaware financial agents capable of real time reasoning, scenario simulation, and cross modal decision making is emphasized. Key challenges in interpretability, data fragility, governance, and regulatory compliance areas critical to production deployment are examined. The proposed taxonomy offers a unified framework for evaluating maturity, aligning infrastructure, and guiding the responsible development of next generation alpha systems.
Authors: Yunho Jin, Gu-Yeon Wei, David Brooks
Abstract: Scaling large language models (LLMs) has driven significant advancements, yet it faces diminishing returns and escalating energy demands. This work introduces test-time compute (TTC)-allocating additional computational resources during inference-as a compelling complement to conventional scaling strategies. Specifically, we investigate whether employing TTC can achieve superior accuracy-energy trade-offs compared to simply increasing model size. Our empirical analysis reveals that TTC surpasses traditional model scaling in accuracy/energy efficiency, with notable gains in tasks demanding complex reasoning rather than mere factual recall. Further, we identify a critical interaction between TTC performance and output sequence length, demonstrating that strategically adjusting compute resources at inference time according to query complexity can substantially enhance efficiency. Our findings advocate for TTC as a promising direction, enabling more sustainable, accurate, and adaptable deployment of future language models without incurring additional pretraining costs.
Authors: Huiliang Zhang, Di Wu, Arnaud Zinflou, Stephane Dellacherie, Mouhamadou Makhtar Dione, Benoit Boulet
Abstract: Multivariate Time Series (MTS) forecasting has a wide range of applications in both industry and academia. Recent advances in Spatial-Temporal Graph Neural Network (STGNN) have achieved great progress in modelling spatial-temporal correlations. Limited by computational complexity, most STGNNs for MTS forecasting focus primarily on short-term and local spatial-temporal dependencies. Although some recent methods attempt to incorporate univariate history into modeling, they still overlook crucial long-term spatial-temporal similarities and correlations across MTS, which are essential for accurate forecasting. To fill this gap, we propose a framework called the Long-term Multivariate History Representation (LMHR) Enhanced STGNN for MTS forecasting. Specifically, a Long-term History Encoder (LHEncoder) is adopted to effectively encode the long-term history into segment-level contextual representations and reduce point-level noise. A non-parametric Hierarchical Representation Retriever (HRetriever) is designed to include the spatial information in the long-term spatial-temporal dependency modelling and pick out the most valuable representations with no additional training. A Transformer-based Aggregator (TAggregator) selectively fuses the sparsely retrieved contextual representations based on the ranking positional embedding efficiently. Experimental results demonstrate that LMHR outperforms typical STGNNs by 10.72% on the average prediction horizons and state-of-the-art methods by 4.12% on several real-world datasets. Additionally, it consistently improves prediction accuracy by 9.8% on the top 10% of rapidly changing patterns across the datasets.
Authors: Heiko Oppel, Andreas Spilz, Michael Munz
Abstract: Denoising diffusion probabilistic models are able to generate synthetic sensor signals. The training process of such a model is controlled by a loss function which measures the difference between the noise that was added in the forward process and the noise that was predicted by the diffusion model. This enables the generation of realistic data. However, the randomness within the process and the loss function itself makes it difficult to estimate the quality of the data. Therefore, we examine multiple similarity metrics and adapt an existing metric to overcome this issue by monitoring the training and synthetisation process using those metrics. The adapted metric can even be fine-tuned on the input data to comply with the requirements of an underlying classification task. We were able to significantly reduce the amount of training epochs without a performance reduction in the classification task. An optimized training process not only saves resources, but also reduces the time for training generative models.
Authors: Kunyun Wang, Bohan Li, Kai Yu, Minyi Guo, Jieru Zhao
Abstract: Diffusion models have emerged as a powerful class of generative models across various modalities, including image, video, and audio synthesis. However, their deployment is often limited by significant inference latency, primarily due to the inherently sequential nature of the denoising process. While existing parallelization strategies attempt to accelerate inference by distributing computation across multiple devices, they typically incur high communication overhead, hindering deployment on commercial hardware. To address this challenge, we propose \textbf{ParaStep}, a novel parallelization method based on a reuse-then-predict mechanism that parallelizes diffusion inference by exploiting similarity between adjacent denoising steps. Unlike prior approaches that rely on layer-wise or stage-wise communication, ParaStep employs lightweight, step-wise communication, substantially reducing overhead. ParaStep achieves end-to-end speedups of up to \textbf{3.88}$\times$ on SVD, \textbf{2.43}$\times$ on CogVideoX-2b, and \textbf{6.56}$\times$ on AudioLDM2-large, while maintaining generation quality. These results highlight ParaStep as a scalable and communication-efficient solution for accelerating diffusion inference, particularly in bandwidth-constrained environments.
Authors: Hong Huang, Dapeng Wu
Abstract: Large language models (LLMs) have made exciting achievements across various domains, yet their deployment on resource-constrained personal devices remains hindered by the prohibitive computational and memory demands of task-specific fine-tuning. While quantization offers a pathway to efficiency, existing methods struggle to balance performance and overhead, either incurring high computational/memory costs or failing to address activation outliers, a critical bottleneck in quantized fine-tuning. To address these challenges, we propose the Outlier Spatial Stability Hypothesis (OSSH): During fine-tuning, certain activation outlier channels retain stable spatial positions across training iterations. Building on OSSH, we propose Quaff, a Quantized parameter-efficient fine-tuning framework for LLMs, optimizing low-precision activation representations through targeted momentum scaling. Quaff dynamically suppresses outliers exclusively in invariant channels using lightweight operations, eliminating full-precision weight storage and global rescaling while reducing quantization errors. Extensive experiments across ten benchmarks validate OSSH and demonstrate Quaff's efficacy. Specifically, on the GPQA reasoning benchmark, Quaff achieves a 1.73x latency reduction and 30% memory savings over full-precision fine-tuning while improving accuracy by 0.6% on the Phi-3 model, reconciling the triple trade-off between efficiency, performance, and deployability. By enabling consumer-grade GPU fine-tuning (e.g., RTX 2080 Super) without sacrificing model utility, Quaff democratizes personalized LLM deployment. The code is available at https://github.com/Little0o0/Quaff.git.
Authors: Varun Raaghav, Dimitrios Bikos, Antonio Rago, Francesca Toni, Maria Charalambides
Abstract: Composites are amongst the most important materials manufactured today, as evidenced by their use in countless applications. In order to establish the suitability of composites in specific applications, finite element (FE) modelling, a numerical method based on partial differential equations, is the industry standard for assessing their mechanical properties. However, FE modelling is exceptionally costly from a computational viewpoint, a limitation which has led to efforts towards applying AI models to this task. However, in these approaches: the chosen model architectures were rudimentary, feed-forward neural networks giving limited accuracy; the studies focus on predicting elastic mechanical properties, without considering material strength limits; and the models lacked transparency, hindering trustworthiness by users. In this paper, we show that convolutional neural networks (CNNs) equipped with methods from explainable AI (XAI) can be successfully deployed to solve this problem. Our approach uses customised CNNs trained on a dataset we generate using transverse tension tests in FE modelling to predict composites' mechanical properties, i.e., Young's modulus and yield strength. We show empirically that our approach achieves high accuracy, outperforming a baseline, ResNet-34, in estimating the mechanical properties. We then use SHAP and Integrated Gradients, two post-hoc XAI methods, to explain the predictions, showing that the CNNs use the critical geometrical features that influence the composites' behaviour, thus allowing engineers to verify that the models are trustworthy by representing the science of composites.
Authors: Zaifa Xue, Tao Zhang, Tuo Xu, Huaixin Liang, Le Gao
Abstract: GraphSAGE is a widely used graph neural network. The introduction of causal inference has improved its robust performance and named as Causal GraphSAGE. However, Causal GraphSAGE focuses on measuring causal weighting among individual nodes, but neglecting the cooperative relationships among sampling nodes as a whole. To address this issue, this paper proposes Cooperative Causal GraphSAGE (CoCa-GraphSAGE), which combines cooperative game theory with Causal GraphSAGE. Initially, a cooperative causal structure model is constructed in the case of cooperation based on the graph structure. Subsequently, Cooperative Causal sampling (CoCa-sampling) algorithm is proposed, employing the Shapley values to calculate the cooperative contribution based on causal weights of the nodes sets. CoCa-sampling guides the selection of nodes with significant cooperative causal effects during the neighborhood sampling process, thus integrating the selected neighborhood features under cooperative relationships, which takes the sampled nodes as a whole and generates more stable target node embeddings. Experiments on publicly available datasets show that the proposed method has comparable classification performance to the compared methods and outperforms under perturbations, demonstrating the robustness improvement by CoCa-sampling.
Authors: Maheak Dave, Aniket Kumar Singh, Aryan Pareek, Harshita Jha, Debasis Chaudhuri, Manish Pratap Singh
Abstract: Deep Neural Networks have achieved remarkable achievements across various domains, however balancing performance and generalization still remains a challenge while training these networks. In this paper, we propose a novel framework that uses a cyclic optimization strategy to concurrently optimize the model and its input data for better training, rethinking the traditional training paradigm. Central to our approach is Iterative Constructive Perturbation (ICP), which leverages the model's loss to iteratively perturb the input, progressively constructing an enhanced representation over some refinement steps. This ICP input is then fed back into the model to produce improved intermediate features, which serve as a target in a self-distillation framework against the original features. By alternately altering the model's parameters to the data and the data to the model, our method effectively addresses the gap between fitting and generalization, leading to enhanced performance. Extensive experiments demonstrate that our approach not only mitigates common performance bottlenecks in neural networks but also demonstrates significant improvements across training variations.
Authors: Yihong Tang, Menglin Kong, Lijun Sun
Abstract: Generating synthetic data that faithfully captures the statistical structure of real-world distributions is a fundamental challenge in data modeling. Classical approaches often depend on strong parametric assumptions or manual structural design and struggle in high-dimensional or heterogeneous domains. Recent progress in Large Language Models (LLMs) reveals their potential as flexible, high-dimensional priors over real-world distributions. However, when applied to data synthesis, standard LLM-based sampling is inefficient, constrained by fixed context limits, and fails to ensure statistical alignment. Given this, we introduce LLMSynthor, a general framework for data synthesis that transforms LLMs into structure-aware simulators guided by distributional feedback. LLMSynthor treats the LLM as a nonparametric copula simulator for modeling high-order dependencies and introduces LLM Proposal Sampling to generate grounded proposal distributions that improve sampling efficiency without requiring rejection. By minimizing discrepancies in the summary statistics space, the iterative synthesis loop aligns real and synthetic data while gradually uncovering and refining the latent generative structure. We evaluate LLMSynthor in both controlled and real-world settings using heterogeneous datasets in privacy-sensitive domains (e.g., e-commerce, population, and mobility) that encompass both structured and unstructured formats. The synthetic data produced by LLMSynthor shows high statistical fidelity, practical utility, and cross-data adaptability, positioning it as a valuable tool across economics, social science, urban studies, and beyond.
Authors: Chih-Yu Chang, Milad Azvar, Chinedum Okwudire, Raed Al Kontar
Abstract: Bayesian optimization (BO) is a sequential decision-making tool widely used for optimizing expensive black-box functions. Recently, Large Language Models (LLMs) have shown remarkable adaptability in low-data regimes, making them promising tools for black-box optimization by leveraging contextual knowledge to propose high-quality query points. However, relying solely on LLMs as optimization agents introduces risks due to their lack of explicit surrogate modeling and calibrated uncertainty, as well as their inherently opaque internal mechanisms. This structural opacity makes it difficult to characterize or control the exploration-exploitation trade-off, ultimately undermining theoretical tractability and reliability. To address this, we propose LLINBO: LLM-in-the-Loop BO, a hybrid framework for BO that combines LLMs with statistical surrogate experts (e.g., Gaussian Processes (GP)). The core philosophy is to leverage contextual reasoning strengths of LLMs for early exploration, while relying on principled statistical models to guide efficient exploitation. Specifically, we introduce three mechanisms that enable this collaboration and establish their theoretical guarantees. We end the paper with a real-life proof-of-concept in the context of 3D printing. The code to reproduce the results can be found at https://github.com/UMDataScienceLab/LLM-in-the-Loop-BO.
URLs: https://github.com/UMDataScienceLab/LLM-in-the-Loop-BO.
Authors: Orhun Vural, Bunyamin Ozaydin, Khalid Y. Aram, James Booth, Brittany F. Lindsey, Abdulaziz Ahmed
Abstract: This study develops deep learning models to forecast the number of patients in the emergency department (ED) boarding phase six hours in advance, aiming to support proactive operational decision-making using only non-clinical, operational, and contextual features. Data were collected from five sources: ED tracking systems, inpatient census records, weather reports, federal holiday calendars, and local event schedules. After feature engineering, the data were aggregated at an hourly level, cleaned, and merged into a unified dataset for model training. Several time series deep learning models, including ResNetPlus, TSTPlus, TSiTPlus (from the tsai library), and N-BEATSx, were trained using Optuna and grid search for hyperparameter tuning. The average ED boarding count was 28.7, with a standard deviation of 11.2. N-BEATSx achieved the best performance, with a mean absolute error of 2.10, mean squared error of 7.08, root mean squared error of 2.66, and a coefficient of determination of 0.95. The model maintained stable accuracy even during periods of extremely high boarding counts, defined as values exceeding one, two, or three standard deviations above the mean. Results show that accurate six-hour-ahead forecasts are achievable without using patient-level clinical data. While strong performance was observed even with a basic feature set, the inclusion of additional features improved prediction stability under extreme conditions. This framework offers a practical and generalizable approach for hospital systems to anticipate boarding levels and help mitigate ED overcrowding.
Authors: Ben Cohen, Emaad Khwaja, Youssef Doubli, Salahidine Lemaachi, Chris Lettieri, Charles Masson, Hugo Miccinilli, Elise Ram\'e, Qiqi Ren, Afshin Rostamizadeh, Jean Ogier du Terrail, Anna-Monica Toon, Kan Wang, Stephan Xie, David Asker, Ameet Talwalkar, Othmane Abou-Amal
Abstract: We introduce Toto, a time series forecasting foundation model with 151 million parameters. Toto uses a modern decoder-only architecture coupled with architectural innovations designed to account for specific challenges found in multivariate observability time series data. Toto's pre-training corpus is a mixture of observability data, open datasets, and synthetic data, and is 4-10$\times$ larger than those of leading time series foundation models. Additionally, we introduce BOOM, a large-scale benchmark consisting of 350 million observations across 2,807 real-world time series. For both Toto and BOOM, we source observability data exclusively from Datadog's own telemetry and internal observability metrics. Extensive evaluations demonstrate that Toto achieves state-of-the-art performance on both BOOM and on established general purpose time series forecasting benchmarks. Toto's model weights, inference code, and evaluation scripts, as well as BOOM's data and evaluation code, are all available as open source under the Apache 2.0 License available at https://huggingface.co/Datadog/Toto-Open-Base-1.0 and https://github.com/DataDog/toto.
URLs: https://huggingface.co/Datadog/Toto-Open-Base-1.0, https://github.com/DataDog/toto.
Authors: Mingquan Feng, Yixin Huang, Yifan Fu, Shaobo Wang, Junchi Yan
Abstract: The design of optimization algorithms for neural networks remains a critical challenge, with most existing methods relying on heuristic adaptations of gradient-based approaches. This paper introduces KO (Kinetics-inspired Optimizer), a novel neural optimizer inspired by kinetic theory and partial differential equation (PDE) simulations. We reimagine the training dynamics of network parameters as the evolution of a particle system governed by kinetic principles, where parameter updates are simulated via a numerical scheme for the Boltzmann transport equation (BTE) that models stochastic particle collisions. This physics-driven approach inherently promotes parameter diversity during optimization, mitigating the phenomenon of parameter condensation, i.e. collapse of network parameters into low-dimensional subspaces, through mechanisms analogous to thermal diffusion in physical systems. We analyze this property, establishing both a mathematical proof and a physical interpretation. Extensive experiments on image classification (CIFAR-10/100, ImageNet) and text classification (IMDB, Snips) tasks demonstrate that KO consistently outperforms baseline optimizers (e.g., Adam, SGD), achieving accuracy improvements while computation cost remains comparable.
Authors: Iman Kazemian, Paritosh Ramanan, Murat Yildirim
Abstract: Data engineering pipelines are essential - albeit costly - components of predictive analytics frameworks requiring significant engineering time and domain expertise for carrying out tasks such as data ingestion, preprocessing, feature extraction, and feature engineering. In this paper, we propose ADEPT, an automated data engineering pipeline via text embeddings. At the core of the ADEPT framework is a simple yet powerful idea that the entropy of embeddings corresponding to textually dense raw format representation of time series can be intuitively viewed as equivalent (or in many cases superior) to that of numerically dense vector representations obtained by data engineering pipelines. Consequently, ADEPT uses a two step approach that (i) leverages text embeddings to represent the diverse data sources, and (ii) constructs a variational information bottleneck criteria to mitigate entropy variance in text embeddings of time series data. ADEPT provides an end-to-end automated implementation of predictive models that offers superior predictive performance despite issues such as missing data, ill-formed records, improper or corrupted data formats and irregular timestamps. Through exhaustive experiments, we show that the ADEPT outperforms the best existing benchmarks in a diverse set of datasets from large-scale applications across healthcare, finance, science and industrial internet of things. Our results show that ADEPT can potentially leapfrog many conventional data pipeline steps thereby paving the way for efficient and scalable automation pathways for diverse data science applications.
Authors: Yu Liu, Weiyao Tao, Tong Xia, Simon Knight, Tingting Zhu
Abstract: Survival analysis, which estimates the probability of event occurrence over time from censored data, is fundamental in numerous real-world applications, particularly in high-stakes domains such as healthcare and risk assessment. Despite advances in numerous survival models, quantifying the uncertainty of predictions from these models remains underexplored and challenging. The lack of reliable uncertainty quantification limits the interpretability and trustworthiness of survival models, hindering their adoption in clinical decision-making and other sensitive applications. To bridge this gap, in this work, we introduce SurvUnc, a novel meta-model based framework for post-hoc uncertainty quantification for survival models. SurvUnc introduces an anchor-based learning strategy that integrates concordance knowledge into meta-model optimization, leveraging pairwise ranking performance to estimate uncertainty effectively. Notably, our framework is model-agnostic, ensuring compatibility with any survival model without requiring modifications to its architecture or access to its internal parameters. Especially, we design a comprehensive evaluation pipeline tailored to this critical yet overlooked problem. Through extensive experiments on four publicly available benchmarking datasets and five representative survival models, we demonstrate the superiority of SurvUnc across multiple evaluation scenarios, including selective prediction, misprediction detection, and out-of-domain detection. Our results highlight the effectiveness of SurvUnc in enhancing model interpretability and reliability, paving the way for more trustworthy survival predictions in real-world applications.
Authors: Rushit N. Shah, Nikolaos Agadakos, Synthia Sasulski, Ali Farajzadeh, Sanjiban Choudhury, Brian Ziebart
Abstract: Imitation learning often assumes that demonstrations are close to optimal according to some fixed, but unknown, cost function. However, according to satisficing theory, humans often choose acceptable behavior based on their personal (and potentially dynamic) levels of aspiration, rather than achieving (near-) optimality. For example, a lunar lander demonstration that successfully lands without crashing might be acceptable to a novice despite being slow or jerky. Using a margin-based objective to guide deep reinforcement learning, our focused satisficing approach to imitation learning seeks a policy that surpasses the demonstrator's aspiration levels -- defined over trajectories or portions of trajectories -- on unseen demonstrations without explicitly learning those aspirations. We show experimentally that this focuses the policy to imitate the highest quality (portions of) demonstrations better than existing imitation learning methods, providing much higher rates of guaranteed acceptability to the demonstrator, and competitive true returns on a range of environments.
Authors: Runze Zhao, Yue Yu, Adams Yiyue Zhu, Chen Yang, Dongruo Zhou
Abstract: Continuous-time reinforcement learning (CTRL) provides a principled framework for sequential decision-making in environments where interactions evolve continuously over time. Despite its empirical success, the theoretical understanding of CTRL remains limited, especially in settings with general function approximation. In this work, we propose a model-based CTRL algorithm that achieves both sample and computational efficiency. Our approach leverages optimism-based confidence sets to establish the first sample complexity guarantee for CTRL with general function approximation, showing that a near-optimal policy can be learned with a suboptimality gap of $\tilde{O}(\sqrt{d_{\mathcal{R}} + d_{\mathcal{F}}}N^{-1/2})$ using $N$ measurements, where $d_{\mathcal{R}}$ and $d_{\mathcal{F}}$ denote the distributional Eluder dimensions of the reward and dynamic functions, respectively, capturing the complexity of general function approximation in reinforcement learning. Moreover, we introduce structured policy updates and an alternative measurement strategy that significantly reduce the number of policy updates and rollouts while maintaining competitive sample efficiency. We implemented experiments to backup our proposed algorithms on continuous control tasks and diffusion model fine-tuning, demonstrating comparable performance with significantly fewer policy updates and rollouts.
Authors: Marios Andreou, Nan Chen, Erik Bollt
Abstract: Causal inference determines cause-and-effect relationships between variables and has broad applications across disciplines. Traditional time-series methods often reveal causal links only in a time-averaged sense, while ensemble-based information transfer approaches detect the time evolution of short-term causal relationships but are typically limited to low-dimensional systems. In this paper, a new causal inference framework, called assimilative causal inference (ACI), is developed. Fundamentally different from the state-of-the-art methods, ACI uses a dynamical system and a single realization of a subset of the state variables to identify instantaneous causal relationships and the dynamic evolution of the associated causal influence range (CIR). Instead of quantifying how causes influence effects as done traditionally, ACI solves an inverse problem via Bayesian data assimilation, thus tracing causes backward from observed effects with an implicit Bayesian hypothesis. Causality is determined by assessing whether incorporating the information of the effect variables reduces the uncertainty in recovering the potential cause variables. ACI has several desirable features. First, it captures the dynamic interplay of variables, where their roles as causes and effects can shift repeatedly over time. Second, a mathematically justified objective criterion determines the CIR without empirical thresholds. Third, ACI is scalable to high-dimensional problems by leveraging computationally efficient Bayesian data assimilation techniques. Finally, ACI applies to short time series and incomplete datasets. Notably, ACI does not require observations of candidate causes, which is a key advantage since potential drivers are often unknown or unmeasured. The effectiveness of ACI is demonstrated by complex dynamical systems showcasing intermittency and extreme events.
Authors: Rohan Deb, Kiran Thekumparampil, Kousha Kalantari, Gaurush Hiranandani, Shoham Sabach, Branislav Kveton
Abstract: Supervised fine-tuning (SFT) is a standard approach to adapting large language models (LLMs) to new domains. In this work, we improve the statistical efficiency of SFT by selecting an informative subset of training examples. Specifically, for a fixed budget of training examples, which determines the computational cost of fine-tuning, we determine the most informative ones. The key idea in our method is to select examples that maximize information gain, measured by the Hessian of the log-likelihood of the LLM. We approximate it efficiently by linearizing the LLM at the last layer using multinomial logistic regression models. Our approach is computationally efficient, analyzable, and performs well empirically. We demonstrate this on several problems, and back our claims with both quantitative results and an LLM evaluation.
Authors: Juan Nathaniel, Carla Roesch, Jatan Buch, Derek DeSantis, Adam Rupe, Kara Lamb, Pierre Gentine
Abstract: We use a deep Koopman operator-theoretic formalism to develop a novel causal discovery algorithm, Kausal. Causal discovery aims to identify cause-effect mechanisms for better scientific understanding, explainable decision-making, and more accurate modeling. Standard statistical frameworks, such as Granger causality, lack the ability to quantify causal relationships in nonlinear dynamics due to the presence of complex feedback mechanisms, timescale mixing, and nonstationarity. This presents a challenge in studying many real-world systems, such as the Earth's climate. Meanwhile, Koopman operator methods have emerged as a promising tool for approximating nonlinear dynamics in a linear space of observables. In Kausal, we propose to leverage this powerful idea for causal analysis where optimal observables are inferred using deep learning. Causal estimates are then evaluated in a reproducing kernel Hilbert space, and defined as the distance between the marginal dynamics of the effect and the joint dynamics of the cause-effect observables. Our numerical experiments demonstrate Kausal's superior ability in discovering and characterizing causal signals compared to existing approaches of prescribed observables. Lastly, we extend our analysis to observations of El Ni\~no-Southern Oscillation highlighting our algorithm's applicability to real-world phenomena. Our code is available at https://github.com/juannat7/kausal.
Authors: Shreya Gupta, Boyang Huang, Barna Saha, Yinzhan Xu, Christopher Ye
Abstract: Despite the popularity of the Transformer architecture, the standard algorithm for computing Attention suffers from quadratic time complexity in context length $n$. Alman and Song [NeurIPS 2023] showed that when the head dimension $d = \Theta(\log n)$, subquadratic Attention is possible if and only if the inputs have small entries bounded by $B = o(\sqrt{\log n})$ in absolute values, under the Strong Exponential Time Hypothesis ($\mathsf{SETH}$). Equivalently, subquadratic Attention is possible if and only if the softmax is applied with high temperature for $d=\Theta(\log n)$. Running times of these algorithms depend exponentially on $B$ and thus they do not lead to even a polynomial-time algorithm outside the specific range of $B$. This naturally leads to the question: when can Attention be computed efficiently without strong assumptions on temperature? Are there fast attention algorithms that scale polylogarithmically with entry size $B$? In this work, we resolve this question and characterize when fast Attention for arbitrary temperatures is possible. First, for all constant $d = O(1)$, we give the first subquadratic $\tilde{O}(n^{2 - 1/d} \cdot \mathrm{polylog}(B))$ time algorithm for Attention with large $B$. Our result holds even for matrices with large head dimension if they have low rank. In this regime, we also give a similar running time for Attention gradient computation, and therefore for the full LLM training process. Furthermore, we show that any substantial improvement on our algorithm is unlikely. In particular, we show that even when $d = 2^{\Theta(\log^* n)}$, Attention requires $n^{2 - o(1)}$ time under $\mathsf{SETH}$. Finally, in the regime where $d = \mathrm{poly}(n)$, we show that the standard algorithm is optimal under popular fine-grained complexity assumptions.
Authors: Francisco P\'erez-Galarce, Jorge Mart\'inez-Palomera, Karim Pichara, Pablo Huijse, M\'arcio Catelan
Abstract: Over the last two decades, machine learning models have been widely applied and have proven effective in classifying variable stars, particularly with the adoption of deep learning architectures such as convolutional neural networks, recurrent neural networks, and transformer models. While these models have achieved high accuracy, they require high-quality, representative data and a large number of labelled samples for each star type to generalise well, which can be challenging in time-domain surveys. This challenge often leads to models learning and reinforcing biases inherent in the training data, an issue that is not easily detectable when validation is performed on subsamples from the same catalogue. The problem of biases in variable star data has been largely overlooked, and a definitive solution has yet to be established. In this paper, we propose a new approach to improve the reliability of classifiers in variable star classification by introducing a self-regulated training process. This process utilises synthetic samples generated by a physics-enhanced latent space variational autoencoder, incorporating six physical parameters from Gaia Data Release 3. Our method features a dynamic interaction between a classifier and a generative model, where the generative model produces ad-hoc synthetic light curves to reduce confusion during classifier training and populate underrepresented regions in the physical parameter space. Experiments conducted under various scenarios demonstrate that our self-regulated training approach outperforms traditional training methods for classifying variable stars on biased datasets, showing statistically significant improvements.
Authors: Abdellah Aznag, Rachel Cummings, Adam N. Elmachtoub
Abstract: We study a fundamental learning problem over multiple groups with unknown data distributions, where an analyst would like to learn the mean of each group. Moreover, we want to ensure that this data is collected in a relatively fair manner such that the noise of the estimate of each group is reasonable. In particular, we focus on settings where data are collected dynamically, which is important in adaptive experimentation for online platforms or adaptive clinical trials for healthcare. In our model, we employ an active learning framework to sequentially collect samples with bandit feedback, observing a sample in each period from the chosen group. After observing a sample, the analyst updates their estimate of the mean and variance of that group and chooses the next group accordingly. The analyst's objective is to dynamically collect samples to minimize the collective noise of the estimators, measured by the norm of the vector of variances of the mean estimators. We propose an algorithm, Variance-UCB, that sequentially selects groups according to an upper confidence bound on the variance estimate. We provide a general theoretical framework for providing efficient bounds on learning from any underlying distribution where the variances can be estimated reasonably. This framework yields upper bounds on regret that improve significantly upon all existing bounds, as well as a collection of new results for different objectives and distributions than those previously studied.
Authors: Susav Shrestha, Brad Settlemyer, Nikoli Dryden, Narasimha Reddy
Abstract: Accelerating large language model (LLM) inference is critical for real-world deployments requiring high throughput and low latency. Contextual sparsity, where each token dynamically activates only a small subset of the model parameters, shows promise but does not scale to large batch sizes due to union of active neurons quickly approaching dense computation. We introduce Polar Sparsity, highlighting a key shift in sparsity importance from MLP to Attention layers as we scale batch size and sequence length. While MLP layers become more compute-efficient under batching, their sparsity vanishes. In contrast, attention becomes increasingly more expensive at scale, while their head sparsity remains stable and batch-invariant. We develop hardware-efficient, sparsity-aware GPU kernels for selective MLP and Attention computations, delivering up to \(2.2\times\) end-to-end speedups for models like OPT, LLaMA-2 \& 3, across various batch sizes and sequence lengths without compromising accuracy. To our knowledge, this is the first work to demonstrate that contextual sparsity can scale effectively to large batch sizes, delivering substantial inference acceleration with minimal changes, making Polar Sparsity practical for large-scale, high-throughput LLM deployment systems. Our code is available at: https://github.com/susavlsh10/Polar-Sparsity.
Authors: Hootan Mahmoodiyan, Maryam Ahang, Mostafa Abbasi, Homayoun Najjaran
Abstract: Ensuring the reliable operation of power transformers is critical to grid stability. Dissolved Gas Analysis (DGA) is widely used for fault diagnosis, but traditional methods rely on heuristic rules, which may lead to inconsistent results. Machine learning (ML)-based approaches have improved diagnostic accuracy; however, power transformers operate under varying conditions, and differences in transformer type, environmental factors, and operational settings create distribution shifts in diagnostic data. Consequently, direct model transfer between transformers often fails, making techniques for domain adaptation a necessity. To tackle this issue, this work proposes a feature-weighted domain adaptation technique that combines Maximum Mean Discrepancy (MMD) and Correlation Alignment (CORAL) with feature-specific weighting (MCW). Kolmogorov-Smirnov (K-S) statistics are used to assign adaptable weights, prioritizing features with larger distributional discrepancies and thereby improving source and target domain alignment. Experimental evaluations on datasets for power transformers demonstrate the effectiveness of the proposed method, which achieves a 7.9% improvement over Fine-Tuning and a 2.2% improvement over MMD-CORAL (MC). Furthermore, it outperforms both techniques across various training sample sizes, confirming its robustness for domain adaptation.
Authors: Ali Mohajerzarrinkelk, Maryam Ahang, Mehran Zoravar, Mostafa Abbasi, Homayoun Najjaran
Abstract: Precise estimation of the Remaining Useful Life (RUL) of rolling bearings is an important consideration to avoid unexpected failures, reduce downtime, and promote safety and efficiency in industrial systems. Complications in degradation trends, noise presence, and the necessity to detect faults in advance make estimation of RUL a challenging task. This paper introduces a novel framework that combines wavelet-based denoising method, Wavelet Packet Decomposition (WPD), and a customized multi-channel Swin Transformer model (MCSFormer) to address these problems. With attention mechanisms incorporated for feature fusion, the model is designed to learn global and local degradation patterns utilizing hierarchical representations for enhancing predictive performance. Additionally, a customized loss function is developed as a key distinction of this work to differentiate between early and late predictions, prioritizing accurate early detection and minimizing the high operation risks of late predictions. The proposed model was evaluated with the PRONOSTIA dataset using three experiments. Intra-condition experiments demonstrated that MCSFormer outperformed state-of-the-art models, including the Adaptive Transformer, MDAN, and CNN-SRU, achieving 41%, 64%, and 69% lower MAE on average across different operating conditions, respectively. In terms of cross-condition testing, it achieved superior generalization under varying operating conditions compared to the adapted ViT and Swin Transformer. Lastly, the custom loss function effectively reduced late predictions, as evidenced in a 6.3% improvement in the scoring metric while maintaining competitive overall performance. The model's robust noise resistance, generalization capability, and focus on safety make MCSFormer a trustworthy and effective predictive maintenance tool in industrial applications.
Authors: Regol Florence, Schwinn Leo, Sprague Kyle, Coates Mark, Markovich Thomas
Abstract: A significant challenge in maintaining real-world machine learning models is responding to the continuous and unpredictable evolution of data. Most practitioners are faced with the difficult question: when should I retrain or update my machine learning model? This seemingly straightforward problem is particularly challenging for three reasons: 1) decisions must be made based on very limited information - we usually have access to only a few examples, 2) the nature, extent, and impact of the distribution shift are unknown, and 3) it involves specifying a cost ratio between retraining and poor performance, which can be hard to characterize. Existing works address certain aspects of this problem, but none offer a comprehensive solution. Distribution shift detection falls short as it cannot account for the cost trade-off; the scarcity of the data, paired with its unusual structure, makes it a poor fit for existing offline reinforcement learning methods, and the online learning formulation overlooks key practical considerations. To address this, we present a principled formulation of the retraining problem and propose an uncertainty-based method that makes decisions by continually forecasting the evolution of model performance evaluated with a bounded metric. Our experiments addressing classification tasks show that the method consistently outperforms existing baselines on 7 datasets.
Authors: Frederik Wenkel, Wilson Tu, Cassandra Masschelein, Hamed Shirzad, Cian Eastwood, Shawn T. Whitfield, Ihab Bendidi, Craig Russell, Liam Hodgson, Yassir El Mesbahi, Jiarui Ding, Marta M. Fay, Berton Earnshaw, Emmanuel Noutahi, Alisandra K. Denton
Abstract: Accurately predicting cellular responses to genetic perturbations is essential for understanding disease mechanisms and designing effective therapies. Yet exhaustively exploring the space of possible perturbations (e.g., multi-gene perturbations or across tissues and cell types) is prohibitively expensive, motivating methods that can generalize to unseen conditions. In this work, we explore how knowledge graphs of gene-gene relationships can improve out-of-distribution (OOD) prediction across three challenging settings: unseen single perturbations; unseen double perturbations; and unseen cell lines. In particular, we present: (i) TxPert, a new state-of-the-art method that leverages multiple biological knowledge networks to predict transcriptional responses under OOD scenarios; (ii) an in-depth analysis demonstrating the impact of graphs, model architecture, and data on performance; and (iii) an expanded benchmarking framework that strengthens evaluation standards for perturbation modeling.
Authors: Xuefeng Du
Abstract: Ensuring the reliability and safety of machine learning models in open-world deployment is a central challenge in AI safety. This thesis develops both algorithmic and theoretical foundations to address key reliability issues arising from distributional uncertainty and unknown classes, from standard neural networks to modern foundation models like large language models (LLMs). Traditional learning paradigms, such as empirical risk minimization (ERM), assume no distribution shift between training and inference, often leading to overconfident predictions on out-of-distribution (OOD) inputs. This thesis introduces novel frameworks that jointly optimize for in-distribution accuracy and reliability to unseen data. A core contribution is the development of an unknown-aware learning framework that enables models to recognize and handle novel inputs without labeled OOD data. We propose new outlier synthesis methods, VOS, NPOS, and DREAM-OOD, to generate informative unknowns during training. Building on this, we present SAL, a theoretical and algorithmic framework that leverages unlabeled in-the-wild data to enhance OOD detection under realistic deployment conditions. These methods demonstrate that abundant unlabeled data can be harnessed to recognize and adapt to unforeseen inputs, providing formal reliability guarantees. The thesis also extends reliable learning to foundation models. We develop HaloScope for hallucination detection in LLMs, MLLMGuard for defending against malicious prompts in multimodal models, and data cleaning methods to denoise human feedback used for better alignment. These tools target failure modes that threaten the safety of large-scale models in deployment. Overall, these contributions promote unknown-aware learning as a new paradigm, and we hope it can advance the reliability of AI systems with minimal human efforts.
Authors: Ross Nordby
Abstract: To help evaluate and understand the latent capabilities of language models, this paper introduces an approach using optimized input embeddings, or 'soft prompts,' as a metric of conditional distance between a model and a target behavior. The technique aims to facilitate latent capability discovery as a part of automated red teaming/evaluation suites and to provide quantitative feedback about the accessibility of potentially concerning behaviors in a way that may scale to powerful future models, including those which may otherwise be capable of deceptive alignment. An evaluation framework using soft prompts is demonstrated in natural language, chess, and pathfinding, and the technique is extended with generalized conditional soft prompts to aid in constructing task evaluations.
Authors: O. Deniz Kose, Gonzalo Mateos, Yanning Shen
Abstract: The growing enforcement of the right to be forgotten regulations has propelled recent advances in certified (graph) unlearning strategies to comply with data removal requests from deployed machine learning (ML) models. Motivated by the well-documented bias amplification predicament inherent to graph data, here we take a fresh look at graph unlearning and leverage it as a bias mitigation tool. Given a pre-trained graph ML model, we develop a training-free unlearning procedure that offers certifiable bias mitigation via a single-step Newton update on the model weights. This way, we contribute a computationally lightweight alternative to the prevalent training- and optimization-based fairness enhancement approaches, with quantifiable performance guarantees. We first develop a novel fairness-aware nodal feature unlearning strategy along with refined certified unlearning bounds for this setting, whose impact extends beyond the realm of graph unlearning. We then design structural unlearning methods endowed with principled selection mechanisms over nodes and edges informed by rigorous bias analyses. Unlearning these judiciously selected elements can mitigate algorithmic biases with minimal impact on downstream utility (e.g., node classification accuracy). Experimental results over real networks corroborate the bias mitigation efficacy of our unlearning strategies, and delineate markedly favorable utility-complexity trade-offs relative to retraining from scratch using augmented graph data obtained via removals.
Authors: Kungang Li, Xiangyi Chen, Ling Leng, Jiajing Xu, Jiankai Sun, Behnam Rezaei
Abstract: In the realm of online advertising, accurately predicting the conversion rate (CVR) is crucial for enhancing advertising efficiency and user satisfaction. This paper addresses the challenge of CVR prediction while adhering to user privacy preferences and advertiser requirements. Traditional methods face obstacles such as the reluctance of advertisers to share sensitive conversion data and the limitations of model training in secure environments like data clean rooms. We propose a novel model training framework that enables collaborative model training without sharing sample-level gradients with the advertising platform. Our approach introduces several innovative components: (1) utilizing batch-level aggregated gradients instead of sample-level gradients to minimize privacy risks; (2) applying adapter-based parameter-efficient fine-tuning and gradient compression to reduce communication costs; and (3) employing de-biasing techniques to train the model under label differential privacy, thereby maintaining accuracy despite privacy-enhanced label perturbations. Our experimental results, conducted on industrial datasets, demonstrate that our method achieves competitive ROCAUC performance while significantly decreasing communication overhead and complying with both advertiser privacy requirements and user privacy choices. This framework establishes a new standard for privacy-preserving, high-performance CVR prediction in the digital advertising landscape.
Authors: Dave Cook, Tim Klawa
Abstract: AI systems in high-consequence domains such as defense, intelligence, and disaster response must detect rare, high-impact events while operating under tight resource constraints. Traditional annotation strategies that prioritize label volume over informational value introduce redundancy and noise, limiting model generalization. This paper introduces smart-sizing, a training data strategy that emphasizes label diversity, model-guided selection, and marginal utility-based stopping. We implement this through Adaptive Label Optimization (ALO), combining pre-labeling triage, annotator disagreement analysis, and iterative feedback to prioritize labels that meaningfully improve model performance. Experiments show that models trained on 20 to 40 percent of curated data can match or exceed full-data baselines, particularly in rare-class recall and edge-case generalization. We also demonstrate how latent labeling errors embedded in training and validation sets can distort evaluation, underscoring the need for embedded audit tools and performance-aware governance. Smart-sizing reframes annotation as a feedback-driven process aligned with mission outcomes, enabling more robust models with fewer labels and supporting efficient AI development pipelines for frontier models and operational systems.
Authors: Fangzhen Zhao, Chenyi Zhang, Naipeng Dong, Ming Li, Jinxiao Shan
Abstract: Deep neural networks (DNNs) are notoriously hard to understand and difficult to defend. Extracting representative paths (including the neuron activation values and the connections between neurons) from DNNs using software engineering approaches has recently shown to be a promising approach in interpreting the decision making process of blackbox DNNs, as the extracted paths are often effective in capturing essential features. With this in mind, this work investigates a novel approach that extracts critical paths from DNNs and subsequently applies the extracted paths for the anomaly detection task, based on the observation that outliers and adversarial inputs do not usually induce the same activation pattern on those paths as normal (in-distribution) inputs. In our approach, we first identify critical detection paths via genetic evolution and mutation. Since different paths in a DNN often capture different features for the same target class, we ensemble detection results from multiple paths by integrating random subspace sampling and a voting mechanism. Compared with state-of-the-art methods, our experimental results suggest that our method not only outperforms them, but it is also suitable for the detection of a broad range of anomaly types with high accuracy.
Authors: Yangchao Wu, Zongyue Qin, Alex Wong, Stefano Soatto
Abstract: Speculative decoding is a technique to leverage hardware concurrency to improve the efficiency of large-scale autoregressive (AR) Transformer models by enabling multiple steps of token generation in a single forward pass. State-space models (SSMs) are already more efficient than AR Transformers, since their state summarizes all past data with no need to cache or re-process tokens in the sliding window context. However, their state can also comprise thousands of tokens; so, speculative decoding has recently been extended to SSMs. Existing approaches, however, do not leverage the tree-based verification methods, since current SSMs lack the means to compute a token tree efficiently. We propose the first scalable algorithm to perform tree-based speculative decoding in state-space models (SSMs) and hybrid architectures of SSMs and Transformer layers. We exploit the structure of accumulated state transition matrices to facilitate tree-based speculative decoding with minimal overhead to current SSM state update implementations. With the algorithm, we describe a hardware-aware implementation that improves naive application of AR Transformer tree-based speculative decoding methods to SSMs. Furthermore, we outperform vanilla speculative decoding with SSMs even with a baseline drafting model and tree structure on three different benchmarks, opening up opportunities for further speed up with SSM and hybrid model inference. Code will be released upon paper acceptance.
Authors: John L. Zhou, Jonathan C. Kao
Abstract: Offline goal-conditioned reinforcement learning (GCRL) is a promising approach for pretraining generalist policies on large datasets of reward-free trajectories, akin to the self-supervised objectives used to train foundation models for computer vision and natural language processing. However, scaling GCRL to longer horizons remains challenging due to the combination of sparse rewards and discounting, which obscures the comparative advantages of primitive actions with respect to distant goals. Hierarchical RL methods achieve strong empirical results on long-horizon goal-reaching tasks, but their reliance on modular, timescale-specific policies and subgoal generation introduces significant additional complexity and hinders scaling to high-dimensional goal spaces. In this work, we introduce an algorithm to train a flat (non-hierarchical) goal-conditioned policy by bootstrapping on subgoal-conditioned policies with advantage-weighted importance sampling. Our approach eliminates the need for a generative model over the (sub)goal space, which we find is key for scaling to high-dimensional control in large state spaces. We further show that existing hierarchical and bootstrapping-based approaches correspond to specific design choices within our derivation. Across a comprehensive suite of state- and pixel-based locomotion and manipulation benchmarks, our method matches or surpasses state-of-the-art offline GCRL algorithms and scales to complex, long-horizon tasks where prior approaches fail.
Authors: Eric Hanchen Jiang, Haozheng Luo, Shengyuan Pang, Xiaomin Li, Zhenting Qi, Hengli Li, Cheng-Fu Yang, Zongyu Lin, Xinfeng Li, Hao Xu, Kai-Wei Chang, Ying Nian Wu
Abstract: Mathematical reasoning presents a significant challenge for Large Language Models (LLMs), often requiring robust multi step logical consistency. While Chain of Thought (CoT) prompting elicits reasoning steps, it doesn't guarantee correctness, and improving reliability via extensive sampling is computationally costly. This paper introduces the Energy Outcome Reward Model (EORM), an effective, lightweight, post hoc verifier. EORM leverages Energy Based Models (EBMs) to simplify the training of reward models by learning to assign a scalar energy score to CoT solutions using only outcome labels, thereby avoiding detailed annotations. It achieves this by interpreting discriminator output logits as negative energies, effectively ranking candidates where lower energy is assigned to solutions leading to correct final outcomes implicitly favoring coherent reasoning. On mathematical benchmarks (GSM8k, MATH), EORM significantly improves final answer accuracy (e.g., with Llama 3 8B, achieving 90.7% on GSM8k and 63.7% on MATH). EORM effectively leverages a given pool of candidate solutions to match or exceed the performance of brute force sampling, thereby enhancing LLM reasoning outcome reliability through its streamlined post hoc verification process.
Authors: Alvin Heng, Harold Soh
Abstract: Selective classification enhances the reliability of predictive models by allowing them to abstain from making uncertain predictions. In this work, we revisit the design of optimal selection functions through the lens of the Neyman--Pearson lemma, a classical result in statistics that characterizes the optimal rejection rule as a likelihood ratio test. We show that this perspective not only unifies the behavior of several post-hoc selection baselines, but also motivates new approaches to selective classification which we propose here. A central focus of our work is the setting of covariate shift, where the input distribution at test time differs from that at training. This realistic and challenging scenario remains relatively underexplored in the context of selective classification. We evaluate our proposed methods across a range of vision and language tasks, including both supervised learning and vision-language models. Our experiments demonstrate that our Neyman--Pearson-informed methods consistently outperform existing baselines, indicating that likelihood ratio-based selection offers a robust mechanism for improving selective classification under covariate shifts. Our code is publicly available at https://github.com/clear-nus/sc-likelihood-ratios.
Authors: Quan Nguyen, Thanh Nguyen-Tang
Abstract: We study the approximation capabilities and on-convergence behaviors of one-layer transformers on the noiseless and noisy in-context reasoning of next-token prediction. Existing theoretical results focus on understanding the in-context reasoning behaviors for either the first gradient step or when the number of samples is infinite. Furthermore, no convergence rates nor generalization abilities were known. Our work addresses these gaps by showing that there exists a class of one-layer transformers that are provably Bayes-optimal with both linear and ReLU attention. When being trained with gradient descent, we show via a finite-sample analysis that the expected loss of these transformers converges at linear rate to the Bayes risk. Moreover, we prove that the trained models generalize to unseen samples as well as exhibit learning behaviors that were empirically observed in previous works. Our theoretical findings are further supported by extensive empirical validations.
Authors: Longlong Li, Cunquan Qu, Guanghui Wang
Abstract: Conventional Graph Neural Networks (GNNs) aggregate neighbor embeddings as holistic vectors, lacking the ability to identify fine-grained, direction-specific feature relevance. We propose MSH-GNN (Multi-Scale Harmonic Graph Neural Network), a novel architecture that performs feature-wise adaptive message passing through node-specific harmonic projections. For each node, MSH-GNN dynamically projects neighbor features onto frequency-sensitive directions determined by the target node's own representation. These projections are further modulated using learnable sinusoidal encodings at multiple frequencies, enabling the model to capture both smooth and oscillatory structural patterns across scales. A frequency-aware attention pooling mechanism is introduced to emphasize spectrally and structurally salient nodes during readout. Theoretically, we prove that MSH-GNN approximates shift-invariant kernels and matches the expressive power of the 1-Weisfeiler-Lehman (1-WL) test. Empirically, MSH-GNN consistently outperforms state-of-the-art models on a wide range of graph and node classification tasks. Furthermore, in challenging classification settings involving joint variations in graph topology and spectral frequency, MSH-GNN excels at capturing structural asymmetries and high-frequency modulations, enabling more accurate graph discrimination.
Authors: Qingyu Song, Peiyu Liao, Wenqian Zhao, Yiwen Wang, Shoubo Hu, Hui-Ling Zhen, Ning Jiang, Mingxuan Yuan
Abstract: The increasing deployment of Large Language Models (LLMs) on edge devices, driven by model advancements and hardware improvements, offers significant privacy benefits. However, these on-device LLMs inherently face performance limitations due to reduced model capacity and necessary compression techniques. To address this, we introduce a systematic methodology -- encompassing model capability, development efficiency, and system resources -- for evaluating on-device LLMs. Our comprehensive evaluation, encompassing models from 0.5B to 14B parameters and seven post-training quantization (PTQ) methods on commodity laptops, yields several critical insights: 1) System-level metrics exhibit near-linear scaling with effective bits-per-weight (BPW). 2) A practical threshold exists around $\sim$3.5 effective BPW, larger models subjected to low-bit quantization consistently outperform smaller models utilizing higher bit-precision. 3) Quantization with low BPW incurs marginal accuracy loss but significant memory savings. 4) Determined by low-level implementation specifics power consumption on CPU, where computation-intensive operations spend more power than memory-intensive ones. These findings offer crucial insights and practical guidelines for the efficient deployment and optimized configuration of LLMs on resource-constrained edge devices. Our codebase is available at https://github.com/simmonssong/LLMOnDevice.
Authors: Kaiwen Zha, Zhengqi Gao, Maohao Shen, Zhang-Wei Hong, Duane S. Boning, Dina Katabi
Abstract: Reinforcement learning (RL) has recently emerged as a compelling approach for enhancing the reasoning capabilities of large language models (LLMs), where an LLM generator serves as a policy guided by a verifier (reward model). However, current RL post-training methods for LLMs typically use verifiers that are fixed (rule-based or frozen pretrained) or trained discriminatively via supervised fine-tuning (SFT). Such designs are susceptible to reward hacking and generalize poorly beyond their training distributions. To overcome these limitations, we propose Tango, a novel framework that uses RL to concurrently train both an LLM generator and a verifier in an interleaved manner. A central innovation of Tango is its generative, process-level LLM verifier, which is trained via RL and co-evolves with the generator. Importantly, the verifier is trained solely based on outcome-level verification correctness rewards without requiring explicit process-level annotations. This generative RL-trained verifier exhibits improved robustness and superior generalization compared to deterministic or SFT-trained verifiers, fostering effective mutual reinforcement with the generator. Extensive experiments demonstrate that both components of Tango achieve state-of-the-art results among 7B/8B-scale models: the generator attains best-in-class performance across five competition-level math benchmarks and four challenging out-of-domain reasoning tasks, while the verifier leads on the ProcessBench dataset. Remarkably, both components exhibit particularly substantial improvements on the most difficult mathematical reasoning problems. Code is at: https://github.com/kaiwenzha/rl-tango.
Authors: Ivan Smirnov, Shangding Gu
Abstract: Reinforcement learning (RL) has seen significant advancements through the application of various neural network architectures. In this study, we systematically investigate the performance of several neural networks in RL tasks, including Long Short-Term Memory (LSTM), Multi-Layer Perceptron (MLP), Mamba/Mamba-2, Transformer-XL, Gated Transformer-XL, and Gated Recurrent Unit (GRU). Through comprehensive evaluation across continuous control, discrete decision-making, and memory-based environments, we identify architecture-specific strengths and limitations. Our results reveal that: (1) MLPs excel in fully observable continuous control tasks, providing an optimal balance of performance and efficiency; (2) recurrent architectures like LSTM and GRU offer robust performance in partially observable environments with moderate memory requirements; (3) Mamba models achieve a 4.5x higher throughput compared to LSTM and a 3.9x increase over GRU, all while maintaining comparable performance; and (4) only Transformer-XL, Gated Transformer-XL, and Mamba-2 successfully solve the most challenging memory-intensive tasks, with Mamba-2 requiring 8x less memory than Transformer-XL. These findings provide insights for researchers and practitioners, enabling more informed architecture selection based on specific task characteristics and computational constraints. Code is available at: https://github.com/SafeRL-Lab/RLBenchNet
Authors: Yingming Pu, Tao Lin, Hongyu Chen
Abstract: Large Language Model (LLM)-based multi-agent systems (MAS) demonstrate remarkable potential for scientific discovery. Existing approaches, however, often automate scientific discovery using predefined workflows that lack rationality constraints. This often leads to aimless hypothesizing and a failure to consistently link hypotheses with evidence, thereby hindering systematic uncertainty reduction. Overcoming these limitations fundamentally requires systematic uncertainty reduction. We introduce \texttt{PiFlow}, an information-theoretical framework, treating automated scientific discovery as a structured uncertainty reduction problem guided by principles (e.g., scientific laws). In evaluations across three distinct scientific domains -- discovering nanomaterial structures, bio-molecules, and superconductor candidates with targeted properties -- our method significantly improves discovery efficiency, reflected by a 73.55\% increase in the Area Under the Curve (AUC) of property values versus exploration steps, and enhances solution quality by 94.06\% compared to a vanilla agent system. Overall, \texttt{PiFlow} serves as a Plug-and-Play method, establishing a novel paradigm shift in highly efficient automated scientific discovery, paving the way for more robust and accelerated AI-driven research. Code is publicly available at our \href{https://github.com/amair-lab/PiFlow}{GitHub}.
Authors: Sho Sonoda, Yuka Hashimoto, Isao Ishikawa, Masahiro Ikeda
Abstract: Recent theory has reduced the depth dependence of generalization bounds from exponential to polynomial and even depth-independent rates, yet these results remain tied to specific architectures and Euclidean inputs. We present a unified framework for arbitrary \blue{pseudo-metric} spaces in which a depth-\(k\) network is the composition of continuous hidden maps \(f:\mathcal{X}\to \mathcal{X}\) and an output map \(h:\mathcal{X}\to \mathbb{R}\). The resulting bound $O(\sqrt{(\alpha + \log \beta(k))/n})$ isolates the sole depth contribution in \(\beta(k)\), the word-ball growth of the semigroup generated by the hidden layers. By Gromov's theorem polynomial (resp. exponential) growth corresponds to virtually nilpotent (resp. expanding) dynamics, revealing a geometric dichotomy behind existing $O(\sqrt{k})$ (sublinear depth) and $\tilde{O}(1)$ (depth-independent) rates. We further provide covering-number estimates showing that expanding dynamics yield an exponential parameter saving via compositional expressivity. Our results decouple specification from implementation, offering architecture-agnostic and dynamical-systems-aware guarantees applicable to modern deep-learning paradigms such as test-time inference and diffusion models.
Authors: Xin Zhou, Weiqing Wang, Francisco J. Bald\'an, Wray Buntine, Christoph Bergmeir
Abstract: While multimodal data sources are increasingly available from real-world forecasting, most existing research remains on unimodal time series. In this work, we present MoTime, a suite of multimodal time series forecasting datasets that pair temporal signals with external modalities such as text, metadata, and images. Covering diverse domains, MoTime supports structured evaluation of modality utility under two scenarios: 1) the common forecasting task, where varying-length history is available, and 2) cold-start forecasting, where no historical data is available. Experiments show that external modalities can improve forecasting performance in both scenarios, with particularly strong benefits for short series in some datasets, though the impact varies depending on data characteristics. By making datasets and findings publicly available, we aim to support more comprehensive and realistic benchmarks in future multimodal time series forecasting research.
Authors: Nanxu Gong, Sixun Dong, Haoyue Bai, Xinyuan Wang, Wangyang Ying, Yanjie Fu
Abstract: As a widely-used and practical tool, feature engineering transforms raw data into discriminative features to advance AI model performance. However, existing methods usually apply feature selection and generation separately, failing to strive a balance between reducing redundancy and adding meaningful dimensions. To fill this gap, we propose an agentic feature augmentation concept, where the unification of feature generation and selection is modeled as agentic teaming and planning. Specifically, we develop a Multi-Agent System with Long and Short-Term Memory (MAGS), comprising a selector agent to eliminate redundant features, a generator agent to produce informative new dimensions, and a router agent that strategically coordinates their actions. We leverage in-context learning with short-term memory for immediate feedback refinement and long-term memory for globally optimal guidance. Additionally, we employ offline Proximal Policy Optimization (PPO) reinforcement fine-tuning to train the router agent for effective decision-making to navigate a vast discrete feature space. Extensive experiments demonstrate that this unified agentic framework consistently achieves superior task performance by intelligently orchestrating feature selection and generation.
Authors: Sergey Pankov, Georges Harik
Abstract: It is straightforward to design an unbiased gradient estimator that stochastically cuts the backpropagation flow through any part of a computational graph. By cutting the parts that have little effect on the computation, one can potentially save a significant amount of back-propagation computation in exchange for a minimal increase in the stochastic gradient variance, in some situations. Such a situation occurs in the attention mechanism of the transformer architecture. For long sequences, attention becomes the limiting factor, as its compute requirements increase quadratically with sequence length $n$. At the same time, most attention weights become very small, as most attention heads tend to connect a given token with only a small fraction of other tokens in the sequence. These weights become promising targets for cutting backpropagation. We propose a simple probabilistic rule controlled by a single parameter $c$ that cuts backpropagation through most attention weights, leaving at most $c$ interactions per token per attention head. This brings a factor of $c/n$ reduction in the compute required for the attention backpropagation, turning it from quadratic $O(n^2)$ to linear complexity $O(nc)$. We have empirically verified that, for a typical transformer model, cutting $99\%$ of the attention gradient flow (i.e. choosing $c \sim 20-30$) results in relative gradient variance increase of only about $1\%$ for $n \sim 2000$, and it decreases with $n$. This approach is amenable to efficient sparse matrix implementation, thus being promising for making the cost of a backward pass negligible relative to the cost of a forward pass when training a transformer model on long sequences.
Authors: Jeremy Qin
Abstract: Time series forecasting plays a crucial role in various applications, particularly in healthcare, where accurate predictions of future health trajectories can significantly impact clinical decision-making. Ensuring transparency and explainability of the models responsible for these tasks is essential for their adoption in critical settings. Recent work has explored a top-down approach to bi-level transparency, focusing on understanding trends and properties of predicted time series using static features. In this work, we extend this framework by incorporating exogenous time series features alongside static features in a structured manner, while maintaining cohesive interpretation. Our approach leverages the insights of trajectory comprehension to introduce an encoding mechanism for exogenous time series, where they are decomposed into meaningful trends and properties, enabling the extraction of interpretable patterns. Through experiments on several synthetic datasets, we demonstrate that our approach remains predictive while preserving interpretability and robustness. This work represents a step towards developing robust, and generalized time series forecasting models. The code is available at https://github.com/jeremy-qin/TIMEVIEW
Authors: Eray Can Elumar, Cem Tekin, Osman Yagan
Abstract: Recent advances in large language models (LLMs) have enabled automated dataset labeling with minimal human supervision. While majority voting across multiple LLMs can improve label reliability by mitigating individual model biases, it incurs high computational costs due to repeated querying. In this work, we propose a novel online framework, Cost-aware Majority Voting (CaMVo), for efficient and accurate LLM-based dataset annotation. CaMVo adaptively selects a subset of LLMs for each data instance based on contextual embeddings, balancing confidence and cost without requiring pre-training or ground-truth labels. Leveraging a LinUCB-based selection mechanism and a Bayesian estimator over confidence scores, CaMVo estimates a lower bound on labeling accuracy for each LLM and aggregates responses through weighted majority voting. Our empirical evaluation on the MMLU and IMDB Movie Review datasets demonstrates that CaMVo achieves comparable or superior accuracy to full majority voting while significantly reducing labeling costs. This establishes CaMVo as a practical and robust solution for cost-efficient annotation in dynamic labeling environments.
Authors: Zihu Wang, Boxun Xu, Hejia Geng, Peng Li
Abstract: Graph contrastive learning (GCL) has demonstrated great promise for learning generalizable graph representations from unlabeled data. However, conventional GCL approaches face two critical limitations: (1) the restricted expressive capacity of multilayer perceptron (MLP) based encoders, and (2) suboptimal negative samples that either from random augmentations-failing to provide effective 'hard negatives'-or generated hard negatives without addressing the semantic distinctions crucial for discriminating graph data. To this end, we propose Khan-GCL, a novel framework that integrates the Kolmogorov-Arnold Network (KAN) into the GCL encoder architecture, substantially enhancing its representational capacity. Furthermore, we exploit the rich information embedded within KAN coefficient parameters to develop two novel critical feature identification techniques that enable the generation of semantically meaningful hard negative samples for each graph representation. These strategically constructed hard negatives guide the encoder to learn more discriminative features by emphasizing critical semantic differences between graphs. Extensive experiments demonstrate that our approach achieves state-of-the-art performance compared to existing GCL methods across a variety of datasets and tasks.
Authors: Zehong Wang, Zheyuan Liu, Tianyi Ma, Jiazheng Li, Zheyuan Zhang, Xingbo Fu, Yiyang Li, Zhengqing Yuan, Wei Song, Yijun Ma, Qingkai Zeng, Xiusi Chen, Jianan Zhao, Jundong Li, Meng Jiang, Pietro Lio, Nitesh Chawla, Chuxu Zhang, Yanfang Ye
Abstract: Graph-structured data pervades domains such as social networks, biological systems, knowledge graphs, and recommender systems. While foundation models have transformed natural language processing, vision, and multimodal learning through large-scale pretraining and generalization, extending these capabilities to graphs -- characterized by non-Euclidean structures and complex relational semantics -- poses unique challenges and opens new opportunities. To this end, Graph Foundation Models (GFMs) aim to bring scalable, general-purpose intelligence to structured data, enabling broad transfer across graph-centric tasks and domains. This survey provides a comprehensive overview of GFMs, unifying diverse efforts under a modular framework comprising three key components: backbone architectures, pretraining strategies, and adaptation mechanisms. We categorize GFMs by their generalization scope -- universal, task-specific, and domain-specific -- and review representative methods, key innovations, and theoretical insights within each category. Beyond methodology, we examine theoretical foundations including transferability and emergent capabilities, and highlight key challenges such as structural alignment, heterogeneity, scalability, and evaluation. Positioned at the intersection of graph learning and general-purpose AI, GFMs are poised to become foundational infrastructure for open-ended reasoning over structured data. This survey consolidates current progress and outlines future directions to guide research in this rapidly evolving field. Resources are available at https://github.com/Zehong-Wang/Awesome-Foundation-Models-on-Graphs.
URLs: https://github.com/Zehong-Wang/Awesome-Foundation-Models-on-Graphs.
Authors: Sajjad Ghiasvand, Haniyeh Ehsani Oskouie, Mahnoosh Alizadeh, Ramtin Pedarsani
Abstract: Vision-Language Models (VLMs) such as CLIP have shown remarkable performance in cross-modal tasks through large-scale contrastive pre-training. To adapt these large transformer-based models efficiently for downstream tasks, Parameter-Efficient Fine-Tuning (PEFT) techniques like LoRA have emerged as scalable alternatives to full fine-tuning, especially in few-shot scenarios. However, like traditional deep neural networks, VLMs are highly vulnerable to adversarial attacks, where imperceptible perturbations can significantly degrade model performance. Adversarial training remains the most effective strategy for improving model robustness in PEFT. In this work, we propose AdvCLIP-LoRA, the first algorithm designed to enhance the adversarial robustness of CLIP models fine-tuned with LoRA in few-shot settings. Our method formulates adversarial fine-tuning as a minimax optimization problem and provides theoretical guarantees for convergence under smoothness and nonconvex-strong-concavity assumptions. Empirical results across eight datasets using ViT-B/16 and ViT-B/32 models show that AdvCLIP-LoRA significantly improves robustness against common adversarial attacks (e.g., FGSM, PGD), without sacrificing much clean accuracy. These findings highlight AdvCLIP-LoRA as a practical and theoretically grounded approach for robust adaptation of VLMs in resource-constrained settings.
Authors: Shivam Agarwal, Zimin Zhang, Lifan Yuan, Jiawei Han, Hao Peng
Abstract: Entropy minimization (EM) trains the model to concentrate even more probability mass on its most confident outputs. We show that this simple objective alone, without any labeled data, can substantially improve large language models' (LLMs) performance on challenging math, physics, and coding tasks. We explore three approaches: (1) EM-FT minimizes token-level entropy similarly to instruction finetuning, but on unlabeled outputs drawn from the model; (2) EM-RL: reinforcement learning with negative entropy as the only reward to maximize; (3) EM-INF: inference-time logit adjustment to reduce entropy without any training data or parameter updates. On Qwen-7B, EM-RL, without any labeled data, achieves comparable or better performance than strong RL baselines such as GRPO and RLOO that are trained on 60K labeled examples. Furthermore, EM-INF enables Qwen-32B to match or exceed the performance of proprietary models like GPT-4o, Claude 3 Opus, and Gemini 1.5 Pro on the challenging SciCode benchmark, while being 3x more efficient than self-consistency and sequential refinement. Our findings reveal that many pretrained LLMs possess previously underappreciated reasoning capabilities that can be effectively elicited through entropy minimization alone, without any labeled data or even any parameter updates.
Authors: Yang Xu, Swetha Ganesh, Washim Uddin Mondal, Qinbo Bai, Vaneet Aggarwal
Abstract: This paper investigates infinite-horizon average reward Constrained Markov Decision Processes (CMDPs) with general parametrization. We propose a Primal-Dual Natural Actor-Critic algorithm that adeptly manages constraints while ensuring a high convergence rate. In particular, our algorithm achieves global convergence and constraint violation rates of $\tilde{\mathcal{O}}(1/\sqrt{T})$ over a horizon of length $T$ when the mixing time, $\tau_{\mathrm{mix}}$, is known to the learner. In absence of knowledge of $\tau_{\mathrm{mix}}$, the achievable rates change to $\tilde{\mathcal{O}}(1/T^{0.5-\epsilon})$ provided that $T \geq \tilde{\mathcal{O}}\left(\tau_{\mathrm{mix}}^{2/\epsilon}\right)$. Our results match the theoretical lower bound for Markov Decision Processes and establish a new benchmark in the theoretical exploration of average reward CMDPs.
Authors: Tong Cheng, Fu Jie, Xinpeng Ling, Huifa Li, Zhili Chen
Abstract: Graph Neural Networks (GNNs) have been widely used for graph analysis. Federated Graph Learning (FGL) is an emerging learning framework to collaboratively train graph data from various clients. However, since clients are required to upload model parameters to the server in each round, this provides the server with an opportunity to infer each client's data privacy. In this paper, we focus on label distribution attacks(LDAs) that aim to infer the label distributions of the clients' local data. We take the first step to attack client's label distributions in FGL. Firstly, we observe that the effectiveness of LDA is closely related to the variance of node embeddings in GNNs. Next, we analyze the relation between them and we propose a new attack named EC-LDA, which significantly improves the attack effectiveness by compressing node embeddings. Thirdly, extensive experiments on node classification and link prediction tasks across six widely used graph datasets show that EC-LDA outperforms the SOTA LDAs. For example, EC-LDA attains optimal values under both Cos-sim and JS-div evaluation metrics in the CoraFull and LastFM datasets. Finally, we explore the robustness of EC-LDA under differential privacy protection.
Authors: Yunlong Hou, Fengzhuo Zhang, Cunxiao Du, Xuan Zhang, Jiachun Pan, Tianyu Pang, Chao Du, Vincent Y. F. Tan, Zhuoran Yang
Abstract: Speculative decoding has emerged as a popular method to accelerate the inference of Large Language Models (LLMs) while retaining their superior text generation performance. Previous methods either adopt a fixed speculative decoding configuration regardless of the prefix tokens, or train draft models in an offline or online manner to align them with the context. This paper proposes a training-free online learning framework to adaptively choose the configuration of the hyperparameters for speculative decoding as text is being generated. We first formulate this hyperparameter selection problem as a Multi-Armed Bandit problem and provide a general speculative decoding framework BanditSpec. Furthermore, two bandit-based hyperparameter selection algorithms, UCBSpec and EXP3Spec, are designed and analyzed in terms of a novel quantity, the stopping time regret. We upper bound this regret under both stochastic and adversarial reward settings. By deriving an information-theoretic impossibility result, it is shown that the regret performance of UCBSpec is optimal up to universal constants. Finally, extensive empirical experiments with LLaMA3 and Qwen2 demonstrate that our algorithms are effective compared to existing methods, and the throughput is close to the oracle best hyperparameter in simulated real-life LLM serving scenarios with diverse input prompts.
Authors: Weiqin Chen, Xinjie Zhang, Dharmashankar Subramanian, Santiago Paternain
Abstract: Transformer models (TMs) have exhibited remarkable in-context reinforcement learning (ICRL) capabilities, allowing them to generalize to and improve in previously unseen environments without re-training or fine-tuning. This is typically accomplished by imitating the complete learning histories of a source RL algorithm over a substantial amount of pretraining environments, which, however, may transfer suboptimal behaviors inherited from the source algorithm/dataset. Therefore, in this work, we address the issue of inheriting suboptimality from the perspective of dataset preprocessing. Motivated by the success of the weighted empirical risk minimization, we propose a simple yet effective approach, learning history filtering (LHF), to enhance ICRL by reweighting and filtering the learning histories based on their improvement and stability characteristics. To the best of our knowledge, LHF is the first approach to avoid source suboptimality by dataset preprocessing, and can be combined with the current state-of-the-art (SOTA) ICRL algorithms. We substantiate the effectiveness of LHF through a series of experiments conducted on the well-known ICRL benchmarks, encompassing both discrete environments and continuous robotic manipulation tasks, with three SOTA ICRL algorithms (AD, DPT, DICP) as the backbones. LHF exhibits robust performance across a variety of suboptimal scenarios, as well as under varying hyperparameters and sampling strategies. Notably, the superior performance of LHF becomes more pronounced in the presence of noisy data, indicating the significance of filtering learning histories.
Authors: Xiaohou Shi, Ke Li, Aobo Liang, Yan Sun
Abstract: In the past few years, time series foundation models have achieved superior predicting accuracy. However, real-world time series often exhibit significant diversity in their temporal patterns across different time spans and domains, making it challenging for a single model architecture to fit all complex scenarios. In addition, time series data may have multiple variables exhibiting complex correlations between each other. Recent mainstream works have focused on modeling times series in a channel-independent manner in both pretraining and finetuning stages, overlooking the valuable inter-series dependencies. To this end, we propose \textbf{Time Tracker} for better predictions on multivariate time series data. Firstly, we leverage sparse mixture of experts (MoE) within Transformers to handle the modeling of diverse time series patterns, thereby alleviating the learning difficulties of a single model while improving its generalization. Besides, we propose Any-variate Attention, enabling a unified model structure to seamlessly handle both univariate and multivariate time series, thereby supporting channel-independent modeling during pretraining and channel-mixed modeling for finetuning. Furthermore, we design a graph learning module that constructs relations among sequences from frequency-domain features, providing more precise guidance to capture inter-series dependencies in channel-mixed modeling. Based on these advancements, Time Tracker achieves state-of-the-art performance in predicting accuracy, model generalization and adaptability.
Authors: Nanxu Gong, Zijun Li, Sixun Dong, Haoyue Bai, Wangyang Ying, Xinyuan Wang, Yanjie Fu
Abstract: Feature Transformation (FT) crafts new features from original ones via mathematical operations to enhance dataset expressiveness for downstream models. However, existing FT methods exhibit critical limitations: discrete search struggles with enormous combinatorial spaces, impeding practical use; and continuous search, being highly sensitive to initialization and step sizes, often becomes trapped in local optima, restricting global exploration. To overcome these limitations, DIFFT redefines FT as a reward-guided generative task. It first learns a compact and expressive latent space for feature sets using a Variational Auto-Encoder (VAE). A Latent Diffusion Model (LDM) then navigates this space to generate high-quality feature embeddings, its trajectory guided by a performance evaluator towards task-specific optima. This synthesis of global distribution learning (from LDM) and targeted optimization (reward guidance) produces potent embeddings, which a novel semi-autoregressive decoder efficiently converts into structured, discrete features, preserving intra-feature dependencies while allowing parallel inter-feature generation. Extensive experiments on 14 benchmark datasets show DIFFT consistently outperforms state-of-the-art baselines in predictive accuracy and robustness, with significantly lower training and inference times.
Authors: Bo-Han Lai, Pin-Han Huang, Bo-Han Kung, Shang-Tse Chen
Abstract: Lipschitz neural networks are well-known for providing certified robustness in deep learning. In this paper, we present a novel, efficient Block Reflector Orthogonal (BRO) layer that enhances the capability of orthogonal layers on constructing more expressive Lipschitz neural architectures. In addition, by theoretically analyzing the nature of Lipschitz neural networks, we introduce a new loss function that employs an annealing mechanism to increase margin for most data points. This enables Lipschitz models to provide better certified robustness. By employing our BRO layer and loss function, we design BRONet - a simple yet effective Lipschitz neural network that achieves state-of-the-art certified robustness. Extensive experiments and empirical analysis on CIFAR-10/100, Tiny-ImageNet, and ImageNet validate that our method outperforms existing baselines. The implementation is available at \href{https://github.com/ntuaislab/BRONet}{https://github.com/ntuaislab/BRONet}.
URLs: https://github.com/ntuaislab/BRONet, https://github.com/ntuaislab/BRONet
Authors: Jiawei Gu, Ziyue Qiao, Zechao Li
Abstract: The task of graph-level out-of-distribution (OOD) detection is crucial for deploying graph neural networks in real-world settings. In this paper, we observe a significant difference in the relationship between the largest and second-largest eigenvalues of the Laplacian matrix for in-distribution (ID) and OOD graph samples: \textit{OOD samples often exhibit anomalous spectral gaps (the difference between the largest and second-largest eigenvalues)}. This observation motivates us to propose SpecGap, an effective post-hoc approach for OOD detection on graphs. SpecGap adjusts features by subtracting the component associated with the second-largest eigenvalue, scaled by the spectral gap, from the high-level features (i.e., $\mathbf{X}-\left(\lambda_n-\lambda_{n-1}\right) \mathbf{u}_{n-1} \mathbf{v}_{n-1}^T$). SpecGap achieves state-of-the-art performance across multiple benchmark datasets. We present extensive ablation studies and comprehensive theoretical analyses to support our empirical results. As a parameter-free post-hoc method, SpecGap can be easily integrated into existing graph neural network models without requiring any additional training or model modification.
Authors: Zhehao Huang, Xinwen Cheng, Jie Zhang, Jinghao Zheng, Haoran Wang, Zhengbao He, Tao Li, Xiaolin Huang
Abstract: Recent advancements in deep models have highlighted the need for intelligent systems that combine continual learning (CL) for knowledge acquisition with machine unlearning (MU) for data removal, forming the Continual Learning-Unlearning (CLU) paradigm. While existing work treats CL and MU as separate processes, we reveal their intrinsic connection through a unified optimization framework based on Kullback-Leibler divergence minimization. This framework decomposes gradient updates for approximate CLU into four components: learning new knowledge, unlearning targeted data, preserving existing knowledge, and modulation via weight saliency. A critical challenge lies in balancing knowledge update and retention during sequential learning-unlearning cycles. To resolve this stability-plasticity dilemma, we introduce a remain-preserved manifold constraint to induce a remaining Hessian compensation for CLU iterations. A fast-slow weight adaptation mechanism is designed to efficiently approximate the second-order optimization direction, combined with adaptive weighting coefficients and a balanced weight saliency mask, proposing a unified implementation framework for gradient-based CLU. Furthermore, we pioneer task-agnostic CLU scenarios that support fine-grained unlearning at the cross-task category and random sample levels beyond the traditional task-aware setups. Experiments demonstrate that the proposed UG-CLU framework effectively coordinates incremental learning, precise unlearning, and knowledge stability across multiple datasets and model architectures, providing a theoretical foundation and methodological support for dynamic, compliant intelligent systems.
Authors: Jiawei Gu, Ziyue Qiao, Xiao Luo
Abstract: Graph Neural Networks (GNNs) have shown remarkable performance across various domains, yet they often struggle with model bias, particularly in the presence of class imbalance. This bias can lead to suboptimal performance and unfair predictions, especially for underrepresented classes. We introduce NeuBM (Neutral Bias Mitigation), a novel approach to mitigate model bias in GNNs through neutral input calibration. NeuBM leverages a dynamically updated neutral graph to estimate and correct the inherent biases of the model. By subtracting the logits obtained from the neutral graph from those of the input graph, NeuBM effectively recalibrates the model's predictions, reducing bias across different classes. Our method integrates seamlessly into existing GNN architectures and training procedures, requiring minimal computational overhead. Extensive experiments on multiple benchmark datasets demonstrate that NeuBM significantly improves the balanced accuracy and recall of minority classes, while maintaining strong overall performance. The effectiveness of NeuBM is particularly pronounced in scenarios with severe class imbalance and limited labeled data, where traditional methods often struggle. We provide theoretical insights into how NeuBM achieves bias mitigation, relating it to the concept of representation balancing. Our analysis reveals that NeuBM not only adjusts the final predictions but also influences the learning of balanced feature representations throughout the network.
Authors: Adel Javanmard, Rudrajit Das, Alessandro Epasto, Vahab Mirrokni
Abstract: Retraining a model using its own predictions together with the original, potentially noisy labels is a well-known strategy for improving the model performance. While prior works have demonstrated the benefits of specific heuristic retraining schemes, the question of how to optimally combine the model's predictions and the provided labels remains largely open. This paper addresses this fundamental question for binary classification tasks. We develop a principled framework based on approximate message passing (AMP) to analyze iterative retraining procedures for two ground truth settings: Gaussian mixture model (GMM) and generalized linear model (GLM). Our main contribution is the derivation of the Bayes optimal aggregator function to combine the current model's predictions and the given labels, which when used to retrain the same model, minimizes its prediction error. We also quantify the performance of this optimal retraining strategy over multiple rounds. We complement our theoretical results by proposing a practically usable version of the theoretically-optimal aggregator function for linear probing with the cross-entropy loss, and demonstrate its superiority over baseline methods in the high label noise regime.
Authors: Christian Walder, Deep Karkhanis
Abstract: Reinforcement Learning (RL) algorithms sample multiple n>1 solution attempts for each problem and reward them independently. This optimizes for pass@1 performance and prioritizes the strength of isolated samples at the expense of the diversity and collective utility of sets of samples. This under-utilizes the sampling capacity, limiting exploration and eventual improvement on harder examples. As a fix, we propose Pass-at-k Policy Optimization (PKPO), a transformation on the final rewards which leads to direct optimization of pass@k performance, thus optimizing for sets of samples that maximize reward when considered jointly. Our contribution is to derive novel low variance unbiased estimators for pass@k and its gradient, in both the binary and continuous reward settings. We show optimization with our estimators reduces to standard RL with rewards that have been jointly transformed by a stable and efficient transformation function. While previous efforts are restricted to k=n, ours is the first to enable robust optimization of pass@k for any arbitrary k <= n. Moreover, instead of trading off pass@1 performance for pass@k gains, our method allows annealing k during training, optimizing both metrics and often achieving strong pass@1 numbers alongside significant pass@k gains. We validate our reward transformations on toy experiments, which reveal the variance reducing properties of our formulations. We also include real-world examples using the open-source LLM, GEMMA-2. We find that our transformation effectively optimizes for the target k. Furthermore, higher k values enable solving more and harder problems, while annealing k boosts both the pass@1 and pass@k . Crucially, for challenging task sets where conventional pass@1 optimization stalls, our pass@k approach unblocks learning, likely due to better exploration by prioritizing joint utility over the utility of individual samples.
Authors: Haomin Bai, Dingzhi Yu, Shuai Li, Haipeng Luo, Lijun Zhang
Abstract: Group distributionally robust optimization (GDRO) aims to develop models that perform well across $m$ distributions simultaneously. Existing GDRO algorithms can only process a fixed number of samples per iteration, either 1 or $m$, and therefore can not support scenarios where the sample size varies dynamically. To address this limitation, we investigate GDRO with flexible sample queries and cast it as a two-player game: one player solves an online convex optimization problem, while the other tackles a prediction with limited advice (PLA) problem. Within such a game, we propose a novel PLA algorithm, constructing appropriate loss estimators for cases where the sample size is either 1 or not, and updating the decision using follow-the-regularized-leader. Then, we establish the first high-probability regret bound for non-oblivious PLA. Building upon the above approach, we develop a GDRO algorithm that allows an arbitrary and varying sample size per round, achieving a high-probability optimization error bound of $O\left(\frac{1}{t}\sqrt{\sum_{j=1}^t \frac{m}{r_j}\log m}\right)$, where $r_t$ denotes the sample size at round $t$. This result demonstrates that the optimization error decreases as the number of samples increases and implies a consistent sample complexity of $O(m\log (m)/\epsilon^2)$ for any fixed sample size $r\in[m]$, aligning with existing bounds for cases of $r=1$ or $m$. We validate our approach on synthetic binary and real-world multi-class datasets.
Authors: Sampanna Yashwant Kahu
Abstract: Efficient task scheduling is paramount in the Linux kernel, where the Completely Fair Scheduler (CFS) meticulously manages CPU resources to balance high utilization with interactive responsiveness. This research pioneers the use of deep learning techniques to predict the sequence of tasks selected by CFS, aiming to evaluate the feasibility of a more generalized and potentially more adaptive task scheduler for diverse workloads. Our core contributions are twofold: first, the systematic generation and curation of a novel scheduling dataset from a running Linux kernel, capturing real-world CFS behavior; and second, the development, training, and evaluation of a Long Short-Term Memory (LSTM) network designed to accurately forecast the next task to be scheduled. This paper further discusses the practical pathways and implications of integrating such a predictive model into the kernel's scheduling framework. The findings and methodologies presented herein open avenues for data-driven advancements in kernel scheduling, with the full source code provided for reproducibility and further exploration.
Authors: Mathew Vanherreweghe, Lirand\"e Pira, Patrick Rebentrost
Abstract: We introduce cumulative polynomial Kolmogorov-Arnold networks (CP-KAN), a neural architecture combining Chebyshev polynomial basis functions and quadratic unconstrained binary optimization (QUBO). Our primary contribution involves reformulating the degree selection problem as a QUBO task, reducing the complexity from $O(D^N)$ to a single optimization step per layer. This approach enables efficient degree selection across neurons while maintaining computational tractability. The architecture performs well in regression tasks with limited data, showing good robustness to input scales and natural regularization properties from its polynomial basis. Additionally, theoretical analysis establishes connections between CP-KAN's performance and properties of financial time series. Our empirical validation across multiple domains demonstrates competitive performance compared to several traditional architectures tested, especially in scenarios where data efficiency and numerical stability are important. Our implementation, including strategies for managing computational overhead in larger networks is available in Ref.~\citep{cpkan_implementation}.
Authors: Kabir V. Dabholkar, Omri Barak
Abstract: Many natural systems, including neural circuits involved in decision making, can be modeled as high-dimensional dynamical systems with multiple stable states. While existing analytical tools primarily describe behavior near stable equilibria, characterizing separatrices -- the manifolds that delineate boundaries between different basins of attraction -- remains challenging, particularly in high-dimensional settings. Here, we introduce a numerical framework leveraging Koopman Theory combined with Deep Neural Networks to effectively characterize separatrices. Specifically, we approximate Koopman Eigenfunctions (KEFs) associated with real positive eigenvalues, which vanish precisely at the separatrices. Utilizing these scalar KEFs, optimization methods efficiently locate separatrices even in complex systems. We demonstrate our approach on synthetic benchmarks, ecological network models, and recurrent neural networks trained on neuroscience-inspired tasks. Moreover, we illustrate the practical utility of our method by designing optimal perturbations that can shift systems across separatrices, enabling predictions relevant to optogenetic stimulation experiments in neuroscience.
Authors: Peter S\'uken\'ik, Christoph H. Lampert, Marco Mondelli
Abstract: The empirical emergence of neural collapse -- a surprising symmetry in the feature representations of the training data in the penultimate layer of deep neural networks -- has spurred a line of theoretical research aimed at its understanding. However, existing work focuses on data-agnostic models or, when data structure is taken into account, it remains limited to multi-layer perceptrons. Our paper fills both these gaps by analyzing modern architectures in a data-aware regime: we prove that global optima of deep regularized transformers and residual networks (ResNets) with LayerNorm trained with cross entropy or mean squared error loss are approximately collapsed, and the approximation gets tighter as the depth grows. More generally, we formally reduce any end-to-end large-depth ResNet or transformer training into an equivalent unconstrained features model, thus justifying its wide use in the literature even beyond data-agnostic settings. Our theoretical results are supported by experiments on computer vision and language datasets showing that, as the depth grows, neural collapse indeed becomes more prominent.
Authors: Mohamad Mestoukirdi, Mourad Khanfouci
Abstract: This work proposes a new algorithm to mitigate model generalization loss in Vertical Federated Learning (VFL) operating under client reliability constraints within 5G Core Networks (CNs). Recently studied and endorsed by 3GPP, VFL enables collaborative and load-balanced model training and inference across the CN. However, the performance of VFL significantly degrades when the Network Data Analytics Functions (NWDAFs) - which serve as primary clients for VFL model training and inference - experience reliability issues stemming from resource constraints and operational overhead. Unlike edge environments, CN environments adopt fundamentally different data management strategies, characterized by more centralized data orchestration capabilities. This presents opportunities to implement better distributed solutions that take full advantage of the CN data handling flexibility. Leveraging this flexibility, we propose a method that optimizes the vertical feature split among clients while centrally defining their local models based on reliability metrics. Our empirical evaluation demonstrates the effectiveness of our proposed algorithm, showing improved performance over traditional baseline methods.
Authors: Xiaoling Zhou, Wei Ye, Rui Xie, Shikun Zhang
Abstract: Deep learning has seen widespread success in various domains such as science, industry, and society. However, it is acknowledged that certain approaches suffer from non-robustness, relying on spurious correlations for predictions. Addressing these limitations is of paramount importance, necessitating the development of methods that can disentangle spurious correlations. {This study attempts to implement causal models via logit perturbations and introduces a novel Causal Logit Perturbation (CLP) framework to train classifiers with generated causal logit perturbations for individual samples, thereby mitigating the spurious associations between non-causal attributes (i.e., image backgrounds) and classes.} {Our framework employs a} perturbation network to generate sample-wise logit perturbations using a series of training characteristics of samples as inputs. The whole framework is optimized by an online meta-learning-based learning algorithm and leverages human causal knowledge by augmenting metadata in both counterfactual and factual manners. Empirical evaluations on four typical biased learning scenarios, including long-tail learning, noisy label learning, generalized long-tail learning, and subpopulation shift learning, demonstrate that CLP consistently achieves state-of-the-art performance. Moreover, visualization results support the effectiveness of the generated causal perturbations in redirecting model attention towards causal image attributes and dismantling spurious associations.
Authors: Suping Xu, Lin Shang, Keyu Liu, Hengrong Ju, Xibei Yang, Witold Pedrycz
Abstract: Fuzzy rough feature selection (FRFS) is an effective means of addressing the curse of dimensionality in high-dimensional data. By removing redundant and irrelevant features, FRFS helps mitigate classifier overfitting, enhance generalization performance, and lessen computational overhead. However, most existing FRFS algorithms primarily focus on reducing uncertainty in pattern classification, neglecting that lower uncertainty does not necessarily result in improved classification performance, despite it commonly being regarded as a key indicator of feature selection effectiveness in the FRFS literature. To bridge uncertainty characterization and pattern classification, we propose a Margin-aware Fuzzy Rough Feature Selection (MAFRFS) framework that considers both the compactness and separation of label classes. MAFRFS effectively reduces uncertainty in pattern classification tasks, while guiding the feature selection towards more separable and discriminative label class structures. Extensive experiments on 15 public datasets demonstrate that MAFRFS is highly scalable and more effective than FRFS. The algorithms developed using MAFRFS outperform six state-of-the-art feature selection algorithms.
Authors: Idriss Malek, Abhijit Sharma, Salem Lahlou
Abstract: Although Generative Flow Networks (GFlowNets) are designed to capture multiple modes of a reward function, they often suffer from mode collapse in practice, getting trapped in early discovered modes and requiring prolonged training to find diverse solutions. Existing exploration techniques may rely on heuristic novelty signals. We propose Loss-Guided GFlowNets (LGGFN), a novel approach where an auxiliary GFlowNet's exploration is directly driven by the main GFlowNet's training loss. By prioritizing trajectories where the main model exhibits high loss, LGGFN focuses sampling on poorly understood regions of the state space. This targeted exploration significantly accelerates the discovery of diverse, high-reward samples. Empirically, across various benchmarks including grid environments, structured sequence generation, and Bayesian structure learning, LGGFN consistently enhances exploration efficiency and sample diversity compared to baselines. For instance, on a challenging sequence generation task, it discovered over 40 times more unique valid modes while simultaneously reducing the exploration error metric by approximately 99\%.
Authors: Hyunseok Lee, Jeonghoon Kim, Beomjun Kim, Jihoon Tack, Chansong Jo, Jaehong Lee, Cheonbok Park, Sookyo In, Jinwoo Shin, Kang Min Yoo
Abstract: Recent advances in Multimodal Large Language Models (MLLMs) have enabled autonomous agents to interact with computers via Graphical User Interfaces (GUIs), where accurately localizing the coordinates of interface elements (e.g., buttons) is often required for fine-grained actions. However, this remains significantly challenging, leading prior works to rely on large-scale web datasets to improve the grounding accuracy. In this work, we propose Reasoning Graphical User Interface Grounding for Data Efficiency (ReGUIDE), a novel and effective framework for web grounding that enables MLLMs to learn data efficiently through self-generated reasoning and spatial-aware criticism. More specifically, ReGUIDE learns to (i) self-generate a language reasoning process for the localization via online reinforcement learning, and (ii) criticize the prediction using spatial priors that enforce equivariance under input transformations. At inference time, ReGUIDE further boosts performance through a test-time scaling strategy, which combines spatial search with coordinate aggregation. Our experiments demonstrate that ReGUIDE significantly advances web grounding performance across multiple benchmarks, outperforming baselines with substantially fewer training data points (e.g., only 0.2% samples compared to the best open-sourced baselines).
Authors: Chenyu Zheng, Xinyu Zhang, Rongzhen Wang, Wei Huang, Zhi Tian, Weilin Huang, Jun Zhu, Chongxuan Li
Abstract: Diffusion Transformers have emerged as the foundation for vision generative models, but their scalability is limited by the high cost of hyperparameter (HP) tuning at large scales. Recently, Maximal Update Parametrization ($\mu$P) was proposed for vanilla Transformers, which enables stable HP transfer from small to large language models, and dramatically reduces tuning costs. However, it remains unclear whether $\mu$P of vanilla Transformers extends to diffusion Transformers, which differ architecturally and objectively. In this work, we generalize standard $\mu$P to diffusion Transformers and validate its effectiveness through large-scale experiments. First, we rigorously prove that $\mu$P of mainstream diffusion Transformers, including DiT, U-ViT, PixArt-$\alpha$, and MMDiT, aligns with that of the vanilla Transformer, enabling the direct application of existing $\mu$P methodologies. Leveraging this result, we systematically demonstrate that DiT-$\mu$P enjoys robust HP transferability. Notably, DiT-XL-2-$\mu$P with transferred learning rate achieves 2.9 times faster convergence than the original DiT-XL-2. Finally, we validate the effectiveness of $\mu$P on text-to-image generation by scaling PixArt-$\alpha$ from 0.04B to 0.61B and MMDiT from 0.18B to 18B. In both cases, models under $\mu$P outperform their respective baselines while requiring small tuning cost, only 5.5% of one training run for PixArt-$\alpha$ and 3% of consumption by human experts for MMDiT-18B. These results establish $\mu$P as a principled and efficient framework for scaling diffusion Transformers.
Authors: Kun Fang, Qinghua Tao, Mingzhen He, Kexin Lv, Runze Yang, Haibo Hu, Xiaolin Huang, Jie Yang, Longbin Cao
Abstract: Out-of-Distribution (OoD) detection is vital for the reliability of deep neural networks, the key of which lies in effectively characterizing the disparities between OoD and In-Distribution (InD) data. In this work, such disparities are exploited through a fresh perspective of non-linear feature subspace. That is, a discriminative non-linear subspace is learned from InD features to capture representative patterns of InD, while informative patterns of OoD features cannot be well captured in such a subspace due to their different distribution. Grounded on this perspective, we exploit the deviations of InD and OoD features in such a non-linear subspace for effective OoD detection. To be specific, we leverage the framework of Kernel Principal Component Analysis (KPCA) to attain the discriminative non-linear subspace and deploy the reconstruction error on such subspace to distinguish InD and OoD data. Two challenges emerge: (i) the learning of an effective non-linear subspace, i.e., the selection of kernel function in KPCA, and (ii) the computation of the kernel matrix with large-scale InD data. For the former, we reveal two vital non-linear patterns that closely relate to the InD-OoD disparity, leading to the establishment of a Cosine-Gaussian kernel for constructing the subspace. For the latter, we introduce two techniques to approximate the Cosine-Gaussian kernel with significantly cheap computations. In particular, our approximation is further tailored by incorporating the InD data confidence, which is demonstrated to promote the learning of discriminative subspaces for OoD data. Our study presents new insights into the non-linear feature subspace for OoD detection and contributes practical explorations on the associated kernel design and efficient computations, yielding a KPCA detection method with distinctively improved efficacy and efficiency.
Authors: Qianyue Hao, Yiwen Song, Qingmin Liao, Jian Yuan, Yong Li
Abstract: Policy exploration is critical in reinforcement learning (RL), where existing approaches include greedy, Gaussian process, etc. However, these approaches utilize preset stochastic processes and are indiscriminately applied in all kinds of RL tasks without considering task-specific features that influence policy exploration. Moreover, during RL training, the evolution of such stochastic processes is rigid, which typically only incorporates a decay in the variance, failing to adjust flexibly according to the agent's real-time learning status. Inspired by the analyzing and reasoning capability of large language models (LLMs), we design LLM-Explorer to adaptively generate task-specific exploration strategies with LLMs, enhancing the policy exploration in RL. In our design, we sample the learning trajectory of the agent during the RL training in a given task and prompt the LLM to analyze the agent's current policy learning status and then generate a probability distribution for future policy exploration. Updating the probability distribution periodically, we derive a stochastic process specialized for the particular task and dynamically adjusted to adapt to the learning process. Our design is a plug-in module compatible with various widely applied RL algorithms, including the DQN series, DDPG, TD3, and any possible variants developed based on them. Through extensive experiments on the Atari and MuJoCo benchmarks, we demonstrate LLM-Explorer's capability to enhance RL policy exploration, achieving an average performance improvement up to 37.27%. Our code is open-source at https://anonymous.4open.science/r/LLM-Explorer-19BE for reproducibility.
Authors: Johannes Kaiser, Kristian Schwethelm, Daniel Rueckert, Georgios Kaissis
Abstract: Accurately estimating the informativeness of individual samples in a dataset is an important objective in deep learning, as it can guide sample selection, which can improve model efficiency and accuracy by removing redundant or potentially harmful samples. We propose Laplace Sample Information (LSI) measure of sample informativeness grounded in information theory widely applicable across model architectures and learning settings. LSI leverages a Bayesian approximation to the weight posterior and the KL divergence to measure the change in the parameter distribution induced by a sample of interest from the dataset. We experimentally show that LSI is effective in ordering the data with respect to typicality, detecting mislabeled samples, measuring class-wise informativeness, and assessing dataset difficulty. We demonstrate these capabilities of LSI on image and text data in supervised and unsupervised settings. Moreover, we show that LSI can be computed efficiently through probes and transfers well to the training of large models.
Authors: Yiwen Song, Qianyue Hao, Qingmin Liao, Jian Yuan, Yong Li
Abstract: Model ensemble is a useful approach in reinforcement learning (RL) for training effective agents. Despite wide success of RL, training effective agents remains difficult due to the multitude of factors requiring careful tuning, such as algorithm selection, hyperparameter settings, and even random seed choices, all of which can significantly influence an agent's performance. Model ensemble helps overcome this challenge by combining multiple weak agents into a single, more powerful one, enhancing overall performance. However, existing ensemble methods, such as majority voting and Boltzmann addition, are designed as fixed strategies and lack a semantic understanding of specific tasks, limiting their adaptability and effectiveness. To address this, we propose LLM-Ens, a novel approach that enhances RL model ensemble with task-specific semantic understandings driven by large language models (LLMs). Given a task, we first design an LLM to categorize states in this task into distinct 'situations', incorporating high-level descriptions of the task conditions. Then, we statistically analyze the strengths and weaknesses of each individual agent to be used in the ensemble in each situation. During the inference time, LLM-Ens dynamically identifies the changing task situation and switches to the agent that performs best in the current situation, ensuring dynamic model selection in the evolving task condition. Our approach is designed to be compatible with agents trained with different random seeds, hyperparameter settings, and various RL algorithms. Extensive experiments on the Atari benchmark show that LLM-Ens significantly improves the RL model ensemble, surpassing well-known baselines by up to 20.9%. For reproducibility, our code is open-source at https://anonymous.4open.science/r/LLM4RLensemble-F7EE.
URLs: https://anonymous.4open.science/r/LLM4RLensemble-F7EE.
Authors: Yurun Yuan, Fan Chen, Zeyu Jia, Alexander Rakhlin, Tengyang Xie
Abstract: Policy-based methods currently dominate reinforcement learning (RL) pipelines for large language model (LLM) reasoning, leaving value-based approaches largely unexplored. We revisit the classical paradigm of Bellman Residual Minimization and introduce Trajectory Bellman Residual Minimization (TBRM), an algorithm that naturally adapts this idea to LLMs, yielding a simple yet effective off-policy algorithm that optimizes a single trajectory-level Bellman objective using the model's own logits as $Q$-values. TBRM removes the need for critics, importance-sampling ratios, or clipping, and operates with only one rollout per prompt. We prove convergence to the near-optimal KL-regularized policy from arbitrary off-policy data via an improved change-of-trajectory-measure analysis. Experiments on standard mathematical-reasoning benchmarks show that TBRM consistently outperforms policy-based baselines, like PPO and GRPO, with comparable or lower computational and memory overhead. Our results indicate that value-based RL might be a principled and efficient alternative for enhancing reasoning capabilities in LLMs.
Authors: Yuxuan Shu, Vasileios Lampos
Abstract: Multivariable time series forecasting methods can integrate information from exogenous variables, leading to significant prediction accuracy gains. Transformer architecture has been widely applied in various time series forecasting models due to its ability to capture long-range sequential dependencies. However, a na\"ive application of transformers often struggles to effectively model complex relationships among variables over time. To mitigate against this, we propose a novel architecture, namely the Spectral Operator Neural Network (Sonnet). Sonnet applies learnable wavelet transformations to the input and incorporates spectral analysis using the Koopman operator. Its predictive skill relies on the Multivariable Coherence Attention (MVCA), an operation that leverages spectral coherence to model variable dependencies. Our empirical analysis shows that Sonnet yields the best performance on $34$ out of $47$ forecasting tasks with an average mean absolute error (MAE) reduction of $1.1\%$ against the most competitive baseline (different per task). We further show that MVCA -- when put in place of the na\"ive attention used in various deep learning models -- can remedy its deficiencies, reducing MAE by $10.7\%$ on average in the most challenging forecasting tasks.
Authors: Anqiao Ouyang, Hongyi Ke, Qi Wang
Abstract: Invertible neural architectures have recently attracted attention for their compactness, interpretability, and information-preserving properties. In this work, we propose the Fourier-Invertible Neural Encoder (FINE), which combines invertible monotonic activation functions with reversible filter structures, and could be extended using Invertible ResNets. This architecture is examined in learning low-dimensional representations of one-dimensional nonlinear wave interactions and exact circular translation symmetry. Dimensionality is preserved across layers, except for a Fourier truncation step in the latent space, which enables dimensionality reduction while maintaining shift equivariance and interpretability. Our results demonstrate that FINE significantly outperforms classical linear methods such as Discrete Fourier Transformation (DFT) and Proper Orthogonal Decomposition (POD), and achieves reconstruction accuracy better than conventional deep autoencoders with convolutional layers (CNN) - while using substantially smaller models and offering superior physical interpretability. These findings suggest that invertible single-neuron networks, when combined with spectral truncation, offer a promising framework for learning compact and interpretable representations of physics datasets, and symmetry-aware representation learning in physics-informed machine learning.
Authors: Yuanlin Chu, Bo Wang, Xiang Liu, Hong Chen, Aiwei Liu, Xuming Hu
Abstract: Large language models (LLMs) have achieved impressive results on multi-step mathematical reasoning, yet at the cost of high computational overhead. This challenge is particularly acute for test-time scaling methods such as parallel decoding, which increase answer diversity but scale poorly in efficiency. To address this efficiency-accuracy trade-off, we propose SSR (Speculative Parallel Scaling Reasoning), a training-free framework that leverages a key insight: by introducing speculative decoding at the step level, we can accelerate reasoning without sacrificing correctness. SSR integrates two components: a Selective Parallel Module (SPM) that identifies a small set of promising reasoning strategies via model-internal scoring, and Step-level Speculative Decoding (SSD), which enables efficient draft-target collaboration for fine-grained reasoning acceleration. Experiments on three mathematical benchmarks-AIME 2024, MATH-500, and LiveMathBench - demonstrate that SSR achieves strong gains over baselines. For instance, on LiveMathBench, SSR improves pass@1 accuracy by 13.84% while reducing computation to 80.5% of the baseline FLOPs. On MATH-500, SSR reduces compute to only 30% with no loss in accuracy.
Authors: Jacob E. Kooi, Zhao Yang, Vincent Fran\c{c}ois-Lavet
Abstract: Neural network architectures have a large impact in machine learning. In reinforcement learning, network architectures have remained notably simple, as changes often lead to small gains in performance. This work introduces a novel encoder architecture for pixel-based model-free reinforcement learning. The Hadamax (\textbf{Hada}mard \textbf{max}-pooling) encoder achieves state-of-the-art performance by max-pooling Hadamard products between GELU-activated parallel hidden layers. Based on the recent PQN algorithm, the Hadamax encoder achieves state-of-the-art model-free performance in the Atari-57 benchmark. Specifically, without applying any algorithmic hyperparameter modifications, Hadamax-PQN achieves an 80\% performance gain over vanilla PQN and significantly surpasses Rainbow-DQN. For reproducibility, the full code is available on \href{https://github.com/Jacobkooi/Hadamax}{GitHub}.
Authors: Malik Tiomoko, Hamza Cherkaoui, Giuseppe Paolo, Zhang Yili, Yu Meng, Zhang Keli, Hafiz Tiomoko Ali
Abstract: Time series forecasting models often produce systematic, predictable errors even in critical domains such as energy, finance, and healthcare. We introduce a novel post training adaptive optimization framework that improves forecast accuracy without retraining or architectural changes. Our method automatically applies expressive transformations optimized via reinforcement learning, contextual bandits, or genetic algorithms to correct model outputs in a lightweight and model agnostic way. Theoretically, we prove that affine corrections always reduce the mean squared error; practically, we extend this idea with dynamic action based optimization. The framework also supports an optional human in the loop component: domain experts can guide corrections using natural language, which is parsed into actions by a language model. Across multiple benchmarks (e.g., electricity, weather, traffic), we observe consistent accuracy gains with minimal computational overhead. Our interactive demo shows the framework's real time usability. By combining automated post hoc refinement with interpretable and extensible mechanisms, our approach offers a powerful new direction for practical forecasting systems.
Authors: Mounssif Krouka, Chaouki Ben Issaid, Mehdi Bennis
Abstract: Federated learning (FL) faces critical challenges, particularly in heterogeneous environments where non-independent and identically distributed data across clients can lead to unfair and inefficient model performance. In this work, we introduce \textit{DRDM}, a novel algorithm that addresses these issues by combining a distributionally robust optimization (DRO) framework with dynamic regularization to mitigate client drift. \textit{DRDM} frames the training as a min-max optimization problem aimed at maximizing performance for the worst-case client, thereby promoting robustness and fairness. This robust objective is optimized through an algorithm leveraging dynamic regularization and efficient local updates, which significantly reduces the required number of communication rounds. Moreover, we provide a theoretical convergence analysis for convex smooth objectives under partial participation. Extensive experiments on three benchmark datasets, covering various model architectures and data heterogeneity levels, demonstrate that \textit{DRDM} significantly improves worst-case test accuracy while requiring fewer communication rounds than existing state-of-the-art baselines. Furthermore, we analyze the impact of signal-to-noise ratio (SNR) and bandwidth on the energy consumption of participating clients, demonstrating that the number of local update steps can be adaptively selected to achieve a target worst-case test accuracy with minimal total energy cost across diverse communication environments.
Authors: Duncan Bart, Bruno Endres Forlin, Ana-Lucia Varbanescu, Marco Ottavi, Kuan-Hsun Chen
Abstract: Integer quantization has emerged as a critical technique to facilitate deployment on resource-constrained devices. Although they do reduce the complexity of the learning models, their inference performance is often prone to quantization-induced errors. To this end, we introduce InTreeger: an end-to-end framework that takes a training dataset as input, and outputs an architecture-agnostic integer-only C implementation of tree-based machine learning model, without loss of precision. This framework enables anyone, even those without prior experience in machine learning, to generate a highly optimized integer-only classification model that can run on any hardware simply by providing an input dataset and target variable. We evaluated our generated implementations across three different architectures (ARM, x86, and RISC-V), resulting in significant improvements in inference latency. In addition, we show the energy efficiency compared to typical decision tree implementations that rely on floating-point arithmetic. The results underscore the advantages of integer-only inference, making it particularly suitable for energy- and area-constrained devices such as embedded systems and edge computing platforms, while also enabling the execution of decision trees on existing ultra-low power devices.
Authors: Martin Carrasco, Guillermo Bernardez, Marco Montagna, Nina Miolane, Lev Telyatnikov
Abstract: While Graph Neural Networks (GNNs) have proven highly effective at modeling relational data, pairwise connections cannot fully capture multi-way relationships naturally present in complex real-world systems. In response to this, Topological Deep Learning (TDL) leverages more general combinatorial representations -- such as simplicial or cellular complexes -- to accommodate higher-order interactions. Existing TDL methods often extend GNNs through Higher-Order Message Passing (HOMP), but face critical \emph{scalability challenges} due to \textit{(i)} a combinatorial explosion of message-passing routes, and \textit{(ii)} significant complexity overhead from the propagation mechanism. To overcome these limitations, we propose HOPSE (Higher-Order Positional and Structural Encoder) -- a \emph{message passing-free} framework that uses Hasse graph decompositions to derive efficient and expressive encodings over \emph{arbitrary higher-order domains}. Notably, HOPSE scales linearly with dataset size while preserving expressive power and permutation equivariance. Experiments on molecular, expressivity and topological benchmarks show that HOPSE matches or surpasses state-of-the-art performance while achieving up to 7 $times$ speedups over HOMP-based models, opening a new path for scalable TDL.
Authors: Naiqi Li, Yuqiu Xie, Peiyuan Liu, Tao Dai, Yong Jiang, Shu-Tao Xia
Abstract: Low-rank regularization (LRR) has been widely applied in various machine learning tasks, but the associated optimization is challenging. Directly optimizing the rank function under constraints is NP-hard in general. To overcome this difficulty, various relaxations of the rank function were studied. However, optimization of these relaxed LRRs typically depends on singular value decomposition, which is a time-consuming and nondifferentiable operator that cannot be optimized with gradient-based techniques. To address these challenges, in this paper we propose an efficient differentiable approximation of the generalized LRR. The considered LRR form subsumes many popular choices like the nuclear norm, the Schatten-$p$ norm, and various nonconvex relaxations. Our method enables LRR terms to be appended to loss functions in a plug-and-play fashion, and the GPU-friendly operations enable efficient and convenient implementation. Furthermore, convergence analysis is presented, which rigorously shows that both the bias and the variance of our rank estimator rapidly reduce with increased sample size and iteration steps. In the experimental study, the proposed method is applied to various tasks, which demonstrates its versatility and efficiency. Code is available at https://github.com/naiqili/EDLRR.
Authors: Yueheng Li, Guangming Xie, Zongqing Lu
Abstract: Reinforcement Learning (RL) in partially observable environments poses significant challenges due to the complexity of learning under uncertainty. While additional information, such as that available in simulations, can enhance training, effectively leveraging it remains an open problem. To address this, we introduce Guided Policy Optimization (GPO), a framework that co-trains a guider and a learner. The guider takes advantage of privileged information while ensuring alignment with the learner's policy that is primarily trained via imitation learning. We theoretically demonstrate that this learning scheme achieves optimality comparable to direct RL, thereby overcoming key limitations inherent in existing approaches. Empirical evaluations show strong performance of GPO across various tasks, including continuous control with partial observability and noise, and memory-based challenges, significantly outperforming existing methods.
Authors: Marcell T. Kurbucz, Nikolaos Tzivanakis, Nilufer Sari Aslam, Adam M. Sykulski
Abstract: Capturing nonlinear relationships without sacrificing interpretability remains a persistent challenge in regression modeling. We introduce SplitWise, a novel framework that enhances stepwise regression. It adaptively transforms numeric predictors into threshold-based binary features using shallow decision trees, but only when such transformations improve model fit, as assessed by the Akaike Information Criterion (AIC) or Bayesian Information Criterion (BIC). This approach preserves the transparency of linear models while flexibly capturing nonlinear effects. Implemented as a user-friendly R package, SplitWise is evaluated on both synthetic and real-world datasets. The results show that it consistently produces more parsimonious and generalizable models than traditional stepwise and penalized regression techniques.
Authors: Beni Egressy, Jan St\"uhmer
Abstract: While large language models (LLMs) demonstrate impressive capabilities across numerous applications, their robustness remains a critical concern. This paper is motivated by a specific vulnerability: the order sensitivity of LLMs. This vulnerability manifests itself as the order bias observed when LLMs decide between possible options (for example, a preference for the first option) and the tendency of LLMs to provide different answers when options are reordered. The use cases for this scenario extend beyond the classical case of multiple-choice question answering to the use of LLMs as automated evaluators in AI pipelines, comparing output generated by different models. We introduce Set-LLM, a novel architectural adaptation for pretrained LLMs that enables the processing of mixed set-text inputs with permutation invariance guarantees. The adaptations involve a new attention mask and new positional encodings specifically designed for sets. We provide a theoretical proof of invariance and demonstrate through experiments that Set-LLM can be trained effectively, achieving comparable or improved performance and maintaining the runtime of the original model, while eliminating order sensitivity.
Authors: Hossein Zakerinia, Christoph H. Lampert
Abstract: We present new fast-rate generalization bounds for multi-task and meta-learning in the unbalanced setting, i.e. when the tasks have training sets of different sizes, as is typically the case in real-world scenarios. Previously, only standard-rate bounds were known for this situation, while fast-rate bounds were limited to the setting where all training sets are of equal size. Our new bounds are numerically computable as well as interpretable, and we demonstrate their flexibility in handling a number of cases where they give stronger guarantees than previous bounds. Besides the bounds themselves, we also make conceptual contributions: we demonstrate that the unbalanced multi-task setting has different statistical properties than the balanced situation, specifically that proofs from the balanced situation do not carry over to the unbalanced setting. Additionally, we shed light on the fact that the unbalanced situation allows two meaningful definitions of multi-task risk, depending on whether if all tasks should be considered equally important or if sample-rich tasks should receive more weight than sample-poor ones.
Authors: Frederik Baymler Mathiesen, Nikolaus Vertovec, Francesco Fabiano, Luca Laurenti, Alessandro Abate
Abstract: Neural networks hold great potential to act as approximate models of nonlinear dynamical systems, with the resulting neural approximations enabling verification and control of such systems. However, in safety-critical contexts, the use of neural approximations requires formal bounds on their closeness to the underlying system. To address this fundamental challenge, we propose a novel, adaptive, and parallelizable verification method based on certified first-order models. Our approach provides formal error bounds on the neural approximations of dynamical systems, allowing them to be safely employed as surrogates by interpreting the error bound as bounded disturbances acting on the approximated dynamics. We demonstrate the effectiveness and scalability of our method on a range of established benchmarks from the literature, showing that it outperforms the state-of-the-art. Furthermore, we highlight the flexibility of our framework by applying it to two novel scenarios not previously explored in this context: neural network compression and an autoencoder-based deep learning architecture for learning Koopman operators, both yielding compelling results.
Authors: Mahesh Godavarti
Abstract: We introduce a new algebraic structure for multi-dimensional compositional embeddings, built on directional non-commutative monoidal operators. The core contribution of this work is this novel framework, which exhibits appealing theoretical properties (associativity along each dimension and an interchange law ensuring global consistency) while remaining compatible with modern machine learning architectures. Our construction defines a distinct composition operator circ_i for each axis i, ensuring associative combination along each axis without imposing global commutativity. Importantly, all axis-specific operators commute with one another, enforcing a global interchange law that enables consistent crossaxis compositions. This is, to our knowledge, the first approach that provides a common foundation that generalizes classical sequence-modeling paradigms (e.g., structured state-space models (SSMs) and transformer self-attention) to a unified multi-dimensional framework. For example, specific one-dimensional instances of our framework can recover the familiar affine transformation algebra, vanilla self-attention, and the SSM-style recurrence. The higher-dimensional generalizations naturally support recursive, structure-aware operations in embedding spaces. We outline several potential applications unlocked by this structure-including structured positional encodings in Transformers, directional image embeddings, and symbolic modeling of sequences or grids-indicating that it could inform future deep learning model designs. We formally establish the algebraic properties of our framework and discuss efficient implementations. Finally, as our focus is theoretical, we include no experiments here and defer empirical validation to future work, which we plan to undertake.
Authors: Brandon Duderstadt, Zach Nussbaum, Laurens van der Maaten
Abstract: The rapid adoption of generative AI has driven an explosion in the size of datasets consumed and produced by AI models. Traditional methods for unstructured data visualization, such as t-SNE and UMAP, have not kept up with the pace of dataset scaling. This presents a significant challenge for AI explainability, which relies on methods such as t-SNE and UMAP for exploratory data analysis. In this paper, we introduce Negative Or Mean Affinity Discrimination (NOMAD) Projection, the first method for unstructured data visualization via nonlinear dimensionality reduction that can run on multiple GPUs at train time. We provide theory that situates NOMAD Projection as an approximate upper bound on the InfoNC-t-SNE loss, and empirical results that demonstrate NOMAD Projection's superior performance and speed profile compared to existing state-of-the-art methods. We demonstrate the scalability of NOMAD Projection by computing the first complete data map of Multilingual Wikipedia.
Authors: Soham Sane
Abstract: Proximal Policy Optimization (PPO) is a widely used reinforcement learning algorithm that heavily relies on accurate advantage estimates for stable and efficient training. However, raw advantage signals can exhibit significant variance, noise, and scale-related issues, impeding optimal learning performance. To address this challenge, we introduce Advantage Modulation PPO (AM-PPO), a novel enhancement of PPO that adaptively modulates advantage estimates using a dynamic, non-linear scaling mechanism. This adaptive modulation employs an alpha controller that dynamically adjusts the scaling factor based on evolving statistical properties of the advantage signals, such as their norm, variance, and a predefined target saturation level. By incorporating a tanh-based gating function driven by these adaptively scaled advantages, AM-PPO reshapes the advantage signals to stabilize gradient updates and improve the conditioning of the policy gradient landscape. Crucially, this modulation also influences value function training by providing consistent and adaptively conditioned learning targets. Empirical evaluations across standard continuous control benchmarks demonstrate that AM-PPO achieves superior reward trajectories, exhibits sustained learning progression, and significantly reduces the clipping required by adaptive optimizers. These findings underscore the potential of advantage modulation as a broadly applicable technique for enhancing reinforcement learning optimization.
Authors: Christiaan Meijer, E. G. Patrick Bos
Abstract: While eXplainable AI (XAI) has advanced significantly, few methods address interpretability in embedded vector spaces where dimensions represent complex abstractions. We introduce Distance Explainer, a novel method for generating local, post-hoc explanations of embedded spaces in machine learning models. Our approach adapts saliency-based techniques from RISE to explain the distance between two embedded data points by assigning attribution values through selective masking and distance-ranked mask filtering. We evaluate Distance Explainer on cross-modal embeddings (image-image and image-caption pairs) using established XAI metrics including Faithfulness, Sensitivity/Robustness, and Randomization. Experiments with ImageNet and CLIP models demonstrate that our method effectively identifies features contributing to similarity or dissimilarity between embedded data points while maintaining high robustness and consistency. We also explore how parameter tuning, particularly mask quantity and selection strategy, affects explanation quality. This work addresses a critical gap in XAI research and enhances transparency and trustworthiness in deep learning applications utilizing embedded spaces.
Authors: Haruki Settai, Naoya Takeishi, Takehisa Yairi
Abstract: For continuous systems modeled by dynamical equations such as ODEs and SDEs, Bellman's principle of optimality takes the form of the Hamilton-Jacobi-Bellman (HJB) equation, which provides the theoretical target of reinforcement learning (RL). Although recent advances in RL successfully leverage this formulation, the existing methods typically assume the underlying dynamics are known a priori because they need explicit access to the coefficient functions of dynamical equations to update the value function following the HJB equation. We address this inherent limitation of HJB-based RL; we propose a model-free approach still targeting the HJB equation and propose the corresponding temporal difference method. We demonstrate its potential advantages over transition kernel-based formulations, both qualitatively and empirically. The proposed formulation paves the way toward bridging stochastic optimal control and model-free reinforcement learning.
Authors: Adrian Arnaiz-Rodriguez, Federico Errica
Abstract: After a renaissance phase in which researchers revisited the message-passing paradigm through the lens of deep learning, the graph machine learning community shifted its attention towards a deeper and practical understanding of message-passing's benefits and limitations. In this position paper, we notice how the fast pace of progress around the topics of oversmoothing and oversquashing, the homophily-heterophily dichotomy, and long-range tasks, came with the consolidation of commonly accepted beliefs and assumptions that are not always true nor easy to distinguish from each other. We argue that this has led to ambiguities around the investigated problems, preventing researchers from focusing on and addressing precise research questions while causing a good amount of misunderstandings. Our contribution wants to make such common beliefs explicit and encourage critical thinking around these topics, supported by simple but noteworthy counterexamples. The hope is to clarify the distinction between the different issues and promote separate but intertwined research directions to address them.
Authors: Suvadeep Hajra
Abstract: Transformer language models have driven significant progress across various fields, including natural language processing and computer vision. A central component of these models is the self-attention (SA) mechanism, which learns rich vector representations of tokens by modeling their relationships with others in a sequence. However, despite extensive research, transformers continue to suffer from training instability -- often manifesting as spikes or divergence in the training loss during a run. In this work, we identify one source of this instability: SA's limited ability to capture short-range dependencies, especially in tasks like language modeling, where almost every token heavily relies on its nearby neighbors. This limitation causes the pre-softmax logits of SA to grow rapidly, destabilizing training. To address this, we propose decomposing the SA into local (short-range) and global (long-range) attention heads. This decomposed attention, referred to as Long Short-attention (LS-attention), mitigates logit explosion and results in more stable training compared to an equivalent multi-head self-attention (MHSA). Empirical comparisons with two alternative training stabilization methods show that LS-attention reduces the validation perplexity to nearly 2/5 of that achieved by one method and reaches a similar perplexity as the other method using only 1/20 of the GPU hours. Additionally, our experiments demonstrate that LS-attention reduces inference latency by up to 36% compared to a state-of-the-art implementation of equivalent MHSA.
Authors: Julian Oelhaf, Georg Kordowich, Changhun Kim, Paula Andrea Perez-Toro, Andreas Maier, Johann Jager, Siming Bayer
Abstract: Germany's transition to a renewable energy-based power system is reshaping grid operations, requiring advanced monitoring and control to manage decentralized generation. Machine learning (ML) has emerged as a powerful tool for power system protection, particularly for fault detection (FD) and fault line identification (FLI) in transmission grids. However, ML model reliability depends on data quality and availability. Data sparsity resulting from sensor failures, communication disruptions, or reduced sampling rates poses a challenge to ML-based FD and FLI. Yet, its impact has not been systematically validated prior to this work. In response, we propose a framework to assess the impact of data sparsity on ML-based FD and FLI performance. We simulate realistic data sparsity scenarios, evaluate their impact, derive quantitative insights, and demonstrate the effectiveness of this evaluation strategy by applying it to an existing ML-based framework. Results show the ML model remains robust for FD, maintaining an F1-score of 0.999 $\pm$ 0.000 even after a 50x data reduction. In contrast, FLI is more sensitive, with performance decreasing by 55.61% for missing voltage measurements and 9.73% due to communication failures at critical network points. These findings offer actionable insights for optimizing ML models for real-world grid protection. This enables more efficient FD and supports targeted improvements in FLI.
Authors: Marko Tuononen, Duy Vu, Dani Korpi, Vesa Starck, Ville Hautam\"aki
Abstract: Concept discovery in neural networks often targets individual neurons or human-interpretable features, overlooking distributed layer-wide patterns. We study the Neural Activation Pattern (NAP) methodology, which clusters full-layer activation distributions to identify such layer-level concepts. Applied to visual object recognition and radio receiver models, we propose improved normalization, distribution estimation, distance metrics, and varied cluster selection. In the radio receiver model, distinct concepts did not emerge; instead, a continuous activation manifold shaped by Signal-to-Noise Ratio (SNR) was observed -- highlighting SNR as a key learned factor, consistent with classical receiver behavior and supporting physical plausibility. Our enhancements to NAP improved in-distribution vs. out-of-distribution separation, suggesting better generalization and indirectly validating clustering quality. These results underscore the importance of clustering design and activation manifolds in interpreting and troubleshooting neural network behavior.
Authors: Wangyang Ying, Haoyue Bai, Nanxu Gong, Xinyuan Wang, Sixun Dong, Haifeng Chen, Yanjie Fu
Abstract: The data-to-equation (Data2Eqn) task aims to discover interpretable mathematical equations that map observed values to labels, offering physical insights and broad applicability across academic and industrial domains. Genetic programming and traditional deep learning-based approaches suffer from search inefficiency and poor generalization on small task-specific datasets. Foundation models showed promise in this area, but existing approaches suffer from: 1) They are pretrained on general-purpose data distributions, making them less effective for domain-specific tasks; and 2) their training objectives focus on token-level alignment, overlooking mathematical semantics, which can lead to inaccurate equations. To address these issues, we aim to enhance the domain adaptability of foundation models for Data2Eqn tasks. In this work, we propose a reinforcement learning-based finetuning framework that directly optimizes the generation policy of a pretrained model through reward signals derived from downstream numerical fitness. Our method allows the model to adapt to specific and complex data distributions and generate mathematically meaningful equations. Extensive experiments demonstrate that our approach improves both the accuracy and robustness of equation generation under complex distributions.
Authors: Hossein Zakerinia, Jonathan Scott, Christoph H. Lampert
Abstract: Personalized federated learning has emerged as a popular approach to training on devices holding statistically heterogeneous data, known as clients. However, most existing approaches require a client to have labeled data for training or finetuning in order to obtain their own personalized model. In this paper we address this by proposing FLowDUP, a novel method that is able to generate a personalized model using only a forward pass with unlabeled data. The generated model parameters reside in a low-dimensional subspace, enabling efficient communication and computation. FLowDUP's learning objective is theoretically motivated by our new transductive multi-task PAC-Bayesian generalization bound, that provides performance guarantees for unlabeled clients. The objective is structured in such a way that it allows both clients with labeled data and clients with only unlabeled data to contribute to the training process. To supplement our theoretical results we carry out a thorough experimental evaluation of FLowDUP, demonstrating strong empirical performance on a range of datasets with differing sorts of statistically heterogeneous clients. Through numerous ablation studies, we test the efficacy of the individual components of the method.
Authors: Carlos Stein Brito, Daniel McNamee
Abstract: Deploying learned control policies in real-world environments poses a fundamental challenge. When system dynamics change unexpectedly, performance degrades until models are retrained on new data. We introduce Reflexive World Models (RWM), a dual control framework that uses world model predictions as implicit reference trajectories for rapid adaptation. Our method separates the control problem into long-term reward maximization through reinforcement learning and robust motor execution through rapid latent control. This dual architecture achieves significantly faster adaptation with low online computational cost compared to model-based RL baselines, while maintaining near-optimal performance. The approach combines the benefits of flexible policy learning through reinforcement learning with rapid error correction capabilities, providing a principled approach to maintaining performance in high-dimensional continuous control tasks under varying dynamics.
Authors: Yury Belousov, Brian Pulfer, Vitaliy Kinakh, Slava Voloshynovskiy
Abstract: While foundation models demonstrate impressive performance across various tasks, they remain vulnerable to adversarial inputs. Current research explores various approaches to enhance model robustness, with Diffusion Denoised Smoothing emerging as a particularly promising technique. This method employs a pretrained diffusion model to preprocess inputs before model inference. Yet, its effectiveness remains largely unexplored beyond classification. We aim to address this gap by analyzing three datasets with four distinct downstream tasks under three different adversarial attack algorithms. Our findings reveal that while foundation models maintain resilience against conventional transformations, applying high-noise diffusion denoising to clean images without any distortions significantly degrades performance by as high as 57%. Low-noise diffusion settings preserve performance but fail to provide adequate protection across all attack types. Moreover, we introduce a novel attack strategy specifically targeting the diffusion process itself, capable of circumventing defenses in the low-noise regime. Our results suggest that the trade-off between adversarial robustness and performance remains a challenge to be addressed.
Authors: Patrick Cheridito, Jean-Loup Dupret, Donatien Hainaut
Abstract: In this paper, we introduce a model-based deep-learning approach to solve finite-horizon continuous-time stochastic control problems with jumps. We iteratively train two neural networks: one to represent the optimal policy and the other to approximate the value function. Leveraging a continuous-time version of the dynamic programming principle, we derive two different training objectives based on the Hamilton-Jacobi-Bellman equation, ensuring that the networks capture the underlying stochastic dynamics. Empirical evaluations on different problems illustrate the accuracy and scalability of our approach, demonstrating its effectiveness in solving complex, high-dimensional stochastic control tasks.
Authors: Pietro Bartoli, Christian Veronesi, Andrea Giudici, David Siorpaes, Diana Trojaniello, Franco Zappa
Abstract: The rise of IoT has increased the need for on-edge machine learning, with TinyML emerging as a promising solution for resource-constrained devices such as MCU. However, evaluating their performance remains challenging due to diverse architectures and application scenarios. Current solutions have many non-negligible limitations. This work introduces an alternative benchmarking methodology that integrates energy and latency measurements while distinguishing three execution phases pre-inference, inference, and post-inference. Additionally, the setup ensures that the device operates without being powered by an external measurement unit, while automated testing can be leveraged to enhance statistical significance. To evaluate our setup, we tested the STM32N6 MCU, which includes a NPU for executing neural networks. Two configurations were considered: high-performance and Low-power. The variation of the EDP was analyzed separately for each phase, providing insights into the impact of hardware configurations on energy efficiency. Each model was tested 1000 times to ensure statistically relevant results. Our findings demonstrate that reducing the core voltage and clock frequency improve the efficiency of pre- and post-processing without significantly affecting network execution performance. This approach can also be used for cross-platform comparisons to determine the most efficient inference platform and to quantify how pre- and post-processing overhead varies across different hardware implementations.
Authors: H. V. AlquBoj, Hilal AlQuabeh, Velibor Bojkovic, Munachiso Nwadike, Kentaro Inui
Abstract: Grokking, a delayed generalization in neural networks after perfect training performance, has been observed in Transformers and MLPs, but the components driving it remain underexplored. We show that embeddings are central to grokking: introducing them into MLPs induces delayed generalization in modular arithmetic tasks, whereas MLPs without embeddings can generalize immediately. Our analysis identifies two key mechanisms: (1) Embedding update dynamics, where rare tokens stagnate due to sparse gradient updates and weight decay, and (2) Bilinear coupling, where the interaction between embeddings and downstream weights introduces saddle points and increases sensitivity to initialization. To confirm these mechanisms, we investigate frequency-aware sampling, which balances token updates by minimizing gradient variance, and embedding-specific learning rates, derived from the asymmetric curvature of the bilinear loss landscape. We prove that an adaptive learning rate ratio, \(\frac{\eta_E}{\eta_W} \propto \frac{\sigma_{\max}(E)}{\sigma_{\max}(W)} \cdot \frac{f_W}{f_E}\), mitigates bilinear coupling effects, accelerating convergence. Our methods not only improve grokking dynamics but also extend to broader challenges in Transformer optimization, where bilinear interactions hinder efficient training.
Authors: Jacopo Teneggi, Zhenzhen Wang, Paul H. Yi, Tianmin Shu, Jeremias Sulam
Abstract: Machine learning explainability aims to make the decision-making process of black-box models more transparent by finding the most important input features for a given prediction task. Recent works have proposed composing explanations from semantic concepts (e.g., colors, patterns, shapes) that are inherently interpretable to the user of a model. However, these methods generally ignore the communicative context of explanation-the ability of the user to understand the prediction of the model from the explanation. For example, while a medical doctor might understand an explanation in terms of clinical markers, a patient may need a more accessible explanation to make sense of the same diagnosis. In this paper, we address this gap with listener-adaptive explanations. We propose an iterative procedure grounded in principles of pragmatic reasoning and the rational speech act to generate explanations that maximize communicative utility. Our procedure only needs access to pairwise preferences between candidate explanations, relevant in real-world scenarios where a listener model may not be available. We evaluate our method in image classification tasks, demonstrating improved alignment between explanations and listener preferences across three datasets. Furthermore, we perform a user study that demonstrates our explanations increase communicative utility.
Authors: Nick Kocher, Christian Wassermann, Leona Hennig, Jonas Seng, Holger Hoos, Kristian Kersting, Marius Lindauer, Matthias M\"uller
Abstract: Neural Architecture Search (NAS) accelerates progress in deep learning through systematic refinement of model architectures. The downside is increasingly large energy consumption during the search process. Surrogate-based benchmarking mitigates the cost of full training by querying a pre-trained surrogate to obtain an estimate for the quality of the model. Specifically, energy-aware benchmarking aims to make it possible for NAS to favourably trade off model energy consumption against accuracy. Towards this end, we propose three design principles for such energy-aware benchmarks: (i) reliable power measurements, (ii) a wide range of GPU usage, and (iii) holistic cost reporting. We analyse EA-HAS-Bench based on these principles and find that the choice of GPU measurement API has a large impact on the quality of results. Using the Nvidia System Management Interface (SMI) on top of its underlying library influences the sampling rate during the initial data collection, returning faulty low-power estimations. This results in poor correlation with accurate measurements obtained from an external power meter. With this study, we bring to attention several key considerations when performing energy-aware surrogate-based benchmarking and derive first guidelines that can help design novel benchmarks. We show a narrow usage range of the four GPUs attached to our device, ranging from 146 W to 305 W in a single-GPU setting, and narrowing down even further when using all four GPUs. To improve holistic energy reporting, we propose calibration experiments over assumptions made in popular tools, such as Code Carbon, thus achieving reductions in the maximum inaccuracy from 10.3 % to 8.9 % without and to 6.6 % with prior estimation of the expected load on the device.
Authors: Daniel Waxman, Fernando Llorente, Petar M. Djuri\'c
Abstract: We revisit the classical problem of Bayesian ensembles and address the challenge of learning optimal combinations of Bayesian models in an online, continual learning setting. To this end, we reinterpret existing approaches such as Bayesian model averaging (BMA) and Bayesian stacking through a novel empirical Bayes lens, shedding new light on the limitations and pathologies of BMA. Further motivated by insights from online optimization, we propose Online Bayesian Stacking (OBS), a method that optimizes the log-score over predictive distributions to adaptively combine Bayesian models. A key contribution of our work is establishing a novel connection between OBS and portfolio selection, bridging Bayesian ensemble learning with a rich, well-studied theoretical framework that offers efficient algorithms and extensive regret analysis. We further clarify the relationship between OBS and online BMA, showing that they optimize related but distinct cost functions. Through theoretical analysis and empirical evaluation, we identify scenarios where OBS outperforms online BMA and provide principled guidance on when practitioners should prefer one approach over the other.
Authors: Lan V. Truong
Abstract: We study the problem of best-arm identification in stochastic multi-armed bandits under the fixed-confidence setting, with a particular focus on instances that admit multiple optimal arms. While the Track-and-Stop algorithm of Garivier and Kaufmann (2016) is widely conjectured to be instance-optimal, its performance in the presence of multiple optima has remained insufficiently understood. In this work, we revisit the Track-and-Stop strategy and propose a modified stopping rule that ensures instance-optimality even when the set of optimal arms is not a singleton. Our analysis introduces a new information-theoretic lower bound that explicitly accounts for multiple optimal arms, and we demonstrate that our stopping rule tightly matches this bound.
Authors: Youming Tao, Zuyuan Zhang, Dongxiao Yu, Xiuzhen Cheng, Falko Dressler, Di Wang
Abstract: We investigate the problem of finding second-order stationary points (SOSP) in differentially private (DP) stochastic non-convex optimization. Existing methods suffer from two key limitations: (i) inaccurate convergence error rate due to overlooking gradient variance in the saddle point escape analysis, and (ii) dependence on auxiliary private model selection procedures for identifying DP-SOSP, which can significantly impair utility, particularly in distributed settings. To address these issues, we propose a generic perturbed stochastic gradient descent (PSGD) framework built upon Gaussian noise injection and general gradient oracles. A core innovation of our framework is using model drift distance to determine whether PSGD escapes saddle points, ensuring convergence to approximate local minima without relying on second-order information or additional DP-SOSP identification. By leveraging the adaptive DP-SPIDER estimator as a specific gradient oracle, we develop a new DP algorithm that rectifies the convergence error rates reported in prior work. We further extend this algorithm to distributed learning with arbitrarily heterogeneous data, providing the first formal guarantees for finding DP-SOSP in such settings. Our analysis also highlights the detrimental impacts of private selection procedures in distributed learning under high-dimensional models, underscoring the practical benefits of our design. Numerical experiments on real-world datasets validate the efficacy of our approach.
Authors: Harmender Gahlawat, Meirav Zehavi
Abstract: Decision trees are a fundamental tool in machine learning for representing, classifying, and generalizing data. It is desirable to construct ``small'' decision trees, by minimizing either the \textit{size} ($s$) or the \textit{depth} $(d)$ of the \textit{decision tree} (\textsc{DT}). Recently, the parameterized complexity of \textsc{Decision Tree Learning} has attracted a lot of attention. We consider a generalization of \textsc{Decision Tree Learning} where given a \textit{classification instance} $E$ and an integer $t$, the task is to find a ``small'' \textsc{DT} that disagrees with $E$ in at most $t$ examples. We consider two problems: \textsc{DTSO} and \textsc{DTDO}, where the goal is to construct a \textsc{DT} minimizing $s$ and $d$, respectively. We first establish that both \textsc{DTSO} and \textsc{DTDO} are W[1]-hard when parameterized by $s+\delta_{max}$ and $d+\delta_{max}$, respectively, where $\delta_{max}$ is the maximum number of features in which two differently labeled examples can differ. We complement this result by showing that these problems become \textsc{FPT} if we include the parameter $t$. We also consider the kernelization complexity of these problems and establish several positive and negative results for both \textsc{DTSO} and \textsc{DTDO}.
Authors: Cheng Yan, Felix Mohr, Tom Viering
Abstract: Sample-wise learning curves plot performance versus training set size. They are useful for studying scaling laws and speeding up hyperparameter tuning and model selection. Learning curves are often assumed to be well-behaved: monotone (i.e. improving with more data) and convex. By constructing the Learning Curves Database 1.1 (LCDB 1.1), a large-scale database with high-resolution learning curves, we show that learning curves are less often well-behaved than previously thought. Using statistically rigorous methods, we observe significant ill-behavior in approximately 14% of the learning curves, almost twice as much as in previous estimates. We also identify which learners are to blame and show that specific learners are more ill-behaved than others. Additionally, we demonstrate that different feature scalings rarely resolve ill-behavior. We evaluate the impact of ill-behavior on downstream tasks, such as learning curve fitting and model selection, and find it poses significant challenges, underscoring the relevance and potential of LCDB 1.1 as a challenging benchmark for future research.
Authors: Sina Mohammad-Taheri, Matthew J. Colbrook, Simone Brugiapaglia
Abstract: Gradient-based learning imposes (deep) neural networks to be differentiable at all steps. This includes model-based architectures constructed by unrolling iterations of an iterative algorithm onto layers of a neural network, known as algorithm unrolling. However, greedy sparse recovery algorithms depend on the non-differentiable argsort operator, which hinders their integration into neural networks. In this paper, we address this challenge in Orthogonal Matching Pursuit (OMP) and Iterative Hard Thresholding (IHT), two popular representative algorithms in this class. We propose permutation-based variants of these algorithms and approximate permutation matrices using "soft" permutation matrices derived from softsort, a continuous relaxation of argsort. We demonstrate -- both theoretically and numerically -- that Soft-OMP and Soft-IHT, as differentiable counterparts of OMP and IHT and fully compatible with neural network training, effectively approximate these algorithms with a controllable degree of accuracy. This leads to the development of OMP- and IHT-Net, fully trainable network architectures based on Soft-OMP and Soft-IHT, respectively. Finally, by choosing weights as "structure-aware" trainable parameters, we connect our approach to structured sparse recovery and demonstrate its ability to extract latent sparsity patterns from data.
Authors: Davide Scassola, Sebastiano Saccani, Luca Bortolussi
Abstract: Data synthesis is gaining momentum as a privacy-enhancing technology. While single-table tabular data generation has seen considerable progress, current methods for multi-table data often lack the flexibility and expressiveness needed to capture complex relational structures. In particular, they struggle with long-range dependencies and complex foreign-key relationships, such as tables with multiple parent tables or multiple types of links between the same pair of tables. We propose a generative model for relational data that generates the content of a relational dataset given the graph formed by the foreign-key relationships. We do this by learning a deep generative model of the content of the whole relational database by flow matching, where the neural network trained to denoise records leverages a graph neural network to obtain information from connected records. Our method is flexible, as it can support relational datasets with complex structures, and expressive, as the generation of each record can be influenced by any other record within the same connected component. We evaluate our method on several benchmark datasets and show that it achieves state-of-the-art performance in terms of synthetic data fidelity.
Authors: Leonardo N. Coregliano, Maryanthe Malliaris
Abstract: Recently, the authors introduced the theory of high-arity PAC learning, which is well-suited for learning graphs, hypergraphs and relational structures. In the same initial work, the authors proved a high-arity analogue of the Fundamental Theorem of Statistical Learning that almost completely characterizes all notions of high-arity PAC learning in terms of a combinatorial dimension, called the Vapnik--Chervonenkis--Natarajan (VCN${}_k$) $k$-dimension, leaving as an open problem only the characterization of non-partite, non-agnostic high-arity PAC learnability. In this work, we complete this characterization by proving that non-partite non-agnostic high-arity PAC learnability implies a high-arity version of the Haussler packing property, which in turn implies finiteness of VCN${}_k$-dimension. This is done by obtaining direct proofs that classic PAC learnability implies classic Haussler packing property, which in turn implies finite Natarajan dimension and noticing that these direct proofs nicely lift to high-arity.
Authors: Xingyu Zhou, Yulian Wu, Francesco Orabona
Abstract: In this paper, we theoretically investigate the effects of noisy labels in offline alignment, with a focus on the interplay between privacy and robustness against adversarial corruption. Specifically, under linear modeling assumptions, we present a unified analysis covering both reinforcement learning from human feedback (RLHF) and direct preference optimization (DPO) under different privacy-corruption scenarios, such as Local differential privacy-then-Corruption (LTC), where human preference labels are privatized before being corrupted by an adversary, and Corruption-then-Local differential privacy (CTL), where labels are corrupted before privacy protection. Our analysis leverages a reduction framework that reduces the offline alignment problem under linear modeling assumptions to parameter estimation in logistic regression. This framework allows us to establish an interesting separation result between LTC and CTL, demonstrating that LTC presents a greater challenge than CTL in offline alignment, even under linear models. As important by-products, our findings also advance the state-of-the-art theoretical results in offline alignment under privacy-only or corruption-only scenarios.
Authors: Coby Penso, Bar Mahpud, Jacob Goldberger, Or Sheffet
Abstract: Conformal prediction (CP) provides sets of candidate classes with a guaranteed probability of containing the true class. However, it typically relies on a calibration set with clean labels. We address privacy-sensitive scenarios where the aggregator is untrusted and can only access a perturbed version of the true labels. We propose two complementary approaches under local differential privacy (LDP). In the first approach, users do not access the model but instead provide their input features and a perturbed label using a k-ary randomized response. In the second approach, which enforces stricter privacy constraints, users add noise to their conformity score by binary search response. This method requires access to the classification model but preserves both data and label privacy. Both approaches compute the conformal threshold directly from noisy data without accessing the true labels. We prove finite-sample coverage guarantees and demonstrate robust coverage even under severe randomization. This approach unifies strong local privacy with predictive uncertainty control, making it well-suited for sensitive applications such as medical imaging or large language model queries, regardless of whether users can (or are willing to) compute their own scores.
Authors: Jingzhe Liu, Zhigang Hua, Yan Xie, Bingheng Li, Harry Shomer, Yu Song, Kaveh Hassani, Jiliang Tang
Abstract: Temporal Graph Neural Networks (TGNNs) have gained growing attention for modeling and predicting structures in temporal graphs. However, existing TGNNs primarily focus on pairwise interactions while overlooking higher-order structures that are integral to link formation and evolution in real-world temporal graphs. Meanwhile, these models often suffer from efficiency bottlenecks, further limiting their expressive power. To tackle these challenges, we propose a Higher-order structure Temporal Graph Neural Network, which incorporates hypergraph representations into temporal graph learning. In particular, we develop an algorithm to identify the underlying higher-order structures, enhancing the model's ability to capture the group interactions. Furthermore, by aggregating multiple edge features into hyperedge representations, HTGN effectively reduces memory cost during training. We theoretically demonstrate the enhanced expressiveness of our approach and validate its effectiveness and efficiency through extensive experiments on various real-world temporal graphs. Experimental results show that HTGN achieves superior performance on dynamic link prediction while reducing memory costs by up to 50\% compared to existing methods.
Authors: Kanan Kiguchi, Yunhao Tu, Katsuhiro Ajito, Fady Alnajjar, Kazuyuki Murase
Abstract: We propose a novel framework for integrating fragmented multi-modal data in Alzheimer's disease (AD) research using large language models (LLMs) and knowledge graphs. While traditional multimodal analysis requires matched patient IDs across datasets, our approach demonstrates population-level integration of MRI, gene expression, biomarkers, EEG, and clinical indicators from independent cohorts. Statistical analysis identified significant features in each modality, which were connected as nodes in a knowledge graph. LLMs then analyzed the graph to extract potential correlations and generate hypotheses in natural language. This approach revealed several novel relationships, including a potential pathway linking metabolic risk factors to tau protein abnormalities via neuroinflammation (r>0.6, p<0.001), and unexpected correlations between frontal EEG channels and specific gene expression profiles (r=0.42-0.58, p<0.01). Cross-validation with independent datasets confirmed the robustness of major findings, with consistent effect sizes across cohorts (variance <15%). The reproducibility of these findings was further supported by expert review (Cohen's k=0.82) and computational validation. Our framework enables cross modal integration at a conceptual level without requiring patient ID matching, offering new possibilities for understanding AD pathology through fragmented data reuse and generating testable hypotheses for future research.
Authors: Palash Chatterjee, Roni Khardon
Abstract: Continuous time systems are often modeled using discrete time dynamics but this requires a small simulation step to maintain accuracy. In turn, this requires a large planning horizon which leads to computationally demanding planning problems and reduced performance. Previous work in model free reinforcement learning has partially addressed this issue using action repeats where a policy is learned to determine a discrete action duration. Instead we propose to control the continuous decision timescale directly by using temporally-extended actions and letting the planner treat the duration of the action as an additional optimization variable along with the standard action variables. This additional structure has multiple advantages. It speeds up simulation time of trajectories and, importantly, it allows for deep horizon search in terms of primitive actions while using a shallow search depth in the planner. In addition, in the model based reinforcement learning (MBRL) setting, it reduces compounding errors from model learning and improves training time for models. We show that this idea is effective and that the range for action durations can be automatically selected using a multi-armed bandit formulation and integrated into the MBRL framework. An extensive experimental evaluation both in planning and in MBRL, shows that our approach yields faster planning, better solutions, and that it enables solutions to problems that are not solved in the standard formulation.
Authors: Jorge Bacca
Abstract: Deep learning-based models have demonstrated remarkable success in solving illposed inverse problems; however, many fail to strictly adhere to the physical constraints imposed by the measurement process. In this work, we introduce a projection-based correction method to enhance the inference of deep inverse networks by ensuring consistency with the forward model. Specifically, given an initial estimate from a learned reconstruction network, we apply a projection step that constrains the solution to lie within the valid solution space of the inverse problem. We theoretically demonstrate that if the recovery model is a well-trained deep inverse network, the solution can be decomposed into range-space and null-space components, where the projection-based correction reduces to an identity transformation. Extensive simulations and experiments validate the proposed method, demonstrating improved reconstruction accuracy across diverse inverse problems and deep network architectures.
Authors: Pedro P. Santos, Alberto Sardinha, Francisco S. Melo
Abstract: In this work, we contribute the first approach to solve infinite-horizon discounted general-utility Markov decision processes (GUMDPs) in the single-trial regime, i.e., when the agent's performance is evaluated based on a single trajectory. First, we provide some fundamental results regarding policy optimization in the single-trial regime, investigating which class of policies suffices for optimality, casting our problem as a particular MDP that is equivalent to our original problem, as well as studying the computational hardness of policy optimization in the single-trial regime. Second, we show how we can leverage online planning techniques, in particular a Monte-Carlo tree search algorithm, to solve GUMDPs in the single-trial regime. Third, we provide experimental results showcasing the superior performance of our approach in comparison to relevant baselines.
Authors: Jun Wan, Lingrui Mei
Abstract: The rapid advancement of large language models (LLMs) calls for a rigorous theoretical framework to explain their empirical success. While significant progress has been made in understanding LLM behaviors, existing theoretical frameworks remain fragmented in explaining emergent phenomena through a unified mathematical lens. We establish the first formal connection between LLM architectures and Algorithmic Information Theory (AIT) by proving two fundamental results: (1) the training process computationally approximates Solomonoff prior through loss minimization interpreted as program length optimization, and (2) next-token prediction implements approximate Solomonoff induction. We leverage AIT to provide a unified theoretical explanation for in-context learning, few-shot learning, and scaling laws. Furthermore, our theoretical insights lead to a principled method for few-shot example selection that prioritizes samples where models exhibit lower predictive confidence. We demonstrate through experiments on diverse text classification benchmarks that this strategy yields significant performance improvements, particularly for smaller model architectures, when compared to selecting high-confidence examples. Our framework bridges the gap between theoretical foundations and practical LLM behaviors, providing both explanatory power and actionable insights for future model development.
Authors: Zahra Khatti, Daniel P. Robinson, Frank E. Curtis
Abstract: A new strategy for fair supervised machine learning is proposed. The main advantages of the proposed strategy as compared to others in the literature are as follows. (a) We introduce a new smooth nonconvex surrogate to approximate the Heaviside functions involved in discontinuous unfairness measures. The surrogate is based on smoothing methods from the optimization literature, and is new for the fair supervised learning literature. The surrogate is a tight approximation which ensures the trained prediction models are fair, as opposed to other (e.g., convex) surrogates that can fail to lead to a fair prediction model in practice. (b) Rather than rely on regularizers (that lead to optimization problems that are difficult to solve) and corresponding regularization parameters (that can be expensive to tune), we propose a strategy that employs hard constraints so that specific tolerances for unfairness can be enforced without the complications associated with the use of regularization. (c)~Our proposed strategy readily allows for constraints on multiple (potentially conflicting) unfairness measures at the same time. Multiple measures can be considered with a regularization approach, but at the cost of having even more difficult optimization problems to solve and further expense for tuning. By contrast, through hard constraints, our strategy leads to optimization models that can be solved tractably with minimal tuning.
Authors: Taehoon Kim, Henry Gouk, Minyoung Kim, Timothy Hospedales
Abstract: Certifying the IID generalisation ability of deep networks is the first of many requirements for trusting AI in high-stakes applications from medicine to security. However, when instantiating generalisation bounds for deep networks it remains challenging to obtain non-vacuous guarantees, especially when applying contemporary large models on the small scale data prevalent in such high-stakes fields. In this paper, we draw a novel connection between a family of learning methods based on model fusion and generalisation certificates, and surprisingly show that with minor adjustment several existing learning strategies already provide non-trivial generalisation guarantees. Essentially, by focusing on data-driven learning of downstream tasks by fusion rather than fine-tuning, the certified generalisation gap becomes tiny and independent of the base network size, facilitating its certification. Our results show for the first time non-trivial generalisation guarantees for learning with as low as 100 examples, while using vision models such as VIT-B and language models such as mistral-7B. This observation is significant as it has immediate implications for facilitating the certification of existing systems as trustworthy, and opens up new directions for research at the intersection of practice and theory.
Authors: Sarah E. Wessinger, Leslie N. Smith, Jacob Gull, Jonathan Gehman, Zachary Beever, Andrew J. Kammerer
Abstract: Accurately estimating the refractive environment over multiple frequencies within the marine atmospheric boundary layer is crucial for the effective deployment of radar technologies. Traditional parabolic equation simulations, while effective, can be computationally expensive and time-intensive, limiting their practical application. This communication explores a novel approach using deep neural networks to estimate the pattern propagation factor, a critical parameter for characterizing environmental impacts on signal propagation. Image-to-image translation generators designed to ingest modified refractivity data and generate predictions of pattern propagation factors over the same domain were developed. Findings demonstrate that deep neural networks can be trained to analyze multiple frequencies and reasonably predict the pattern propagation factor, offering an alternative to traditional methods.
Authors: Dheeraj Baby, Yifei Tang, Hieu Duy Nguyen, Yu-Xiang Wang, Rohit Pyati
Abstract: In this paper, we study the problem of estimation and learning under temporal distribution shift. Consider an observation sequence of length $n$, which is a noisy realization of a time-varying groundtruth sequence. Our focus is to develop methods to estimate the groundtruth at the final time-step while providing sharp point-wise estimation error rates. We show that, without prior knowledge on the level of temporal shift, a wavelet soft-thresholding estimator provides an optimal estimation error bound for the groundtruth. Our proposed estimation method generalizes existing researches Mazzetto and Upfal (2023) by establishing a connection between the sequence's non-stationarity level and the sparsity in the wavelet-transformed domain. Our theoretical findings are validated by numerical experiments. Additionally, we applied the estimator to derive sparsity-aware excess risk bounds for binary classification under distribution shift and to develop computationally efficient training objectives. As a final contribution, we draw parallels between our results and the classical signal processing problem of total-variation denoising (Mammen and van de Geer,1997; Tibshirani, 2014), uncovering novel optimal algorithms for such task.
Authors: Carlos Rodriguez-Pardo, Leonardo Chiani, Emanuele Borgonovo, Massimo Tavoni
Abstract: We present a neural framework for learning conditional optimal transport (OT) maps between probability distributions. Our approach introduces a conditioning mechanism capable of processing both categorical and continuous conditioning variables simultaneously. At the core of our method lies a hypernetwork that generates transport layer parameters based on these inputs, creating adaptive mappings that outperform simpler conditioning methods. Comprehensive ablation studies demonstrate the superior performance of our method over baseline configurations. Furthermore, we showcase an application to global sensitivity analysis, offering high performance in computing OT-based sensitivity indices. This work advances the state-of-the-art in conditional optimal transport, enabling broader application of optimal transport principles to complex, high-dimensional domains such as generative modeling and black-box model explainability.
Authors: Eric J. Michaud, Asher Parker-Sartori, Max Tegmark
Abstract: We study the problem of creating strong, yet narrow, AI systems. While recent AI progress has been driven by the training of large general-purpose foundation models, the creation of smaller models specialized for narrow domains could be valuable for both efficiency and safety. In this work, we explore two challenges involved in creating such systems, having to do with basic properties of how neural networks learn and structure their representations. The first challenge regards when it is possible to train narrow models from scratch. Through experiments on a synthetic task, we find that it is sometimes necessary to train networks on a wide distribution of data to learn certain narrow skills within that distribution. This effect arises when skills depend on each other hierarchically, and training on a broad distribution introduces a curriculum which substantially accelerates learning. The second challenge regards how to transfer particular skills from large general models into small specialized models. We find that model skills are often not perfectly localized to a particular set of prunable components. However, we find that methods based on pruning can still outperform distillation. We investigate the use of a regularization objective to align desired skills with prunable components while unlearning unnecessary skills.
Authors: Muquan Yu, Mu Nan, Hossein Adeli, Jacob S. Prince, John A. Pyles, Leila Wehbe, Margaret M. Henderson, Michael J. Tarr, Andrew F. Luo
Abstract: Understanding functional representations within higher visual cortex is a fundamental question in computational neuroscience. While artificial neural networks pretrained on large-scale datasets exhibit striking representational alignment with human neural responses, learning image-computable models of visual cortex relies on individual-level, large-scale fMRI datasets. The necessity for expensive, time-intensive, and often impractical data acquisition limits the generalizability of encoders to new subjects and stimuli. BraInCoRL uses in-context learning to predict voxelwise neural responses from few-shot examples without any additional finetuning for novel subjects and stimuli. We leverage a transformer architecture that can flexibly condition on a variable number of in-context image stimuli, learning an inductive bias over multiple subjects. During training, we explicitly optimize the model for in-context learning. By jointly conditioning on image features and voxel activations, our model learns to directly generate better performing voxelwise models of higher visual cortex. We demonstrate that BraInCoRL consistently outperforms existing voxelwise encoder designs in a low-data regime when evaluated on entirely novel images, while also exhibiting strong test-time scaling behavior. The model also generalizes to an entirely new visual fMRI dataset, which uses different subjects and fMRI data acquisition parameters. Further, BraInCoRL facilitates better interpretability of neural signals in higher visual cortex by attending to semantically relevant stimuli. Finally, we show that our framework enables interpretable mappings from natural language queries to voxel selectivity.
Authors: KM Khalid Saifullah, Faiaz Azmain, Habiba Hye
Abstract: Sentiment analysis plays a crucial role in understanding developer interactions, issue resolutions, and project dynamics within software engineering (SE). While traditional SE-specific sentiment analysis tools have made significant strides, they often fail to account for the nuanced and context-dependent language inherent to the domain. This study systematically evaluates the performance of bidirectional transformers, such as BERT, against generative pre-trained transformers, specifically GPT-4o-mini, in SE sentiment analysis. Using datasets from GitHub, Stack Overflow, and Jira, we benchmark the models' capabilities with fine-tuned and default configurations. The results reveal that fine-tuned GPT-4o-mini performs comparable to BERT and other bidirectional models on structured and balanced datasets like GitHub and Jira, achieving macro-averaged F1-scores of 0.93 and 0.98, respectively. However, on linguistically complex datasets with imbalanced sentiment distributions, such as Stack Overflow, the default GPT-4o-mini model exhibits superior generalization, achieving an accuracy of 85.3\% compared to the fine-tuned model's 13.1\%. These findings highlight the trade-offs between fine-tuning and leveraging pre-trained models for SE tasks. The study underscores the importance of aligning model architectures with dataset characteristics to optimize performance and proposes directions for future research in refining sentiment analysis tools tailored to the SE domain.
Authors: Jiwoo Song, Daning Huang
Abstract: This paper presents a novel parametrization approach for aeroelastic systems utilizing Koopman theory, specifically leveraging the Koopman Bilinear Form (KBF) model. To address the limitations of linear parametric dependence in the KBF model, we introduce the Extended KBF (EKBF) model, which enables a global linear representation of aeroelastic dynamics while capturing stronger nonlinear dependence on, e.g., the flutter parameter. The effectiveness of the proposed methodology is demonstrated through two case studies: a 2D academic example and a panel flutter problem. Results show that EKBF effectively interpolates and extrapolates principal eigenvalues, capturing flutter mechanisms, and accurately predicting the flutter boundary even when the data is corrupted by noise. Furthermore, parameterized isostable and isochron identified by EKBF provides valuable insights into the nonlinear flutter system.
Authors: Miguel Lopez-Duran, Julian Fierrez, Aythami Morales, Ruben Tolosana, Oscar Delgado-Mohatar, Alvaro Ortigosa
Abstract: The automatic analysis of document layouts in digital-born PDF documents remains a challenging problem due to the heterogeneous arrangement of textual and nontextual elements and the imprecision of the textual metadata in the Portable Document Format. In this work, we benchmark Graph Neural Network (GNN) architectures for the task of fine-grained layout classification of text blocks from digital native documents. We introduce two graph construction structures: a k-closest-neighbor graph and a fully connected graph, and generate node features via pre-trained text and vision models, thus avoiding manual feature engineering. Three experimental frameworks are evaluated: single-modality (text or visual), concatenated multimodal, and dual-branch multimodal. We evaluated four foundational GNN models and compared them with the baseline. Our experiments are specifically conducted on a rich dataset of public affairs documents that includes more than 20 sources (e.g., regional and national-level official gazettes), 37K PDF documents, with 441K pages in total. Our results demonstrate that GraphSAGE operating on the k-closest-neighbor graph in a dual-branch configuration achieves the highest per-class and overall accuracy, outperforming the baseline in some sources. These findings confirm the importance of local layout relationships and multimodal fusion exploited through GNNs for the analysis of native digital document layouts.
Authors: Aidan Furlong, Xingang Zhao, Robert Salko, Xu Wu
Abstract: Critical heat flux (CHF) marks the transition from nucleate to film boiling, where heat transfer to the working fluid can rapidly deteriorate. Accurate CHF prediction is essential for efficiency, safety, and preventing equipment damage, particularly in nuclear reactors. Although widely used, empirical correlations frequently exhibit discrepancies in comparison with experimental data, limiting their reliability in diverse operational conditions. Traditional machine learning (ML) approaches have demonstrated the potential for CHF prediction but have often suffered from limited interpretability, data scarcity, and insufficient knowledge of physical principles. Hybrid model approaches, which combine data-driven ML with physics-based models, mitigate these concerns by incorporating prior knowledge of the domain. This study integrated a purely data-driven ML model and two hybrid models (using the Biasi and Bowring CHF correlations) within the CTF subchannel code via a custom Fortran framework. Performance was evaluated using two validation cases: a subset of the Nuclear Regulatory Commission CHF database and the Bennett dryout experiments. In both cases, the hybrid models exhibited significantly lower error metrics in comparison with conventional empirical correlations. The pure ML model remained competitive with the hybrid models. Trend analysis of error parity indicates that ML-based models reduce the tendency for CHF overprediction, improving overall accuracy. These results demonstrate that ML-based CHF models can be effectively integrated into subchannel codes and can potentially increase performance in comparison with conventional methods.
Authors: Giovanni Catalani, Jean Fesquet, Xavier Bertrand, Fr\'ed\'eric Tost, Michael Bauerheim, Joseph Morlier
Abstract: This paper introduces a novel surrogate modeling framework for aerodynamic applications based on Neural Fields. The proposed approach, MARIO (Modulated Aerodynamic Resolution Invariant Operator), addresses non parametric geometric variability through an efficient shape encoding mechanism and exploits the discretization-invariant nature of Neural Fields. It enables training on significantly downsampled meshes, while maintaining consistent accuracy during full-resolution inference. These properties allow for efficient modeling of diverse flow conditions, while reducing computational cost and memory requirements compared to traditional CFD solvers and existing surrogate methods. The framework is validated on two complementary datasets that reflect industrial constraints. First, the AirfRANS dataset consists in a two-dimensional airfoil benchmark with non-parametric shape variations. Performance evaluation of MARIO on this case demonstrates an order of magnitude improvement in prediction accuracy over existing methods across velocity, pressure, and turbulent viscosity fields, while accurately capturing boundary layer phenomena and aerodynamic coefficients. Second, the NASA Common Research Model features three-dimensional pressure distributions on a full aircraft surface mesh, with parametric control surface deflections. This configuration confirms MARIO's accuracy and scalability. Benchmarking against state-of-the-art methods demonstrates that Neural Field surrogates can provide rapid and accurate aerodynamic predictions under the computational and data limitations characteristic of industrial applications.
Authors: Xin Zhang, Ziruo Zhang, Jiawei Du, Zuozhu Liu, Joey Tianyi Zhou
Abstract: Multimodal Dataset Distillation (MDD) seeks to condense large-scale image-text datasets into compact surrogates while retaining their effectiveness for cross-modal learning. Despite recent progress, existing MDD approaches often suffer from \textit{\textbf{Modality Collapse}}, characterized by over-concentrated intra-modal representations and enlarged distributional gap across modalities. In this paper, at the first time, we identify this issue as stemming from a fundamental conflict between the over-compression behavior inherent in dataset distillation and the cross-modal supervision imposed by contrastive objectives. To alleviate modality collapse, we introduce \textbf{RepBlend}, a novel MDD framework that weakens overdominant cross-modal supervision via representation blending, thereby significantly enhancing intra-modal diversity. Additionally, we observe that current MDD methods impose asymmetric supervision across modalities, resulting in biased optimization. To address this, we propose symmetric projection trajectory matching, which synchronizes the optimization dynamics using modality-specific projection heads, thereby promoting balanced supervision and enhancing cross-modal alignment. Experiments on Flickr-30K and MS-COCO show that RepBlend consistently outperforms prior state-of-the-art MDD methods, achieving significant gains in retrieval performance (e.g., +9.4 IR@10, +6.3 TR@10 under the 100-pair setting) and offering up to 6.7$\times$ distillation speedup.
Authors: Dionissios T. Hristopulos, Anastassia Baxevani, Giorgio Kaniadakis
Abstract: Asymmetric, non-Gaussian probability distributions are often observed in the analysis of natural and engineering datasets. The lognormal distribution is a standard model for data with skewed frequency histograms and fat tails. However, the lognormal law severely restricts the asymptotic dependence of the probability density and the hazard function for high values. Herein we present a family of three-parameter non-Gaussian probability density functions that are based on generalized kappa-exponential and kappa-logarithm functions and investigate its mathematical properties. These kappa-lognormal densities represent continuous deformations of the lognormal with lighter right tails, controlled by the parameter kappa. In addition, bimodal distributions are obtained for certain parameter combinations. We derive closed-form analytic expressions for the main statistical functions of the kappa-lognormal distribution. For the moments, we derive bounds that are based on hypergeometric functions as well as series expansions. Explicit expressions for the gradient and Hessian of the negative log-likelihood are obtained to facilitate numerical maximum-likelihood estimates of the kappa-lognormal parameters from data. We also formulate a joint probability density function for kappa-lognormal stochastic processes by applying Jacobi's multivariate theorem to a latent Gaussian process. Estimation of the kappa-lognormal distribution based on synthetic and real data is explored. Furthermore, we investigate applications of kappa-lognormal processes with different covariance kernels in time series forecasting and spatial interpolation using warped Gaussian process regression. Our results are of practical interest for modeling skewed distributions in various scientific and engineering fields.
Authors: Sahil Tomar, Rajeshwar Tripathi, Sandeep Kumar
Abstract: Bone fractures are a leading cause of morbidity and disability worldwide, imposing significant clinical and economic burdens on healthcare systems. Traditional X ray interpretation is time consuming and error prone, while existing machine learning and deep learning solutions often demand extensive feature engineering, large, annotated datasets, and high computational resources. To address these challenges, a distributed hybrid quantum classical pipeline is proposed that first applies Principal Component Analysis (PCA) for dimensionality reduction and then leverages a 4 qubit quantum amplitude encoding circuit for feature enrichment. By fusing eight PCA derived features with eight quantum enhanced features into a 16 dimensional vector and then classifying with different machine learning models achieving 99% accuracy using a public multi region X ray dataset on par with state of the art transfer learning models while reducing feature extraction time by 82%.
Authors: Xigui Li, Yuanye Zhou, Feiyang Xiao, Xin Guo, Chen Jiang, Tan Pan, Xingmeng Zhang, Cenyu Liu, Zeyun Miao, Jianchao Ge, Xiansheng Wang, Qimeng Wang, Yichi Zhang, Wenbo Zhang, Fengping Zhu, Limei Han, Yuan Qi, Chensen Lin, Yuan Cheng
Abstract: Intracranial aneurysms (IAs) are serious cerebrovascular lesions found in approximately 5\% of the general population. Their rupture may lead to high mortality. Current methods for assessing IA risk focus on morphological and patient-specific factors, but the hemodynamic influences on IA development and rupture remain unclear. While accurate for hemodynamic studies, conventional computational fluid dynamics (CFD) methods are computationally intensive, hindering their deployment in large-scale or real-time clinical applications. To address this challenge, we curated a large-scale, high-fidelity aneurysm CFD dataset to facilitate the development of efficient machine learning algorithms for such applications. Based on 427 real aneurysm geometries, we synthesized 10,660 3D shapes via controlled deformation to simulate aneurysm evolution. The authenticity of these synthetic shapes was confirmed by neurosurgeons. CFD computations were performed on each shape under eight steady-state mass flow conditions, generating a total of 85,280 blood flow dynamics data covering key parameters. Furthermore, the dataset includes segmentation masks, which can support tasks that use images, point clouds or other multimodal data as input. Additionally, we introduced a benchmark for estimating flow parameters to assess current modeling methods. This dataset aims to advance aneurysm research and promote data-driven approaches in biofluids, biomedical engineering, and clinical risk assessment. The code and dataset are available at: https://github.com/Xigui-Li/Aneumo.
Authors: Pierre-Marc Jodoin, Manon Edde, Gabriel Girard, F\'elix Dumais, Guillaume Theaud, Matthieu Dumont, Jean-Christophe Houde, Yoan David, Maxime Descoteaux
Abstract: Over the years, ComBAT has become the standard method for harmonizing MRI-derived measurements, with its ability to compensate for site-related additive and multiplicative biases while preserving biological variability. However, ComBAT relies on a set of assumptions that, when violated, can result in flawed harmonization. In this paper, we thoroughly review ComBAT's mathematical foundation, outlining these assumptions, and exploring their implications for the demographic composition necessary for optimal results. Through a series of experiments involving a slightly modified version of ComBAT called Pairwise-ComBAT tailored for normative modeling applications, we assess the impact of various population characteristics, including population size, age distribution, the absence of certain covariates, and the magnitude of additive and multiplicative factors. Based on these experiments, we present five essential recommendations that should be carefully considered to enhance consistency and supporting reproducibility, two essential factors for open science, collaborative research, and real-life clinical deployment.
Authors: Subrata Biswas, Mohammad Nur Hossain Khan, Bashima Islam
Abstract: Spoken Language Understanding (SLU) systems must balance performance and efficiency, particularly in resource-constrained environments. Existing methods apply distillation and quantization separately, leading to suboptimal compression as distillation ignores quantization constraints. We propose QUADS, a unified framework that optimizes both through multi-stage training with a pre-tuned model, enhancing adaptability to low-bit regimes while maintaining accuracy. QUADS achieves 71.13\% accuracy on SLURP and 99.20\% on FSC, with only minor degradations of up to 5.56\% compared to state-of-the-art models. Additionally, it reduces computational complexity by 60--73$\times$ (GMACs) and model size by 83--700$\times$, demonstrating strong robustness under extreme quantization. These results establish QUADS as a highly efficient solution for real-world, resource-constrained SLU applications.
Authors: Xuejun Sun, Yiran Song, Xiaochen Zhou, Ruilie Cai, Yu Zhang, Xinyi Li, Rui Peng, Jialiu Xie, Yuanyuan Yan, Muyao Tang, Prem Lakshmanane, Baiming Zou, James S. Hagood, Raymond J. Pickles, Didong Li, Fei Zou, Xiaojing Zheng
Abstract: Respiratory viral infections pose a global health burden, yet the cellular immune responses driving protection or pathology remain unclear. Natural infection cohorts often lack pre-exposure baseline data and structured temporal sampling. In contrast, inoculation and vaccination trials generate insightful longitudinal transcriptomic data. However, the scattering of these datasets across platforms, along with inconsistent metadata and preprocessing procedure, hinders AI-driven discovery. To address these challenges, we developed the Human Respiratory Viral Immunization LongitudinAl Gene Expression (HR-VILAGE-3K3M) repository: an AI-ready, rigorously curated dataset that integrates 14,136 RNA-seq profiles from 3,178 subjects across 66 studies encompassing over 2.56 million cells. Spanning vaccination, inoculation, and mixed exposures, the dataset includes microarray, bulk RNA-seq, and single-cell RNA-seq from whole blood, PBMCs, and nasal swabs, sourced from GEO, ImmPort, and ArrayExpress. We harmonized subject-level metadata, standardized outcome measures, applied unified preprocessing pipelines with rigorous quality control, and aligned all data to official gene symbols. To demonstrate the utility of HR-VILAGE-3K3M, we performed predictive modeling of vaccine responders and evaluated batch-effect correction methods. Beyond these initial demonstrations, it supports diverse systems immunology applications and benchmarking of feature selection and transfer learning algorithms. Its scale and heterogeneity also make it ideal for pretraining foundation models of the human immune response and for advancing multimodal learning frameworks. As the largest longitudinal transcriptomic resource for human respiratory viral immunization, it provides an accessible platform for reproducible AI-driven research, accelerating systems immunology and vaccine development against emerging viral threats.
Authors: Ningning Yao, Huan Xi, Lang Chen, Zhe Song, Jian Li, Yulei Chen, Baocai Guo, Yuanhang Zhang, Tong Zhu, Pengfei Li, Daniel Rosenfeld, John H. Seinfeld, Shaocai Yu
Abstract: Despite policymakers deploying various tools to mitigate emissions of ozone (O\textsubscript{3}) precursors, such as nitrogen oxides (NO\textsubscript{x}), carbon monoxide (CO), and volatile organic compounds (VOCs), the effectiveness of policy combinations remains uncertain. We employ an integrated framework that couples structural break detection with machine learning to pinpoint effective interventions across the building, electricity, industrial, and transport sectors, identifying treatment effects as abrupt changes without prior assumptions about policy treatment assignment and timing. Applied to two decades of global O\textsubscript{3} precursor emissions data, we detect 78, 77, and 78 structural breaks for NO\textsubscript{x}, CO, and VOCs, corresponding to cumulative emission reductions of 0.96-0.97 Gt, 2.84-2.88 Gt, and 0.47-0.48 Gt, respectively. Sector-level analysis shows that electricity sector structural policies cut NO\textsubscript{x} by up to 32.4\%, while in buildings, developed countries combined adoption subsidies with carbon taxes to achieve 42.7\% CO reductions and developing countries used financing plus fuel taxes to secure 52.3\%. VOCs abatement peaked at 38.5\% when fossil-fuel subsidy reforms were paired with financial incentives. Finally, hybrid strategies merging non-price measures (subsidies, bans, mandates) with pricing instruments delivered up to an additional 10\% co-benefit. These findings guide the sequencing and complementarity of context-specific policy portfolios for O\textsubscript{3} precursor mitigation.
Authors: Janis Zenkner, Tobias Sesterhenn, Christian Bartelt
Abstract: Abstraction and reasoning in program synthesis has seen significant progress through both inductive and transductive paradigms. Inductive approaches generate a program or latent function from input-output examples, which can then be applied to new inputs. Transductive approaches directly predict output values for given inputs, effectively serving as the function themselves. Current approaches combine inductive and transductive models via isolated ensembling, but they do not explicitly model the interaction between both paradigms. In this work, we introduce \acs{tiips}, a novel framework that unifies transductive and inductive strategies by explicitly modeling their interactions through a cooperative mechanism: an inductive model generates programs, while a transductive model constrains, guides, and refines the search to improve synthesis accuracy and generalization. We evaluate \acs{tiips} on two widely studied program synthesis domains: string and list manipulation. Our results show that \acs{tiips} solves more tasks and yields functions that more closely match optimal solutions in syntax and semantics, particularly in out-of-distribution settings, yielding state-of-the-art performance. We believe that explicitly modeling the synergy between inductive and transductive reasoning opens promising avenues for general-purpose program synthesis and broader applications.
Authors: Fatemeh Chajaei, Hossein Bagheri
Abstract: Three-dimensional reconstruction of buildings, particularly at Level of Detail 1 (LOD1), plays a crucial role in various applications such as urban planning, urban environmental studies, and designing optimized transportation networks. This study focuses on assessing the potential of LiDAR data for accurate 3D building reconstruction at LOD1 and extracting morphological features from these models. Four deep semantic segmentation models, U-Net, Attention U-Net, U-Net3+, and DeepLabV3+, were used, applying transfer learning to extract building footprints from LiDAR data. The results showed that U-Net3+ and Attention U-Net outperformed the others, achieving IoU scores of 0.833 and 0.814, respectively. Various statistical measures, including maximum, range, mode, median, and the 90th percentile, were used to estimate building heights, resulting in the generation of 3D models at LOD1. As the main contribution of the research, the impact of segmentation accuracy on the quality of 3D building modeling and the accuracy of morphological features like building area and external wall surface area was investigated. The results showed that the accuracy of building identification (segmentation performance) significantly affects the 3D model quality and the estimation of morphological features, depending on the height calculation method. Overall, the UNet3+ method, utilizing the 90th percentile and median measures, leads to accurate height estimation of buildings and the extraction of morphological features.
Authors: Andrey Alexandrov, Giovanni Acampora, Giovanni De Lellis, Antonia Di Crescenzo, Chiara Errico, Daria Morozova, Valeri Tioukov, Autilia Vittiello
Abstract: Accurately tracking particles and determining their position along the optical axis is a major challenge in optical microscopy, especially when extremely high precision is needed. In this study, we introduce a deep learning approach using convolutional neural networks (CNNs) that can determine axial positions from dual-focal plane images without relying on predefined models. Our method achieves an axial localization accuracy of 40 nanometers - six times better than traditional single-focal plane techniques. The model's simple design and strong performance make it suitable for a wide range of uses, including dark matter detection, proton therapy for cancer, and radiation protection in space. It also shows promise in fields like biological imaging, materials science, and environmental monitoring. This work highlights how machine learning can turn complex image data into reliable, precise information, offering a flexible and powerful tool for many scientific applications.
Authors: Yan Wang, Ling Ding, Tien N Nguyen, Shaohua Wang, Yanan Zheng
Abstract: Large Language Models for code often entail significant computational complexity, which grows significantly with the length of the input code sequence. We propose LeanCode for code simplification to reduce training and prediction time, leveraging code contexts in utilizing attention scores to represent the tokens' importance. We advocate for the selective removal of tokens based on the average context-aware attention scores rather than average scores across all inputs. LeanCode uses the attention scores of `CLS' tokens within the encoder for classification tasks, such as code search. It also employs the encoder-decoder attention scores to determine token significance for sequence-to-sequence tasks like code summarization.Our evaluation shows LeanCode's superiority over the SOTAs DietCode and Slimcode, with improvements of 60% and 16% for code search, and 29% and 27% for code summarization, respectively.
Authors: Richard Khoury, Maxence Verhaverbeke, Julie A. Gramaccia
Abstract: The 5W1H questions -- who, what, when, where, why and how -- are commonly used in journalism to ensure that an article describes events clearly and systematically. Answering them is a crucial prerequisites for tasks such as summarization, clustering, and news aggregation. In this paper, we design the first automated extraction pipeline to get 5W1H information from French news articles. To evaluate the performance of our algo- rithm, we also create a corpus of 250 Quebec news articles with 5W1H answers marked by four human annotators. Our results demonstrate that our pipeline performs as well in this task as the large language model GPT-4o.
Authors: Minglu Zhao, Dehong Xu, Deqian Kong, Wen-Hao Zhang, Ying Nian Wu
Abstract: The hippocampus orchestrates spatial navigation through collective place cell encodings that form cognitive maps. We reconceptualize the population of place cells as position embeddings approximating multi-scale symmetric random walk transition kernels: the inner product $\langle h(x, t), h(y, t) \rangle = q(y|x, t)$ represents normalized transition probabilities, where $h(x, t)$ is the embedding at location $ x $, and $q(y|x, t)$ is the normalized symmetric transition probability over time $t$. The time parameter $\sqrt{t}$ defines a spatial scale hierarchy, mirroring the hippocampal dorsoventral axis. $q(y|x, t)$ defines spatial adjacency between $x$ and $y$ at scale or resolution $\sqrt{t}$, and the pairwise adjacency relationships $(q(y|x, t), \forall x, y)$ are reduced into individual embeddings $(h(x, t), \forall x)$ that collectively form a map of the environment at sale $\sqrt{t}$. Our framework employs gradient ascent on $q(y|x, t) = \langle h(x, t), h(y, t)\rangle$ with adaptive scale selection, choosing the time scale with maximal gradient at each step for trap-free, smooth trajectories. Efficient matrix squaring $P_{2t} = P_t^2$ builds global representations from local transitions $P_1$ without memorizing past trajectories, enabling hippocampal preplay-like path planning. This produces robust navigation through complex environments, aligning with hippocampal navigation. Experimental results show that our model captures place cell properties -- field size distribution, adaptability, and remapping -- while achieving computational efficiency. By modeling collective transition probabilities rather than individual place fields, we offer a biologically plausible, scalable framework for spatial navigation.
Authors: Soo Min Kwon, Alec S. Xu, Can Yaras, Laura Balzano, Qing Qu
Abstract: This work aims to demystify the out-of-distribution (OOD) capabilities of in-context learning (ICL) by studying linear regression tasks parameterized with low-rank covariance matrices. With such a parameterization, we can model distribution shifts as a varying angle between the subspace of the training and testing covariance matrices. We prove that a single-layer linear attention model incurs a test risk with a non-negligible dependence on the angle, illustrating that ICL is not robust to such distribution shifts. However, using this framework, we also prove an interesting property of ICL: when trained on task vectors drawn from a union of low-dimensional subspaces, ICL can generalize to any subspace within their span, given sufficiently long prompt lengths. This suggests that the OOD generalization ability of Transformers may actually stem from the new task lying within the span of those encountered during training. We empirically show that our results also hold for models such as GPT-2, and conclude with (i) experiments on how our observations extend to nonlinear function classes and (ii) results on how LoRA has the ability to capture distribution shifts.
Authors: Xi Wang, Jiaqian Hu, Safinah Ali
Abstract: We present MAATS, a Multi Agent Automated Translation System that leverages the Multidimensional Quality Metrics (MQM) framework as a fine-grained signal for error detection and refinement. MAATS employs multiple specialized AI agents, each focused on a distinct MQM category (e.g., Accuracy, Fluency, Style, Terminology), followed by a synthesis agent that integrates the annotations to iteratively refine translations. This design contrasts with conventional single-agent methods that rely on self-correction. Evaluated across diverse language pairs and Large Language Models (LLMs), MAATS outperforms zero-shot and single-agent baselines with statistically significant gains in both automatic metrics and human assessments. It excels particularly in semantic accuracy, locale adaptation, and linguistically distant language pairs. Qualitative analysis highlights its strengths in multi-layered error diagnosis, omission detection across perspectives, and context-aware refinement. By aligning modular agent roles with interpretable MQM dimensions, MAATS narrows the gap between black-box LLMs and human translation workflows, shifting focus from surface fluency to deeper semantic and contextual fidelity.
Authors: Drishya Karki, Michiel Kamphuis, Angelecia Frey
Abstract: EasyMath is a compact benchmark for practical math reasoning in small language models. It covers thirteen categories, from basic arithmetic and order of operations to word problems, algebraic expressions, edge cases, and omits specialist topics. We tested 23 models (14M to 4B parameters) using exact, numerical, and symbolic checks on free-form answers in a zero-shot setting. Accuracy rises with size and training, chain-of-thought adds modest gains, and consistency improves at scale.
Authors: So Won Jeong, Claire Donnat
Abstract: Graph Neural Networks (GNNs) are increasingly used in conjunction with unsupervised learning techniques to learn powerful node representations, but their deployment is hindered by their high sensitivity to hyperparameter tuning and the absence of established methodologies for selecting the optimal models. To address these challenges, we propose LOBSTUR-GNN ({\bf Lo}cal {\bf B}oot{\bf s}trap for {\bf T}uning {\bf U}nsupervised {\bf R}epresentations in GNNs) i), a novel framework designed to adapt bootstrapping techniques for unsupervised graph representation learning. LOBSTUR-GNN tackles two main challenges: (a) adapting the bootstrap edge and feature resampling process to account for local graph dependencies in creating alternative versions of the same graph, and (b) establishing robust metrics for evaluating learned representations without ground-truth labels. Using locally bootstrapped resampling and leveraging Canonical Correlation Analysis (CCA) to assess embedding consistency, LOBSTUR provides a principled approach for hyperparameter tuning in unsupervised GNNs. We validate the effectiveness and efficiency of our proposed method through extensive experiments on established academic datasets, showing an 65.9\% improvement in the classification accuracy compared to an uninformed selection of hyperparameters. Finally, we deploy our framework on a real-world application, thereby demonstrating its validity and practical utility in various settings. \footnote{The code is available at \href{https://github.com/sowonjeong/lobstur-graph-bootstrap}{github.com/sowonjeong/lobstur-graph-bootstrap}.}
Authors: Ryan Solgi, Kai Zhen, Rupak Vignesh Swaminathan, Nathan Susanj, Athanasios Mouchtaris, Siegfried Kunzmann, Zheng Zhang
Abstract: The efficient implementation of large language models (LLMs) is crucial for deployment on resource-constrained devices. Low-rank tensor compression techniques, such as tensor-train (TT) networks, have been widely studied for over-parameterized neural networks. However, their applications to compress pre-trained large language models (LLMs) for downstream tasks (post-training) remains challenging due to the high-rank nature of pre-trained LLMs and the lack of access to pretraining data. In this study, we investigate low-rank tensorized LLMs during fine-tuning and propose sparse augmented tensor networks (Saten) to enhance their performance. The proposed Saten framework enables full model compression. Experimental results demonstrate that Saten enhances both accuracy and compression efficiency in tensorized language models, achieving state-of-the-art performance.
Authors: Fadel M. Megahed, Ying-Ju Chen, L. Allision Jones-Farmer, Younghwa Lee, Jiawei Brooke Wang, Inez M. Zwetsloot
Abstract: This study introduces a framework for evaluating consistency in large language model (LLM) binary text classification, addressing the lack of established reliability assessment methods. Adapting psychometric principles, we determine sample size requirements, develop metrics for invalid responses, and evaluate intra- and inter-rater reliability. Our case study examines financial news sentiment classification across 14 LLMs (including claude-3-7-sonnet, gpt-4o, deepseek-r1, gemma3, llama3.2, phi4, and command-r-plus), with five replicates per model on 1,350 articles. Models demonstrated high intra-rater consistency, achieving perfect agreement on 90-98% of examples, with minimal differences between expensive and economical models from the same families. When validated against StockNewsAPI labels, models achieved strong performance (accuracy 0.76-0.88), with smaller models like gemma3:1B, llama3.2:3B, and claude-3-5-haiku outperforming larger counterparts. All models performed at chance when predicting actual market movements, indicating task constraints rather than model limitations. Our framework provides systematic guidance for LLM selection, sample size planning, and reliability assessment, enabling organizations to optimize resources for classification tasks.
Authors: Shashwat Khandelwal, Shreejith Shanker
Abstract: Recent research has highlighted the vulnerability of in-vehicle network protocols such as controller area networks (CAN) and proposed machine learning-based intrusion detection systems (IDSs) as an effective mitigation technique. However, their efficient integration into vehicular architecture is non-trivial, with existing methods relying on electronic control units (ECUs)-coupled IDS accelerators or dedicated ECUs as IDS accelerators. Here, initiating IDS requires complete reception of a CAN message from the controller, incurring data movement and software overheads. In this paper, we present SecCAN, a novel CAN controller architecture that embeds IDS capability within the datapath of the controller. This integration allows IDS to tap messages directly from within the CAN controller as they are received from the bus, removing overheads incurred by existing ML-based IDSs. A custom-quantised machine-learning accelerator is developed as the IDS engine and embedded into SecCAN's receive data path, with optimisations to overlap the IDS inference with the protocol's reception window. We implement SecCAN on AMD XCZU7EV FPGA to quantify its performance and benefits in hardware, using multiple attack datasets. We show that SecCAN can completely hide the IDS latency within the CAN reception window for all CAN packet sizes and detect multiple attacks with state-of-the-art accuracy with zero software overheads on the ECU and low energy overhead (73.7 uJ per message) for IDS inference. Also, SecCAN incurs limited resource overhead compared to a standard CAN controller (< 30% LUT, < 1% FF), making it ideally suited for automotive deployment.
Authors: Sil Hamilton, Rebecca M. M. Hicke, Matthew Wilkens, David Mimno
Abstract: Although the context length of large language models (LLMs) has increased to millions of tokens, evaluating their effectiveness beyond needle-in-a-haystack approaches has proven difficult. We argue that novels provide a case study of subtle, complicated structure and long-range semantic dependencies often over 128k tokens in length. Inspired by work on computational novel analysis, we release the Too Long, Didn't Model (TLDM) benchmark, which tests a model's ability to report plot summary, storyworld configuration, and elapsed narrative time. We find that none of seven tested frontier LLMs retain stable understanding beyond 64k tokens. Our results suggest language model developers must look beyond "lost in the middle" benchmarks when evaluating model performance in complex long-context scenarios. To aid in further development we release the TLDM benchmark together with reference code and data.
Authors: Hao Tang, Kevin Ellis, Suhas Lohit, Michael J. Jones, Moitreya Chatterjee
Abstract: The task of estimating the world model describing the dynamics of a real world process assumes immense importance for anticipating and preparing for future outcomes. For applications such as video surveillance, robotics applications, autonomous driving, etc. this objective entails synthesizing plausible visual futures, given a few frames of a video to set the visual context. Towards this end, we propose ProgGen, which undertakes the task of video frame prediction by representing the dynamics of the video using a set of neuro-symbolic, human-interpretable set of states (one per frame) by leveraging the inductive biases of Large (Vision) Language Models (LLM/VLM). In particular, ProgGen utilizes LLM/VLM to synthesize programs: (i) to estimate the states of the video, given the visual context (i.e. the frames); (ii) to predict the states corresponding to future time steps by estimating the transition dynamics; (iii) to render the predicted states as visual RGB-frames. Empirical evaluations reveal that our proposed method outperforms competing techniques at the task of video frame prediction in two challenging environments: (i) PhyWorld (ii) Cart Pole. Additionally, ProgGen permits counter-factual reasoning and interpretable video generation attesting to its effectiveness and generalizability for video generation tasks.
Authors: Xiaoyin Chen, Jiarui Lu, Minsu Kim, Dinghuai Zhang, Jian Tang, Alexandre Pich\'e, Nicolas Gontier, Yoshua Bengio, Ehsan Kamalloo
Abstract: Reinforcement learning (RL) has proven effective for fine-tuning large language models (LLMs), significantly enhancing their reasoning abilities in domains such as mathematics and code generation. A crucial factor influencing RL fine-tuning success is the training curriculum: the order in which training problems are presented. While random curricula serve as common baselines, they remain suboptimal; manually designed curricula often rely heavily on heuristics, and online filtering methods can be computationally prohibitive. To address these limitations, we propose Self-Evolving Curriculum (SEC), an automatic curriculum learning method that learns a curriculum policy concurrently with the RL fine-tuning process. Our approach formulates curriculum selection as a non-stationary Multi-Armed Bandit problem, treating each problem category (e.g., difficulty level or problem type) as an individual arm. We leverage the absolute advantage from policy gradient methods as a proxy measure for immediate learning gain. At each training step, the curriculum policy selects categories to maximize this reward signal and is updated using the TD(0) method. Across three distinct reasoning domains: planning, inductive reasoning, and mathematics, our experiments demonstrate that SEC significantly improves models' reasoning capabilities, enabling better generalization to harder, out-of-distribution test problems. Additionally, our approach achieves better skill balance when fine-tuning simultaneously on multiple reasoning domains. These findings highlight SEC as a promising strategy for RL fine-tuning of LLMs.
Authors: Ghasem Pasandi, Kishor Kunal, Varun Tej, Kunjal Shan, Hanfei Sun, Sumit Jain, Chunhui Li, Chenhui Deng, Teodor-Dumitru Ene, Haoxing Ren, Sreedhar Pratty
Abstract: This paper presents JARVIS, a novel multi-agent framework that leverages Large Language Models (LLMs) and domain expertise to generate high-quality scripts for specialized Electronic Design Automation (EDA) tasks. By combining a domain-specific LLM trained with synthetically generated data, a custom compiler for structural verification, rule enforcement, code fixing capabilities, and advanced retrieval mechanisms, our approach achieves significant improvements over state-of-the-art domain-specific models. Our framework addresses the challenges of data scarcity and hallucination errors in LLMs, demonstrating the potential of LLMs in specialized engineering domains. We evaluate our framework on multiple benchmarks and show that it outperforms existing models in terms of accuracy and reliability. Our work sets a new precedent for the application of LLMs in EDA and paves the way for future innovations in this field.
Authors: Meenal Parakh, Alexandre Kirchmeyer, Beining Han, Jia Deng
Abstract: Generalizing control policies to novel embodiments remains a fundamental challenge in enabling scalable and transferable learning in robotics. While prior works have explored this in locomotion, a systematic study in the context of manipulation tasks remains limited, partly due to the lack of standardized benchmarks. In this paper, we introduce a benchmark for learning cross-embodiment manipulation, focusing on two foundational tasks-reach and push-across a diverse range of morphologies. The benchmark is designed to test generalization along three axes: interpolation (testing performance within a robot category that shares the same link structure), extrapolation (testing on a robot with a different link structure), and composition (testing on combinations of link structures). On the benchmark, we evaluate the ability of different RL policies to learn from multiple morphologies and to generalize to novel ones. Our study aims to answer whether morphology-aware training can outperform single-embodiment baselines, whether zero-shot generalization to unseen morphologies is feasible, and how consistently these patterns hold across different generalization regimes. The results highlight the current limitations of multi-embodiment learning and provide insights into how architectural and training design choices influence policy generalization.
Authors: Zixuan Ke, Austin Xu, Yifei Ming, Xuan-Phi Nguyen, Caiming Xiong, Shafiq Joty
Abstract: Multi-agent systems (MAS) leveraging the impressive capabilities of Large Language Models (LLMs) hold significant potential for tackling complex tasks. However, most current MAS depend on manually designed agent roles and communication protocols. These manual designs often fail to align with the underlying LLMs' strengths and struggle to adapt to novel tasks. Recent automatic MAS approaches attempt to mitigate these limitations but typically necessitate a validation-set for tuning and yield static MAS designs lacking adaptability during inference. We introduce SELF-MAS, the first self-supervised, inference-time only framework for automatic MAS design. SELF-MAS employs meta-level design to iteratively generate, evaluate, and refine MAS configurations tailored to each problem instance, without requiring a validation set. Critically, it enables dynamic agent composition and problem decomposition through meta-feedback on solvability and completeness. Experiments across math, graduate-level QA, and software engineering benchmarks, using both closed-source and open-source LLM back-bones of varying sizes, demonstrate that SELF-MAS outperforms both manual and automatic MAS baselines, achieving a 7.44% average accuracy improvement over the next strongest baseline while maintaining cost-efficiency. These findings underscore the promise of meta-level self-supervised design for creating effective and adaptive MAS.
Authors: Fernando Lucatelli Nunes, Gordon Plotkin, Matthijs V\'ak\'ar
Abstract: Combinatory Homomorphic Automatic Differentiation (CHAD) was originally formulated as a semantics-driven source transformation for reverse-mode AD in total programming languages. We extend this framework to partial languages with features such as potentially non-terminating operations, real-valued conditionals, and iteration constructs like while-loops, while preserving CHAD's structure-preserving semantics principle. A key contribution is the introduction of iteration-extensive indexed categories, which allow iteration in the base category to lift to parameterized initial algebras in the indexed category. This enables iteration to be interpreted in the Grothendieck construction of the target language in a principled way. The resulting fibred iterative structure cleanly models iteration in the categorical semantics. Consequently, the extended CHAD transformation remains the unique structure-preserving functor (an iterative Freyd category morphism) from the freely generated iterative Freyd category of the source language to the Grothendieck construction of the target's syntactic semantics, mapping each primitive operation to its derivative. We prove the correctness of this transformation using the universal property of the source language's syntax, showing that the transformed programs compute correct reverse-mode derivatives. Our development also contributes to understanding iteration constructs within dependently typed languages and categories of containers. As our primary motivation and application, we generalize CHAD to languages with data types, partial features, and iteration, providing the first rigorous categorical semantics for reverse-mode CHAD in such settings and formally guaranteeing the correctness of the source-to-source CHAD technique.
Authors: Anupama Sridhar, Alexander Johansen
Abstract: First-order adaptive optimization methods like Adam are the default choices for training modern deep neural networks. Despite their empirical success, the theoretical understanding of these methods in non-smooth settings, particularly in Deep ReLU networks, remains limited. ReLU activations create exponentially many region boundaries where standard smoothness assumptions break down. \textbf{We derive the first \(\tilde{O}\!\bigl(\sqrt{d_{\mathrm{eff}}/n}\bigr)\) generalization bound for Adam in Deep ReLU networks and the first global-optimal convergence for Adam in the non smooth, non convex relu landscape without a global PL or convexity assumption.} Our analysis is based on stratified Morse theory and novel results in Kakeya sets. We develop a multi-layer refinement framework that progressively tightens bounds on region crossings. We prove that the number of region crossings collapses from exponential to near-linear in the effective dimension. Using a Kakeya based method, we give a tighter generalization bound than PAC-Bayes approaches and showcase convergence using a mild uniform low barrier assumption.
Authors: Ya-Yun Huang, Joseph McClernon, Jason A. Oliver, Matthew M. Engelhard
Abstract: Daily environments have profound influence on our health and behavior. Recent work has shown that digital envirotyping, where computer vision is applied to images of daily environments taken during ecological momentary assessment (EMA), can be used to identify meaningful relationships between environmental features and health outcomes of interest. To systematically study such effects on an individual level, it is helpful to group images into distinct environments encountered in an individual's daily life; these may then be analyzed, further grouped into related environments with similar features, and linked to health outcomes. Here we introduce infinite hierarchical contrastive clustering to address this challenge. Building on the established contrastive clustering framework, our method a) allows an arbitrary number of clusters without requiring the full Dirichlet Process machinery by placing a stick-breaking prior on predicted cluster probabilities; and b) encourages distinct environments to form well-defined sub-clusters within each cluster of related environments by incorporating a participant-specific prediction loss. Our experiments show that our model effectively identifies distinct personal environments and groups these environments into meaningful environment types. We then illustrate how the resulting clusters can be linked to various health outcomes, highlighting the potential of our approach to advance the envirotyping paradigm.
Authors: Ze Wang, Jingang Qu, Zhenyu Gao, Pascal Morin
Abstract: This work demonstrates an airflow inertial based odometry system with multi-sensor data fusion, including thermal anemometer, IMU, ESC, and barometer. This goal is challenging because low-cost IMUs and barometers have significant bias, and anemometer measurements are very susceptible to interference from spinning propellers and ground effects. We employ a GRU-based deep neural network to estimate relative air speed from noisy and disturbed anemometer measurements, and an observer with bias model to fuse the sensor data and thus estimate the state of aerial vehicle. A complete flight data, including takeoff and landing on the ground, shows that the approach is able to decouple the downwash induced wind speed caused by propellers and the ground effect, and accurately estimate the flight speed in a wind-free indoor environment. IMU, and barometer bias are effectively estimated, which significantly reduces the position integration drift, which is only 5.7m for 203s manual random flight. The open source is available on https://github.com/SyRoCo-ISIR/Flight-Speed-Estimation-Airflow.
URLs: https://github.com/SyRoCo-ISIR/Flight-Speed-Estimation-Airflow.
Authors: Feiyang Cai, Jiahui Bai, Tao Tang, Joshua Luo, Tianyu Zhu, Ling Liu, Feng Luo
Abstract: Precise recognition, editing, and generation of molecules are essential prerequisites for both chemists and AI systems tackling various chemical tasks. We present MolLangBench, a comprehensive benchmark designed to evaluate fundamental molecule-language interface tasks: language-prompted molecular structure recognition, editing, and generation. To ensure high-quality, unambiguous, and deterministic outputs, we construct the recognition tasks using automated cheminformatics tools, and curate editing and generation tasks through rigorous expert annotation and validation. MolLangBench supports the evaluation of models that interface language with different molecular representations, including linear strings, molecular images, and molecular graphs. Evaluations of state-of-the-art models reveal significant limitations: the strongest model (o3) achieves $79.2\%$ and $78.5\%$ accuracy on recognition and editing tasks, which are intuitively simple for humans, and performs even worse on the generation task, reaching only $29.0\%$ accuracy. These results highlight the shortcomings of current AI systems in handling even preliminary molecular recognition and manipulation tasks. We hope MolLangBench will catalyze further research toward more effective and reliable AI systems for chemical applications.
Authors: Cheng Qian, Hongyi Du, Hongru Wang, Xiusi Chen, Yuji Zhang, Avirup Sil, Chengxiang Zhai, Kathleen McKeown, Heng Ji
Abstract: Recent progress in large language models (LLMs) has enabled substantial advances in solving mathematical problems. However, existing benchmarks often fail to reflect the complexity of real-world problems, which demand open-ended, interdisciplinary reasoning and integration of computational tools. To address this gap, we introduce ModelingBench, a novel benchmark featuring real-world-inspired, open-ended problems from math modeling competitions across diverse domains, ranging from urban traffic optimization to ecosystem resource planning. These tasks require translating natural language into formal mathematical formulations, applying appropriate tools, and producing structured, defensible reports. ModelingBench also supports multiple valid solutions, capturing the ambiguity and creativity of practical modeling. We also present ModelingAgent, a multi-agent framework that coordinates tool use, supports structured workflows, and enables iterative self-refinement to generate well-grounded, creative solutions. To evaluate outputs, we further propose ModelingJudge, an expert-in-the-loop system leveraging LLMs as domain-specialized judges assessing solutions from multiple expert perspectives. Empirical results show that ModelingAgent substantially outperforms strong baselines and often produces solutions indistinguishable from those of human experts. Together, our work provides a comprehensive framework for evaluating and advancing real-world problem-solving in open-ended, interdisciplinary modeling challenges.
Authors: Yuhang Zhou, Jing Zhu, Shengyi Qian, Zhuokai Zhao, Xiyao Wang, Xiaoyu Liu, Ming Li, Paiheng Xu, Wei Ai, Furong Huang
Abstract: Large Language Models (LLMs) are increasingly aligned with human preferences through Reinforcement Learning from Human Feedback (RLHF). Among RLHF methods, Group Relative Policy Optimization (GRPO) has gained attention for its simplicity and strong performance, notably eliminating the need for a learned value function. However, GRPO implicitly assumes a balanced domain distribution and uniform semantic alignment across groups - assumptions that rarely hold in real-world datasets. When applied to multi-domain, imbalanced data, GRPO disproportionately optimizes for dominant domains, neglecting underrepresented ones and resulting in poor generalization and fairness. We propose Domain-Informed Self-Consistency Policy Optimization (DISCO), a principled extension to GRPO that addresses inter-group imbalance with two key innovations. Domain-aware reward scaling counteracts frequency bias by reweighting optimization based on domain prevalence. Difficulty-aware reward scaling leverages prompt-level self-consistency to identify and prioritize uncertain prompts that offer greater learning value. Together, these strategies promote more equitable and effective policy learning across domains. Extensive experiments across multiple LLMs and skewed training distributions show that DISCO improves generalization, outperforms existing GRPO variants by 5% on Qwen3 models, and sets new state-of-the-art results on multi-domain alignment benchmarks.
Authors: Hao Wang, Pinzhi Huang, Jihan Yang, Saining Xie, Daisuke Kawahara
Abstract: The rapid evolution of multimodal large language models (MLLMs) has significantly enhanced their real-world applications. However, achieving consistent performance across languages, especially when integrating cultural knowledge, remains a significant challenge. To better assess this issue, we introduce two new benchmarks: KnowRecall and VisRecall, which evaluate cross-lingual consistency in MLLMs. KnowRecall is a visual question answering benchmark designed to measure factual knowledge consistency in 15 languages, focusing on cultural and historical questions about global landmarks. VisRecall assesses visual memory consistency by asking models to describe landmark appearances in 9 languages without access to images. Experimental results reveal that state-of-the-art MLLMs, including proprietary ones, still struggle to achieve cross-lingual consistency. This underscores the need for more robust approaches that produce truly multilingual and culturally aware models.
Authors: Alessandro dos Santos Ferreira, Ana Paula Marques Ramos, Jos\'e Marcato Junior, Wesley Nunes Gon\c{c}alves
Abstract: Urban forests play a key role in enhancing environmental quality and supporting biodiversity in cities. Mapping and monitoring these green spaces are crucial for urban planning and conservation, yet accurately detecting trees is challenging due to complex landscapes and the variability in image resolution caused by different satellite sensors or UAV flight altitudes. While deep learning architectures have shown promise in addressing these challenges, their effectiveness remains strongly dependent on the availability of large and manually labeled datasets, which are often expensive and difficult to obtain in sufficient quantity. In this work, we propose a novel pipeline that integrates domain adaptation with GANs and Diffusion models to enhance the quality of low-resolution aerial images. Our proposed pipeline enhances low-resolution imagery while preserving semantic content, enabling effective tree segmentation without requiring large volumes of manually annotated data. Leveraging models such as pix2pix, Real-ESRGAN, Latent Diffusion, and Stable Diffusion, we generate realistic and structurally consistent synthetic samples that expand the training dataset and unify scale across domains. This approach not only improves the robustness of segmentation models across different acquisition conditions but also provides a scalable and replicable solution for remote sensing scenarios with scarce annotation resources. Experimental results demonstrated an improvement of over 50% in IoU for low-resolution images, highlighting the effectiveness of our method compared to traditional pipelines.
Authors: Sona Elza Simon, Preethi Jyothi
Abstract: Effective cross-lingual transfer remains a critical challenge in scaling the benefits of large language models from high-resource to low-resource languages. Towards this goal, prior studies have explored many approaches to combine task knowledge from task-specific data in a (high-resource) source language and language knowledge from unlabeled text in a (low-resource) target language. One notable approach proposed composable sparse fine-tuning (SFT) for cross-lingual transfer that learns task-specific and language-specific sparse masks to select a subset of the pretrained model's parameters that are further fine-tuned. These sparse fine-tuned vectors (SFTs) are subsequently composed with the pretrained model to facilitate zero-shot cross-lingual transfer to a task in a target language, using only task-specific data from a source language. These sparse masks for SFTs were identified using a simple magnitude-based pruning. In our work, we introduce DeFT-X, a novel composable SFT approach that denoises the weight matrices of a pretrained model before magnitude pruning using singular value decomposition, thus yielding more robust SFTs. We evaluate DeFT-X on a diverse set of extremely low-resource languages for sentiment classification (NusaX) and natural language inference (AmericasNLI) and demonstrate that it performs at par or outperforms SFT and other prominent cross-lingual transfer baselines.
Authors: Jason Yang, Wenda Chu, Daniel Khalil, Raul Astudillo, Bruce J. Wittmann, Frances H. Arnold, Yisong Yue
Abstract: Protein fitness optimization involves finding a protein sequence that maximizes desired quantitative properties in a combinatorially large design space of possible sequences. Recent developments in steering protein generative models (e.g diffusion models, language models) offer a promising approach. However, by and large, past studies have optimized surrogate rewards and/or utilized large amounts of labeled data for steering, making it unclear how well existing methods perform and compare to each other in real-world optimization campaigns where fitness is measured by low-throughput wet-lab assays. In this study, we explore fitness optimization using small amounts (hundreds) of labeled sequence-fitness pairs and comprehensively evaluate strategies such as classifier guidance and posterior sampling for guiding generation from different discrete diffusion models of protein sequences. We also demonstrate how guidance can be integrated into adaptive sequence selection akin to Thompson sampling in Bayesian optimization, showing that plug-and-play guidance strategies offer advantages compared to alternatives such as reinforcement learning with protein language models.
Authors: Muniba Noreen (Faculty of Electrical,Electronics Engineering, University of Engineering,Technology Taxila, Pakistan), Furqan Shaukat (Faculty of Electrical,Electronics Engineering, University of Engineering,Technology Taxila, Pakistan)
Abstract: Lung cancer remains among the deadliest types of cancer in recent decades, and early lung nodule detection is crucial for improving patient outcomes. The limited availability of annotated medical imaging data remains a bottleneck in developing accurate computer-aided diagnosis (CAD) systems. Self-supervised learning can help leverage large amounts of unlabeled data to develop more robust CAD systems. With the recent advent of transformer-based architecture and their ability to generalize to unseen tasks, there has been an effort within the healthcare community to adapt them to various medical downstream tasks. Thus, we propose a novel "LungNodule-SSM" method, which utilizes selfsupervised learning with DINOv2 as a backbone to enhance lung nodule detection and classification without annotated data. Our methodology has two stages: firstly, the DINOv2 model is pre-trained on unlabeled CT scans to learn robust feature representations, then secondly, these features are fine-tuned using transformer-based architectures for lesionlevel detection and accurate lung nodule diagnosis. The proposed method has been evaluated on the challenging LUNA 16 dataset, consisting of 888 CT scans, and compared with SOTA methods. Our experimental results show the superiority of our proposed method with an accuracy of 98.37%, explaining its effectiveness in lung nodule detection. The source code, datasets, and pre-processed data can be accessed using the link:https://github.com/EMeRALDsNRPU/Lung-Nodule-SSM-Self-Supervised-Lung-Nodule-Detection-and-Classification/tree/main
Authors: Haiduo Huang, Jiangcheng Song, Yadong Zhang, Pengju Ren
Abstract: Recent advances in knowledge distillation have emphasized the importance of decoupling different knowledge components. While existing methods utilize momentum mechanisms to separate task-oriented and distillation gradients, they overlook the inherent conflict between target-class and non-target-class knowledge flows. Furthermore, low-confidence dark knowledge in non-target classes introduces noisy signals that hinder effective knowledge transfer. To address these limitations, we propose DeepKD, a novel training framework that integrates dual-level decoupling with adaptive denoising. First, through theoretical analysis of gradient signal-to-noise ratio (GSNR) characteristics in task-oriented and non-task-oriented knowledge distillation, we design independent momentum updaters for each component to prevent mutual interference. We observe that the optimal momentum coefficients for task-oriented gradient (TOG), target-class gradient (TCG), and non-target-class gradient (NCG) should be positively related to their GSNR. Second, we introduce a dynamic top-k mask (DTM) mechanism that gradually increases K from a small initial value to incorporate more non-target classes as training progresses, following curriculum learning principles. The DTM jointly filters low-confidence logits from both teacher and student models, effectively purifying dark knowledge during early training. Extensive experiments on CIFAR-100, ImageNet, and MS-COCO demonstrate DeepKD's effectiveness. Our code is available at https://github.com/haiduo/DeepKD.
Authors: Yuante Li, Xu Yang, Xiao Yang, Minrui Xu, Xisen Wang, Weiqing Liu, Jiang Bian
Abstract: Financial markets pose fundamental challenges for asset return prediction due to their high dimensionality, non-stationarity, and persistent volatility. Despite advances in large language models and multi-agent systems, current quantitative research pipelines suffer from limited automation, weak interpretability, and fragmented coordination across key components such as factor mining and model innovation. In this paper, we propose R&D-Agent for Quantitative Finance, in short RD-Agent(Q), the first data-centric multi-agent framework designed to automate the full-stack research and development of quantitative strategies via coordinated factor-model co-optimization. RD-Agent(Q) decomposes the quant process into two iterative stages: a Research stage that dynamically sets goal-aligned prompts, formulates hypotheses based on domain priors, and maps them to concrete tasks, and a Development stage that employs a code-generation agent, Co-STEER, to implement task-specific code, which is then executed in real-market backtests. The two stages are connected through a feedback stage that thoroughly evaluates experimental outcomes and informs subsequent iterations, with a multi-armed bandit scheduler for adaptive direction selection. Empirically, RD-Agent(Q) achieves up to 2X higher annualized returns than classical factor libraries using 70% fewer factors, and outperforms state-of-the-art deep time-series models on real markets. Its joint factor-model optimization delivers a strong balance between predictive accuracy and strategy robustness. Our code is available at: https://github.com/microsoft/RD-Agent.
Authors: Mohit Sharma, Adam Fishman, Vikash Kumar, Chris Paxton, Oliver Kroemer
Abstract: Robots in the real world need to perceive and move to goals in complex environments without collisions. Avoiding collisions is especially difficult when relying on sensor perception and when goals are among clutter. Diffusion policies and other generative models have shown strong performance in solving local planning problems, but often struggle at avoiding all of the subtle constraint violations that characterize truly challenging global motion planning problems. In this work, we propose an approach for learning global motion planning using diffusion policies, allowing the robot to generate full trajectories through complex scenes and reasoning about multiple obstacles along the path. Our approach uses cascaded hierarchical models which unify global prediction and local refinement together with online plan repair to ensure the trajectories are collision free. Our method outperforms (by ~5%) a wide variety of baselines on challenging tasks in multiple domains including navigation and manipulation.
Authors: Diego Granziol, Donald Flynn
Abstract: We investigate the theoretical foundations of data poisoning attacks in machine learning models. Our analysis reveals that the Hessian with respect to the input serves as a diagnostic tool for detecting poisoning, exhibiting spectral signatures that characterize compromised datasets. We use random matrix theory (RMT) to develop a theory for the impact of poisoning proportion and regularisation on attack efficacy in linear regression. Through QR stepwise regression, we study the spectral signatures of the Hessian in multi-output regression. We perform experiments on deep networks to show experimentally that this theory extends to modern convolutional and transformer networks under the cross-entropy loss. Based on these insights we develop preliminary algorithms to determine if a network has been poisoned and remedies which do not require further training.
Authors: Jeonghye Kim, Sojeong Rhee, Minbeom Kim, Dohyung Kim, Sangmook Lee, Youngchul Sung, Kyomin Jung
Abstract: Recent advances in LLM agents have largely built on reasoning backbones like ReAct, which interleave thought and action in complex environments. However, ReAct often produces ungrounded or incoherent reasoning steps, leading to misalignment between the agent's actual state and goal. Our analysis finds that this stems from ReAct's inability to maintain consistent internal beliefs and goal alignment, causing compounding errors and hallucinations. To address this, we introduce ReflAct, a novel backbone that shifts reasoning from merely planning next actions to continuously reflecting on the agent's state relative to its goal. By explicitly grounding decisions in states and enforcing ongoing goal alignment, ReflAct dramatically improves strategic reliability. This design delivers substantial empirical gains: ReflAct surpasses ReAct by 27.7% on average, achieving a 93.3% success rate in ALFWorld. Notably, ReflAct even outperforms ReAct with added enhancement modules (e.g., Reflexion, WKM), showing that strengthening the core reasoning backbone is key to reliable agent performance.
Authors: Rina Tazaki, Tomoyuki Akiyama, Akira Furui
Abstract: Automated epileptic seizure detection from electroencephalogram (EEG) remains challenging due to significant individual differences in EEG patterns across patients. While existing studies achieve high accuracy with patient-specific approaches, they face difficulties in generalizing to new patients. To address this, we propose a detection framework combining domain adversarial training with a convolutional neural network (CNN) and a bidirectional long short-term memory (BiLSTM). First, the CNN extracts local patient-invariant features through domain adversarial training, which optimizes seizure detection accuracy while minimizing patient-specific characteristics. Then, the BiLSTM captures temporal dependencies in the extracted features to model seizure evolution patterns. Evaluation using EEG recordings from 20 patients with focal epilepsy demonstrated superior performance over non-adversarial methods, achieving high detection accuracy across different patients. The integration of adversarial training with temporal modeling enables robust cross-patient seizure detection.
Authors: Otto Tabell, Santtu Tikka, Juha Karvanen
Abstract: Data fusion, the process of combining observational and experimental data, can enable the identification of causal effects that would otherwise remain non-identifiable. Although identification algorithms have been developed for specific scenarios, do-calculus remains the only general-purpose tool for causal data fusion, particularly when variables are present in some data sources but not others. However, approaches based on do-calculus may encounter computational challenges as the number of variables increases and the causal graph grows in complexity. Consequently, there exists a need to reduce the size of such models while preserving the essential features. For this purpose, we propose pruning (removing unnecessary variables) and clustering (combining variables) as preprocessing operations for causal data fusion. We generalize earlier results on a single data source and derive conditions for applying pruning and clustering in the case of multiple data sources. We give sufficient conditions for inferring the identifiability or non-identifiability of a causal effect in a larger graph based on a smaller graph and show how to obtain the corresponding identifying functional for identifiable causal effects. Examples from epidemiology and social science demonstrate the use of the results.
Authors: Andy K. Zhang, Joey Ji, Celeste Menders, Riya Dulepet, Thomas Qin, Ron Y. Wang, Junrong Wu, Kyleen Liao, Jiliang Li, Jinghan Hu, Sara Hong, Nardos Demilew, Shivatmica Murgai, Jason Tran, Nishka Kacheria, Ethan Ho, Denis Liu, Lauren McLane, Olivia Bruvik, Dai-Rong Han, Seungwoo Kim, Akhil Vyas, Cuiyuanxiu Chen, Ryan Li, Weiran Xu, Jonathan Z. Ye, Prerit Choudhary, Siddharth M. Bhatia, Vikram Sivashankar, Yuxuan Bao, Dawn Song, Dan Boneh, Daniel E. Ho, Percy Liang
Abstract: AI agents have the potential to significantly alter the cybersecurity landscape. To help us understand this change, we introduce the first framework to capture offensive and defensive cyber-capabilities in evolving real-world systems. Instantiating this framework with BountyBench, we set up 25 systems with complex, real-world codebases. To capture the vulnerability lifecycle, we define three task types: Detect (detecting a new vulnerability), Exploit (exploiting a specific vulnerability), and Patch (patching a specific vulnerability). For Detect, we construct a new success indicator, which is general across vulnerability types and provides localized evaluation. We manually set up the environment for each system, including installing packages, setting up server(s), and hydrating database(s). We add 40 bug bounties, which are vulnerabilities with monetary awards from \$10 to \$30,485, and cover 9 of the OWASP Top 10 Risks. To modulate task difficulty, we devise a new strategy based on information to guide detection, interpolating from identifying a zero day to exploiting a specific vulnerability. We evaluate 5 agents: Claude Code, OpenAI Codex CLI, and custom agents with GPT-4.1, Gemini 2.5 Pro Preview, and Claude 3.7 Sonnet Thinking. Given up to three attempts, the top-performing agents are Claude Code (5% on Detect, mapping to \$1,350), Custom Agent with Claude 3.7 Sonnet Thinking (5% on Detect, mapping to \$1,025; 67.5% on Exploit), and OpenAI Codex CLI (5% on Detect, mapping to \$2,400; 90% on Patch, mapping to \$14,422). OpenAI Codex CLI and Claude Code are more capable at defense, achieving higher Patch scores of 90% and 87.5%, compared to Exploit scores of 32.5% and 57.5% respectively; in contrast, the custom agents are relatively balanced between offense and defense, achieving Exploit scores of 40-67.5% and Patch scores of 45-60%.
Authors: Itsuki Yazawa, Seitaro Yoneda, Akira Furui
Abstract: Electromyogram (EMG) signals recorded from the skin surface enable intuitive control of assistive devices such as prosthetic limbs. However, in EMG-based motion recognition, collecting comprehensive training data for all target motions remains challenging, particularly for complex combined motions. This paper proposes a method to efficiently recognize combined motions using synthetic EMG data generated through convex combinations of basic motion patterns. Instead of measuring all possible combined motions, the proposed method utilizes measured basic motion data along with synthetically combined motion data for training. This approach expands the range of recognizable combined motions while minimizing the required training data collection. We evaluated the effectiveness of the proposed method through an upper limb motion classification experiment with eight subjects. The experimental results demonstrated that the proposed method improved the classification accuracy for unseen combined motions by approximately 17%.
Authors: Yao Du, Huawei Fan, Xingang Wang
Abstract: A new machine learning scheme, termed versatile reservoir computing, is proposed for sustaining the dynamics of heterogeneous complex networks. We show that a single, small-scale reservoir computer trained on time series from a subset of elements is able to replicate the dynamics of any element in a large-scale complex network, though the elements are of different intrinsic parameters and connectivities. Furthermore, by substituting failed elements with the trained machine, we demonstrate that the collective dynamics of the network can be preserved accurately over a finite time horizon. The capability and effectiveness of the proposed scheme are validated on three representative network models: a homogeneous complex network of non-identical phase oscillators, a heterogeneous complex network of non-identical phase oscillators, and a heterogeneous complex network of non-identical chaotic oscillators.
Authors: Yassir Fathullah, Mark J. F. Gales
Abstract: This paper explores generalised probabilistic modelling and uncertainty estimation in comparative LLM-as-a-judge frameworks. We show that existing Product-of-Experts methods are specific cases of a broader framework, enabling diverse modelling options. Furthermore, we propose improved uncertainty estimates for individual comparisons, enabling more efficient selection and achieving strong performance with fewer evaluations. We also introduce a method for estimating overall ranking uncertainty. Finally, we demonstrate that combining absolute and comparative scoring improves performance. Experiments show that the specific expert model has a limited impact on final rankings but our proposed uncertainty estimates, especially the probability of reordering, significantly improve the efficiency of systems reducing the number of needed comparisons by ~50%. Furthermore, ranking-level uncertainty metrics can be used to identify low-performing predictions, where the nature of the probabilistic model has a notable impact on the quality of the overall uncertainty.
Authors: Andre Dourson, Kylie Taylor, Xiaoli Qiao, Michael Fitzke
Abstract: Self-supervised learning has emerged as a powerful paradigm for training deep neural networks, particularly in medical imaging where labeled data is scarce. While current approaches typically rely on synthetic augmentations of single images, we propose VET-DINO, a framework that leverages a unique characteristic of medical imaging: the availability of multiple standardized views from the same study. Using a series of clinical veterinary radiographs from the same patient study, we enable models to learn view-invariant anatomical structures and develop an implied 3D understanding from 2D projections. We demonstrate our approach on a dataset of 5 million veterinary radiographs from 668,000 canine studies. Through extensive experimentation, including view synthesis and downstream task performance, we show that learning from real multi-view pairs leads to superior anatomical understanding compared to purely synthetic augmentations. VET-DINO achieves state-of-the-art performance on various veterinary imaging tasks. Our work establishes a new paradigm for self-supervised learning in medical imaging that leverages domain-specific properties rather than merely adapting natural image techniques.
Authors: Zhengyi Li, Yue Guan, Kang Yang, Yu Feng, Ning Liu, Yu Yu, Jingwen Leng, Minyi Guo
Abstract: The wide deployment of the generative pre-trained transformer (GPT) has raised privacy concerns for both clients and servers. While cryptographic primitives can be employed for secure GPT inference to protect the privacy of both parties, they introduce considerable performance overhead.To accelerate secure inference, this study proposes a public decoding and secure verification approach that utilizes public GPT models, motivated by the observation that securely decoding one and multiple tokens takes a similar latency. The client uses the public model to generate a set of tokens, which are then securely verified by the private model for acceptance. The efficiency of our approach depends on the acceptance ratio of tokens proposed by the public model, which we improve from two aspects: (1) a private sampling protocol optimized for cryptographic primitives and (2) model alignment using knowledge distillation. Our approach improves the efficiency of secure decoding while maintaining the same level of privacy and generation quality as standard secure decoding. Experiments demonstrate a $2.1\times \sim 6.0\times$ speedup compared to standard decoding across three pairs of public-private models and different network conditions.
Authors: Om Khangaonkar, Hamed Pirsiavash
Abstract: By pretraining to synthesize coherent images from perturbed inputs, generative models inherently learn to understand object boundaries and scene compositions. How can we repurpose these generative representations for general-purpose perceptual organization? We finetune Stable Diffusion and MAE (encoder+decoder) for category-agnostic instance segmentation using our instance coloring loss exclusively on a narrow set of object types (indoor furnishings and cars). Surprisingly, our models exhibit strong zero-shot generalization, accurately segmenting objects of types and styles unseen in finetuning (and in many cases, MAE's ImageNet-1K pretraining too). Our best-performing models closely approach the heavily supervised SAM when evaluated on unseen object types and styles, and outperform it when segmenting fine structures and ambiguous boundaries. In contrast, existing promptable segmentation architectures or discriminatively pretrained models fail to generalize. This suggests that generative models learn an inherent grouping mechanism that transfers across categories and domains, even without internet-scale pretraining. Code, pretrained models, and demos are available on our website.
Authors: Seokjun Lee, Seung-Hyun Kong
Abstract: Oversteer, wherein a vehicle's rear tires lose traction and induce unintentional excessive yaw, poses critical safety challenges. Failing to control oversteer often leads to severe traffic accidents. Although recent autonomous driving efforts have attempted to handle oversteer through stabilizing maneuvers, the majority rely on expert-defined trajectories or assume obstacle-free environments, limiting real-world applicability. This paper introduces a novel end-to-end (E2E) autonomous driving approach that tackles oversteer control and collision avoidance simultaneously. Existing E2E techniques, including Imitation Learning (IL), Reinforcement Learning (RL), and Hybrid Learning (HL), generally require near-optimal demonstrations or extensive experience. Yet even skilled human drivers struggle to provide perfect demonstrations under oversteer, and high transition variance hinders accumulating sufficient data. Hence, we present Q-Compared Soft Actor-Critic (QC-SAC), a new HL algorithm that effectively learns from suboptimal demonstration data and adapts rapidly to new conditions. To evaluate QC-SAC, we introduce a benchmark inspired by real-world driver training: a vehicle encounters sudden oversteer on a slippery surface and must avoid randomly placed obstacles ahead. Experimental results show QC-SAC attains near-optimal driving policies, significantly surpassing state-of-the-art IL, RL, and HL baselines. Our method demonstrates the world's first safe autonomous oversteer control with obstacle avoidance.
Authors: Kaito Ariu, Po-An Wang, Alexandre Proutiere, Kenshi Abe
Abstract: We study the policy testing problem in discounted Markov decision processes (MDPs) under the fixed-confidence setting. The goal is to determine whether the value of a given policy exceeds a specified threshold while minimizing the number of observations. We begin by deriving an instance-specific lower bound that any algorithm must satisfy. This lower bound is characterized as the solution to an optimization problem with non-convex constraints. We propose a policy testing algorithm inspired by this optimization problem--a common approach in pure exploration problems such as best-arm identification, where asymptotically optimal algorithms often stem from such optimization-based characterizations. As for other pure exploration tasks in MDPs, however, the non-convex constraints in the lower-bound problem present significant challenges, raising doubts about whether statistically optimal and computationally tractable algorithms can be designed. To address this, we reformulate the lower-bound problem by interchanging the roles of the objective and the constraints, yielding an alternative problem with a non-convex objective but convex constraints. Strikingly, this reformulated problem admits an interpretation as a policy optimization task in a newly constructed reversed MDP. Leveraging recent advances in policy gradient methods, we efficiently solve this problem and use it to design a policy testing algorithm that is statistically optimal--matching the instance-specific lower bound on sample complexity--while remaining computationally tractable. We validate our approach with numerical experiments.
Authors: Enric Junqu\'e de Fortuny
Abstract: Despite the wide-scale usage and development of emotion classification datasets in NLP, the field lacks a standardized, large-scale resource that follows a psychologically grounded taxonomy. Existing datasets either use inconsistent emotion categories, suffer from limited sample size, or focus on specific domains. The Super Emotion Dataset addresses this gap by harmonizing diverse text sources into a unified framework based on Shaver's empirically validated emotion taxonomy, enabling more consistent cross-domain emotion recognition research.
Authors: Xabier de Zuazo, Eva Navas, Ibon Saratxaga, Mathieu Bourguignon, Nicola Molinaro
Abstract: Understanding the neural mechanisms underlying speech production is essential for both advancing cognitive neuroscience theory and developing practical communication technologies. In this study, we investigated magnetoencephalography signals to decode phones from brain activity during speech production and perception (passive listening and voice playback) tasks. Using a dataset comprising 17 participants, we performed pairwise phone classification, extending our analysis to 15 phonetic pairs. Multiple machine learning approaches, including regularized linear models and neural network architectures, were compared to determine their effectiveness in decoding phonetic information. Our results demonstrate significantly higher decoding accuracy during speech production (76.6%) compared to passive listening and playback modalities (~51%), emphasizing the richer neural information available during overt speech. Among the models, the Elastic Net classifier consistently outperformed more complex neural networks, highlighting the effectiveness of traditional regularization techniques when applied to limited and high-dimensional MEG datasets. Besides, analysis of specific brain frequency bands revealed that low-frequency oscillations, particularly Delta (0.2-3 Hz) and Theta (4-7 Hz), contributed the most substantially to decoding accuracy, suggesting that these bands encode critical speech production-related neural processes. Despite using advanced denoising methods, it remains unclear whether decoding solely reflects neural activity or if residual muscular or movement artifacts also contributed, indicating the need for further methodological refinement. Overall, our findings underline the critical importance of examining overt speech production paradigms, which, despite their complexity, offer opportunities to improve brain-computer interfaces to help individuals with severe speech impairments.
Authors: Seitaro Yoneda, Akira Furui
Abstract: In electromyogram (EMG)-based motion recognition, a subject-specific classifier is typically trained with sufficient labeled data. However, this process demands extensive data collection over extended periods, burdening the subject. To address this, utilizing information from pre-training on multiple subjects for the training of the target subject could be beneficial. This paper proposes an inter-subject variance transfer learning method based on a Bayesian approach. This method is founded on the simple hypothesis that while the means of EMG features vary greatly across subjects, their variances may exhibit similar patterns. Our approach transfers variance information, acquired through pre-training on multiple source subjects, to a target subject within a Bayesian updating framework, thereby allowing accurate classification using limited target calibration data. A coefficient was also introduced to adjust the amount of information transferred for efficient transfer learning. Experimental evaluations using two EMG datasets demonstrated the effectiveness of our variance transfer strategy and its superiority compared to existing methods.
Authors: Leon Chlon, Maggie Chlon, MarcAntonio M. Awada
Abstract: Real-world multimodal systems routinely face missing-input scenarios, and in reality, robots lose audio in a factory or a clinical record omits lab tests at inference time. Standard fusion layers either preserve robustness or calibration but never both. We introduce Adaptive Entropy-Gated Contrastive Fusion (AECF), a single light-weight layer that (i) adapts its entropy coefficient per instance, (ii) enforces monotone calibration across all modality subsets, and (iii) drives a curriculum mask directly from training-time entropy. On AV-MNIST and MS-COCO, AECF improves masked-input mAP by +18 pp at a 50% drop rate while reducing ECE by up to 200%, yet adds 1% run-time. All back-bones remain frozen, making AECF an easy drop-in layer for robust, calibrated multimodal inference.
Authors: Pritam Anand
Abstract: This paper explores Uncertainty Quantification (UQ) in SVM predictions, particularly for regression and forecasting tasks. Unlike the Neural Network, the SVM solutions are typically more stable, sparse, optimal and interpretable. However, there are only few literature which addresses the UQ in SVM prediction. At first, we provide a comprehensive summary of existing Prediction Interval (PI) estimation and probabilistic forecasting methods developed in the SVM framework and evaluate them against the key properties expected from an ideal PI model. We find that none of the existing SVM PI models achieves a sparse solution. To introduce sparsity in SVM model, we propose the Sparse Support Vector Quantile Regression (SSVQR) model, which constructs PIs and probabilistic forecasts by solving a pair of linear programs. Further, we develop a feature selection algorithm for PI estimation using SSVQR that effectively eliminates a significant number of features while improving PI quality in case of high-dimensional dataset. Finally we extend the SVM models in Conformal Regression setting for obtaining more stable prediction set with finite test set guarantees. Extensive experiments on artificial, real-world benchmark datasets compare the different characteristics of both existing and proposed SVM-based PI estimation methods and also highlight the advantages of the feature selection in PI estimation. Furthermore, we compare both, the existing and proposed SVM-based PI estimation models, with modern deep learning models for probabilistic forecasting tasks on benchmark datasets. Furthermore, SVM models show comparable or superior performance to modern complex deep learning models for probabilistic forecasting task in our experiments.
Authors: Nikita Kotelevskii, Mohsen Guizani, Eric Moulines, Maxim Panov
Abstract: Conformal prediction enables the construction of high-coverage prediction sets for any pre-trained model, guaranteeing that the true label lies within the set with a specified probability. However, these sets do not provide probability estimates for individual labels, limiting their practical use. In this paper, we propose, to the best of our knowledge, the first method for assigning calibrated probabilities to elements of a conformal prediction set. Our approach frames this as an adaptive calibration problem, selecting an input-specific temperature parameter to match the desired coverage level. Experiments on several challenging image classification datasets demonstrate that our method maintains coverage guarantees while significantly reducing expected calibration error.
Authors: David Nordstr\"om, Johan Edstedt, Fredrik Kahl, Georg B\"okman
Abstract: Recent efforts at scaling computer vision models have established Vision Transformers (ViTs) as the leading architecture. ViTs incorporate weight sharing over image patches as an important inductive bias. In this work, we show that ViTs benefit from incorporating equivariance under the octic group, i.e., reflections and 90-degree rotations, as a further inductive bias. We develop new architectures, octic ViTs, that use octic-equivariant layers and put them to the test on both supervised and self-supervised learning. Through extensive experiments on DeiT-III and DINOv2 training on ImageNet-1K, we show that octic ViTs yield more computationally efficient networks while also improving performance. In particular, we achieve approximately 40% reduction in FLOPs for ViT-H while simultaneously improving both classification and segmentation results.
Authors: Michal Kucha\v{r}, Jarom\'ir Fi\v{s}er, Cyril Oswald, Tom\'a\v{s} Vyhl\'idal
Abstract: The paper presents a decision support system for the long-term preservation of aeronautical heritage exhibited/stored in sheltered sites. The aeronautical heritage is characterized by diverse materials of which this heritage is constituted. Heritage aircraft are made of ancient aluminum alloys, (ply)wood, and particularly fabrics. The decision support system (DSS) designed, starting from a conceptual model, is knowledge-based on degradation/corrosion mechanisms of prevailing materials of aeronautical heritage. In the case of historical aircraft wooden parts, this knowledge base is filled in by the damage function models developed within former European projects. Model-based corrosion prediction is implemented within the new DSS for ancient aluminum alloys. The novelty of this DSS consists of supporting multi-material heritage protection and tailoring to peculiarities of aircraft exhibition/storage hangars and the needs of aviation museums. The novel DSS is tested on WWII aircraft heritage exhibited in the Aviation Museum Kbely, Military History Institute Prague, Czech Republic.
Authors: Shubhrangshu Debsarkar, Bijoy Kundu
Abstract: Dynamic FDG PET imaging study of n = 52 rats including 26 control Wistar-Kyoto (WKY) rats and 26 experimental spontaneously hypertensive rats (SHR) were performed using a Siemens microPET and Albira trimodal scanner longitudinally at 1, 2, 3, 5, 9, 12 and 18 months of age. A 15-parameter dual output model correcting for spill over contamination and partial volume effects with peak fitting cost functions was developed for simultaneous estimation of model corrected blood input function (MCIF) and kinetic rate constants for dynamic FDG PET images of rat heart in vivo. Major drawbacks of this model are its dependence on manual annotations for the Image Derived Input Function (IDIF) and manual determination of crucial model parameters to compute MCIF. To overcome these limitations, we performed semi-automated segmentation and then formulated a Long-Short-Term Memory (LSTM) cell network to train and predict MCIF in test data using a concatenation of IDIFs and myocardial inputs and compared them with reference-modeled MCIF. Thresholding along 2D plane slices with two thresholds, with T1 representing high-intensity myocardium, and T2 representing lower-intensity rings, was used to segment the area of the LV blood pool. The resultant IDIF and myocardial TACs were used to compute the corresponding reference (model) MCIF for all data sets. The segmented IDIF and the myocardium formed the input for the LSTM network. A k-fold cross validation structure with a 33:8:11 split and 5 folds was utilized to create the model and evaluate the performance of the LSTM network for all datasets. To overcome the sparseness of data as time steps increase, midpoint interpolation was utilized to increase the density of datapoints beyond time = 10 minutes. The model utilizing midpoint interpolation was able to achieve a 56.4% improvement over previous Mean Squared Error (MSE).
Authors: Tom Silver, Rajat Kumar Jenamani, Ziang Liu, Ben Dodson, Tapomayukh Bhattacharjee
Abstract: Generalist robots must personalize in-the-wild to meet the diverse needs and preferences of long-term users. How can we enable flexible personalization without sacrificing safety or competency? This paper proposes Coloring Between the Lines (CBTL), a method for personalization that exploits the null space of constraint satisfaction problems (CSPs) used in robot planning. CBTL begins with a CSP generator that ensures safe and competent behavior, then incrementally personalizes behavior by learning parameterized constraints from online interaction. By quantifying uncertainty and leveraging the compositionality of planning constraints, CBTL achieves sample-efficient adaptation without environment resets. We evaluate CBTL in (1) three diverse simulation environments; (2) a web-based user study; and (3) a real-robot assisted feeding system, finding that CBTL consistently achieves more effective personalization with fewer interactions than baselines. Our results demonstrate that CBTL provides a unified and practical approach for continual, flexible, active, and safe robot personalization. Website: https://emprise.cs.cornell.edu/cbtl/
Authors: Debarshi Brahma, Anuska Roy, Soma Biswas
Abstract: Recently, Vision-Language foundation models like CLIP and ALIGN, which are pre-trained on large-scale data have shown remarkable zero-shot generalization to diverse datasets with different classes and even domains. In this work, we take a step further and analyze whether these models can be adapted to target datasets having very different distributions and classes compared to what these models have been trained on, using only a few labeled examples from the target dataset. In such scenarios, finetuning large pretrained models is challenging due to problems of overfitting as well as loss of generalization, and has not been well explored in prior literature. Since, the pre-training data of such models are unavailable, it is difficult to comprehend the performance on various downstream datasets. First, we try to answer the question: Given a target dataset with a few labelled examples, can we estimate whether further fine-tuning can enhance the performance compared to zero-shot evaluation? by analyzing the common vision-language embedding space. Based on the analysis, we propose a novel prompt-tuning method, PromptMargin for adapting such large-scale VLMs directly on the few target samples. PromptMargin effectively tunes the text as well as visual prompts for this task, and has two main modules: 1) Firstly, we use a selective augmentation strategy to complement the few training samples in each task; 2) Additionally, to ensure robust training in the presence of unfamiliar class names, we increase the inter-class margin for improved class discrimination using a novel Multimodal Margin Regularizer. Extensive experiments and analysis across fifteen target benchmark datasets, with varying degrees of distribution shifts from natural images, shows the effectiveness of the proposed framework over the existing state-of-the-art approaches applied to this setting. github.com/debarshigit/PromptMargin.
Authors: Kaiyuan Chen, Shuangyu Xie, Zehan Ma, Ken Goldberg
Abstract: Vision-Language Models (VLMs) acquire real-world knowledge and general reasoning ability through Internet-scale image-text corpora. They can augment robotic systems with scene understanding and task planning, and assist visuomotor policies that are trained on robot trajectory data. We explore the reverse paradigm - using rich, real, multi-modal robot trajectory data to enhance and evaluate VLMs. In this paper, we present Robo2VLM, a Visual Question Answering (VQA) dataset generation framework for VLMs. Given a human tele-operated robot trajectory, Robo2VLM derives ground-truth from non-visual and non-descriptive sensory modalities, such as end-effector pose, gripper aperture, and force sensing. Based on these modalities, it segments the robot trajectory into a sequence of manipulation phases. At each phase, Robo2VLM uses scene and interaction understanding to identify 3D properties of the robot, task goal, and the target object. The properties are used to generate representative VQA queries - images with textural multiple-choice questions - based on spatial, goal-conditioned, and interaction reasoning question templates. We curate Robo2VLM-1, a large-scale in-the-wild dataset with 684,710 questions covering 463 distinct scenes and 3,396 robotic manipulation tasks from 176k real robot trajectories. Results suggest that Robo2VLM-1 can benchmark and improve VLM capabilities in spatial and interaction reasoning.
Authors: Anna R. Flowers, Christopher T. Franck, Micka\"el Binois, Chiwoo Park, Robert B. Gramacy
Abstract: Gaussian processes (GPs) furnish accurate nonlinear predictions with well-calibrated uncertainty. However, the typical GP setup has a built-in stationarity assumption, making it ill-suited for modeling data from processes with sudden changes, or "jumps" in the output variable. The "jump GP" (JGP) was developed for modeling data from such processes, combining local GPs and latent "level" variables under a joint inferential framework. But joint modeling can be fraught with difficulty. We aim to simplify by suggesting a more modular setup, eschewing joint inference but retaining the main JGP themes: (a) learning optimal neighborhood sizes that locally respect manifolds of discontinuity; and (b) a new cluster-based (latent) feature to capture regions of distinct output levels on both sides of the manifold. We show that each of (a) and (b) separately leads to dramatic improvements when modeling processes with jumps. In tandem (but without requiring joint inference) that benefit is compounded, as illustrated on real and synthetic benchmark examples from the recent literature.
Authors: Kaiyuan Chen, Letian Fu, David Huang, Yanxiang Zhang, Lawrence Yunliang Chen, Huang Huang, Kush Hari, Ashwin Balakrishna, Ted Xiao, Pannag R Sanketi, John Kubiatowicz, Ken Goldberg
Abstract: Recent results suggest that very large datasets of teleoperated robot demonstrations can be used to train transformer-based models that have the potential to generalize to new scenes, robots, and tasks. However, curating, distributing, and loading large datasets of robot trajectories, which typically consist of video, textual, and numerical modalities - including streams from multiple cameras - remains challenging. We propose Robo-DM, an efficient open-source cloud-based data management toolkit for collecting, sharing, and learning with robot data. With Robo-DM, robot datasets are stored in a self-contained format with Extensible Binary Meta Language (EBML). Robo-DM can significantly reduce the size of robot trajectory data, transfer costs, and data load time during training. Compared to the RLDS format used in OXE datasets, Robo-DM's compression saves space by up to 70x (lossy) and 3.5x (lossless). Robo-DM also accelerates data retrieval by load-balancing video decoding with memory-mapped decoding caches. Compared to LeRobot, a framework that also uses lossy video compression, Robo-DM is up to 50x faster when decoding sequentially. We physically evaluate a model trained by Robo-DM with lossy compression, a pick-and-place task, and In-Context Robot Transformer. Robo-DM uses 75x compression of the original dataset and does not suffer reduction in downstream task accuracy.
Authors: Xin Huang, Ruibin Li, Tong Jia, Wei Zheng, Ya Wang
Abstract: Vision-Language Models (VLMs) are essential for multimodal tasks, especially compositional reasoning (CR) tasks, which require distinguishing fine-grained semantic differences between visual and textual embeddings. However, existing methods primarily fine-tune the model by generating text-based hard negative samples, neglecting the importance of image-based negative samples, which results in insufficient training of the visual encoder and ultimately impacts the overall performance of the model. Moreover, negative samples are typically treated uniformly, without considering their difficulty levels, and the alignment of positive samples is insufficient, which leads to challenges in aligning difficult sample pairs. To address these issues, we propose Adaptive Hard Negative Perturbation Learning (AHNPL). AHNPL translates text-based hard negatives into the visual domain to generate semantically disturbed image-based negatives for training the model, thereby enhancing its overall performance. AHNPL also introduces a contrastive learning approach using a multimodal hard negative loss to improve the model's discrimination of hard negatives within each modality and a dynamic margin loss that adjusts the contrastive margin according to sample difficulty to enhance the distinction of challenging sample pairs. Experiments on three public datasets demonstrate that our method effectively boosts VLMs' performance on complex CR tasks. The source code is available at https://github.com/nynu-BDAI/AHNPL.
Authors: Haocheng Ju, Bin Dong
Abstract: Mathematical Information Retrieval (MIR) is the task of retrieving information from mathematical documents and plays a key role in various applications, including theorem search in mathematical libraries, answer retrieval on math forums, and premise selection in automated theorem proving. However, a unified benchmark for evaluating these diverse retrieval tasks has been lacking. In this paper, we introduce MIRB (Mathematical Information Retrieval Benchmark) to assess the MIR capabilities of retrieval models. MIRB includes four tasks: semantic statement retrieval, question-answer retrieval, premise retrieval, and formula retrieval, spanning a total of 12 datasets. We evaluate 13 retrieval models on this benchmark and analyze the challenges inherent to MIR. We hope that MIRB provides a comprehensive framework for evaluating MIR systems and helps advance the development of more effective retrieval models tailored to the mathematical domain.
Authors: Wei Liu, Ruochen Zhou, Yiyun Deng, Yuzhen Huang, Junteng Liu, Yuntian Deng, Yizhe Zhang, Junxian He
Abstract: Large Reasoning Models (LRMs) have shown remarkable capabilities in solving complex problems through reinforcement learning (RL), particularly by generating long reasoning traces. However, these extended outputs often exhibit substantial redundancy, which limits the efficiency of LRMs. In this paper, we investigate RL-based approaches to promote reasoning efficiency. Specifically, we first present a unified framework that formulates various efficient reasoning methods through the lens of length-based reward shaping. Building on this perspective, we propose a novel Length-bAsed StEp Reward shaping method (LASER), which employs a step function as the reward, controlled by a target length. LASER surpasses previous methods, achieving a superior Pareto-optimal balance between performance and efficiency. Next, we further extend LASER based on two key intuitions: (1) The reasoning behavior of the model evolves during training, necessitating reward specifications that are also adaptive and dynamic; (2) Rather than uniformly encouraging shorter or longer chains of thought (CoT), we posit that length-based reward shaping should be difficulty-aware i.e., it should penalize lengthy CoTs more for easy queries. This approach is expected to facilitate a combination of fast and slow thinking, leading to a better overall tradeoff. The resulting method is termed LASER-D (Dynamic and Difficulty-aware). Experiments on DeepSeek-R1-Distill-Qwen-1.5B, DeepSeek-R1-Distill-Qwen-7B, and DeepSeek-R1-Distill-Qwen-32B show that our approach significantly enhances both reasoning performance and response length efficiency. For instance, LASER-D and its variant achieve a +6.1 improvement on AIME2024 while reducing token usage by 63%. Further analysis reveals our RL-based compression produces more concise reasoning patterns with less redundant "self-reflections". Resources are at https://github.com/hkust-nlp/Laser.
Authors: Tiasa Singha Roy, Aditeya Baral, Ayush Rajesh Jhaveri, Yusuf Baig
Abstract: Large language models (LLMs) demonstrate considerable potential in various natural language tasks but face significant challenges in mathematical reasoning, particularly in executing precise, multi-step logic. However, current evaluation frameworks judge their performance solely based on accuracy, which only accounts for the final answer. This study explores these pitfalls by employing a novel evaluation framework. We propose an evaluation metric called the MAPLE score, which holistically quantifies reasoning misalignment by integrating error rates, redundancy, and validity.
Authors: David Thulke, Jakob Kemmler, Christian Dugast, Hermann Ney
Abstract: Large language models that use retrieval augmented generation have the potential to unlock valuable knowledge for researchers, policymakers, and the public by making long and technical climate-related documents more accessible. While this approach can help alleviate factual hallucinations by relying on retrieved passages as additional context, its effectiveness depends on whether the model's output remains faithful to these passages. To address this, we explore the automatic assessment of faithfulness of different models in this setting. We then focus on ClimateGPT, a large language model specialised in climate science, to examine which factors in its instruction fine-tuning impact the model's faithfulness. By excluding unfaithful subsets of the model's training data, we develop ClimateGPT Faithful+, which achieves an improvement in faithfulness from 30% to 57% in supported atomic claims according to our automatic metric.
Authors: Zihao Li, Xu Wang, Yuzhe Yang, Ziyu Yao, Haoyi Xiong, Mengnan Du
Abstract: Large Language Models (LLMs) demonstrate the ability to solve reasoning and mathematical problems using the Chain-of-Thought (CoT) technique. Expanding CoT length, as seen in models such as DeepSeek-R1, significantly enhances this reasoning for complex problems, but requires costly and high-quality long CoT data and fine-tuning. This work, inspired by the deep thinking paradigm of DeepSeek-R1, utilizes a steering technique to enhance the reasoning ability of an LLM without external datasets. Our method first employs Sparse Autoencoders (SAEs) to extract interpretable features from vanilla CoT. These features are then used to steer the LLM's internal states during generation. Recognizing that many LLMs do not have corresponding pre-trained SAEs, we further introduce a novel SAE-free steering algorithm, which directly computes steering directions from the residual activations of an LLM, obviating the need for an explicit SAE. Experimental results demonstrate that both our SAE-based and subsequent SAE-free steering algorithms significantly enhance the reasoning capabilities of LLMs.
Authors: Yousef Al-Jazzazi, Haya Diwan, Jinrui Gou, Cameron Musco, Christopher Musco, Torsten Suel
Abstract: Nearest neighbor search is central in machine learning, information retrieval, and databases. For high-dimensional datasets, graph-based methods such as HNSW, DiskANN, and NSG have become popular thanks to their empirical accuracy and efficiency. These methods construct a directed graph over the dataset and perform beam search on the graph to find nodes close to a given query. While significant work has focused on practical refinements and theoretical understanding of graph-based methods, many questions remain. We propose a new distance-based termination condition for beam search to replace the commonly used condition based on beam width. We prove that, as long as the search graph is navigable, our resulting Adaptive Beam Search method is guaranteed to approximately solve the nearest-neighbor problem, establishing a connection between navigability and the performance of graph-based search. We also provide extensive experiments on our new termination condition for both navigable graphs and approximately navigable graphs used in practice, such as HNSW and Vamana graphs. We find that Adaptive Beam Search outperforms standard beam search over a range of recall values, data sets, graph constructions, and target number of nearest neighbors. It thus provides a simple and practical way to improve the performance of popular methods.
Authors: Zhengjia Zhuo, Viswanath Nagarajan
Abstract: Optimal decision tree (\odt) is a fundamental problem arising in applications such as active learning, entity identification, and medical diagnosis. An instance of \odt is given by $m$ hypotheses, out of which an unknown ``true'' hypothesis is drawn according to some probability distribution. An algorithm needs to identify the true hypothesis by making queries: each query incurs a cost and has a known response for each hypothesis. The goal is to minimize the expected query cost to identify the true hypothesis. We consider the most general setting with arbitrary costs, probabilities and responses. \odt is NP-hard to approximate better than $\ln m$ and there are $O(\ln m)$ approximation algorithms known for it. However, these algorithms and/or their analyses are quite complex. Moreover, the leading constant factors are large. We provide a simple algorithm and analysis for \odt, proving an approximation ratio of $8 \ln m$.
Authors: Ruijie Zheng, Jing Wang, Scott Reed, Johan Bjorck, Yu Fang, Fengyuan Hu, Joel Jang, Kaushil Kundalia, Zongyu Lin, Loic Magne, Avnish Narayan, You Liang Tan, Guanzhi Wang, Qi Wang, Jiannan Xiang, Yinzhen Xu, Seonghyeon Ye, Jan Kautz, Furong Huang, Yuke Zhu, Linxi Fan
Abstract: We introduce $\textbf{F}$uture $\textbf{LA}$tent $\textbf{RE}$presentation Alignment ($\textbf{FLARE}$), a novel framework that integrates predictive latent world modeling into robot policy learning. By aligning features from a diffusion transformer with latent embeddings of future observations, $\textbf{FLARE}$ enables a diffusion transformer policy to anticipate latent representations of future observations, allowing it to reason about long-term consequences while generating actions. Remarkably lightweight, $\textbf{FLARE}$ requires only minimal architectural modifications -- adding a few tokens to standard vision-language-action (VLA) models -- yet delivers substantial performance gains. Across two challenging multitask simulation imitation learning benchmarks spanning single-arm and humanoid tabletop manipulation, $\textbf{FLARE}$ achieves state-of-the-art performance, outperforming prior policy learning baselines by up to 26%. Moreover, $\textbf{FLARE}$ unlocks the ability to co-train with human egocentric video demonstrations without action labels, significantly boosting policy generalization to a novel object with unseen geometry with as few as a single robot demonstration. Our results establish $\textbf{FLARE}$ as a general and scalable approach for combining implicit world modeling with high-frequency robotic control.
Authors: Jinyang Wu, Chonghua Liao, Mingkuan Feng, Shuai Zhang, Zhengqi Wen, Pengpeng Shao, Huazhe Xu, Jianhua Tao
Abstract: Reinforcement learning (RL) has emerged as an effective method for training reasoning models. However, existing RL approaches typically bias the model's output distribution toward reward-maximizing paths without introducing external knowledge. This limits their exploration capacity and results in a narrower reasoning capability boundary compared to base models. To address this limitation, we propose TAPO (Thought-Augmented Policy Optimization), a novel framework that augments RL by incorporating external high-level guidance ("thought patterns"). By adaptively integrating structured thoughts during training, TAPO effectively balances model-internal exploration and external guidance exploitation. Extensive experiments show that our approach significantly outperforms GRPO by 99% on AIME, 41% on AMC, and 17% on Minerva Math. Notably, these high-level thought patterns, abstracted from only 500 prior samples, generalize effectively across various tasks and models. This highlights TAPO's potential for broader applications across multiple tasks and domains. Our further analysis reveals that introducing external guidance produces powerful reasoning models with superior explainability of inference behavior and enhanced output readability.
Authors: Maike Behrendt, Stefan Sylvius Wagner, Stefan Harmeling
Abstract: The [CLS] token in BERT is commonly used as a fixed-length representation for classification tasks, yet prior work has shown that both other tokens and intermediate layers encode valuable contextual information. In this work, we propose MaxPoolBERT, a lightweight extension to BERT that refines the [CLS] representation by aggregating information across layers and tokens. Specifically, we explore three modifications: (i) max-pooling the [CLS] token across multiple layers, (ii) enabling the [CLS] token to attend over the entire final layer using an additional multi-head attention (MHA) layer, and (iii) combining max-pooling across the full sequence with MHA. Our approach enhances BERT's classification accuracy (especially on low-resource tasks) without requiring pre-training or significantly increasing model size. Experiments on the GLUE benchmark show that MaxPoolBERT consistently achieves a better performance on the standard BERT-base model.
Authors: Pingqing Zheng (Katie), Jiayin Qin (Katie), Fuqi Zhang (Katie), Shang Wu (Katie), Yu Cao (Katie), Caiwen Ding (Katie), Yang (Katie), Zhao
Abstract: Large Language Models (LLMs) have demonstrated their potential in hardware design tasks, such as Hardware Description Language (HDL) generation and debugging. Yet, their performance in real-world, repository-level HDL projects with thousands or even tens of thousands of code lines is hindered. To this end, we propose HDLxGraph, a novel framework that integrates Graph Retrieval Augmented Generation (Graph RAG) with LLMs, introducing HDL-specific graph representations by incorporating Abstract Syntax Trees (ASTs) and Data Flow Graphs (DFGs) to capture both code graph view and hardware graph view. HDLxGraph utilizes a dual-retrieval mechanism that not only mitigates the limited recall issues inherent in similarity-based semantic retrieval by incorporating structural information, but also enhances its extensibility to various real-world tasks by a task-specific retrieval finetuning. Additionally, to address the lack of comprehensive HDL search benchmarks, we introduce HDLSearch, a multi-granularity evaluation dataset derived from real-world repository-level projects. Experimental results demonstrate that HDLxGraph significantly improves average search accuracy, debugging efficiency and completion quality by 12.04%, 12.22% and 5.04% compared to similarity-based RAG, respectively. The code of HDLxGraph and collected HDLSearch benchmark are available at https://github.com/Nick-Zheng-Q/HDLxGraph.
Authors: Tianqi Du, Zeming Wei, Quan Chen, Chenheng Zhang, Yisen Wang
Abstract: The rapid advancement of large language models (LLMs) has demonstrated milestone success in a variety of tasks, yet their potential for generating harmful content has raised significant safety concerns. Existing safety evaluation approaches typically operate directly on textual responses, overlooking the rich information embedded in the model's internal representations. In this paper, we propose Safety Representation Ranking (SRR), a listwise ranking framework that selects safe responses using hidden states from the LLM itself. SRR encodes both instructions and candidate completions using intermediate transformer representations and ranks candidates via a lightweight similarity-based scorer. Our approach directly leverages internal model states and supervision at the list level to capture subtle safety signals. Experiments across multiple benchmarks show that SRR significantly improves robustness to adversarial prompts. Our code will be available upon publication.
Authors: Luqin Gan, Tarek M. Zikry, Genevera I. Allen
Abstract: As machine learning systems are increasingly used in high-stakes domains, there is a growing emphasis placed on making them interpretable to improve trust in these systems. In response, a range of interpretable machine learning (IML) methods have been developed to generate human-understandable insights into otherwise black box models. With these methods, a fundamental question arises: Are these interpretations reliable? Unlike with prediction accuracy or other evaluation metrics for supervised models, the proximity to the true interpretation is difficult to define. Instead, we ask a closely related question that we argue is a prerequisite for reliability: Are these interpretations stable? We define stability as findings that are consistent or reliable under small random perturbations to the data or algorithms. In this study, we conduct the first systematic, large-scale empirical stability study on popular machine learning global interpretations for both supervised and unsupervised tasks on tabular data. Our findings reveal that popular interpretation methods are frequently unstable, notably less stable than the predictions themselves, and that there is no association between the accuracy of machine learning predictions and the stability of their associated interpretations. Moreover, we show that no single method consistently provides the most stable interpretations across a range of benchmark datasets. Overall, these results suggest that interpretability alone does not warrant trust, and underscores the need for rigorous evaluation of interpretation stability in future work. To support these principles, we have developed and released an open source IML dashboard and Python package to enable researchers to assess the stability and reliability of their own data-driven interpretations and discoveries.
Authors: Gaurav Srivastava, Zhenyu Bi, Meng Lu, Xuan Wang
Abstract: Large language models (LLMs) have improved significantly in their reasoning through extensive training on massive datasets. However, relying solely on additional data for improvement is becoming increasingly impractical, highlighting the need for models to autonomously enhance their reasoning without external supervision. In this paper, we propose Debate, Train, Evolve (DTE), a novel ground truth-free training framework that uses multi-agent debate traces to evolve a single language model. We also introduce a new prompting strategy Reflect-Critique-Refine, to improve debate quality by explicitly instructing agents to critique and refine their reasoning. Extensive evaluations on five reasoning benchmarks with six open-weight models show that our DTE framework achieve substantial improvements, with an average accuracy gain of 8.92% on the challenging GSM-PLUS dataset. Furthermore, we observe strong cross-domain generalization, with an average accuracy gain of 5.8% on all other benchmarks, suggesting that our method captures general reasoning capabilities.
Authors: Xiaoxue Yang, Bozhidar Stevanoski, Matthieu Meeus, Yves-Alexandre de Montjoye
Abstract: Large language models (LLMs) are rapidly deployed in real-world applications ranging from chatbots to agentic systems. Alignment is one of the main approaches used to defend against attacks such as prompt injection and jailbreaks. Recent defenses report near-zero Attack Success Rates (ASR) even against Greedy Coordinate Gradient (GCG), a white-box attack that generates adversarial suffixes to induce attacker-desired outputs. However, this search space over discrete tokens is extremely large, making the task of finding successful attacks difficult. GCG has, for instance, been shown to converge to local minima, making it sensitive to initialization choices. In this paper, we assess the future-proof robustness of these defenses using a more informed threat model: attackers who have access to some information about the alignment process. Specifically, we propose an informed white-box attack leveraging the intermediate model checkpoints to initialize GCG, with each checkpoint acting as a stepping stone for the next one. We show this approach to be highly effective across state-of-the-art (SOTA) defenses and models. We further show our informed initialization to outperform other initialization methods and show a gradient-informed checkpoint selection strategy to greatly improve attack performance and efficiency. Importantly, we also show our method to successfully find universal adversarial suffixes -- single suffixes effective across diverse inputs. Our results show that, contrary to previous beliefs, effective adversarial suffixes do exist against SOTA alignment-based defenses, that these can be found by existing attack methods when adversaries exploit alignment knowledge, and that even universal suffixes exist. Taken together, our results highlight the brittleness of current alignment-based methods and the need to consider stronger threat models when testing the safety of LLMs.
Authors: Adam Gould, Francesca Toni
Abstract: We introduce Gradual Abstract Argumentation for Case-Based Reasoning (Gradual AA-CBR), a data-driven, neurosymbolic classification model in which the outcome is determined by an argumentation debate structure that is learned simultaneously with neural-based feature extractors. Each argument in the debate is an observed case from the training data, favouring their labelling. Cases attack or support those with opposing or agreeing labellings, with the strength of each argument and relationship learned through gradient-based methods. This argumentation debate structure provides human-aligned reasoning, improving model interpretability compared to traditional neural networks (NNs). Unlike the existing purely symbolic variant, Abstract Argumentation for Case-Based Reasoning (AA-CBR), Gradual AA-CBR is capable of multi-class classification, automatic learning of feature and data point importance, assigning uncertainty values to outcomes, using all available data points, and does not require binary features. We show that Gradual AA-CBR performs comparably to NNs whilst significantly outperforming existing AA-CBR formulations.
Authors: Taiye Chen, Zeming Wei, Ang Li, Yisen Wang
Abstract: Large Language Models (LLMs) are known to be vulnerable to jailbreaking attacks, wherein adversaries exploit carefully engineered prompts to induce harmful or unethical responses. Such threats have raised critical concerns about the safety and reliability of LLMs in real-world deployment. While existing defense mechanisms partially mitigate such risks, subsequent advancements in adversarial techniques have enabled novel jailbreaking methods to circumvent these protections, exposing the limitations of static defense frameworks. In this work, we explore defending against evolving jailbreaking threats through the lens of context retrieval. First, we conduct a preliminary study demonstrating that even a minimal set of safety-aligned examples against a particular jailbreak can significantly enhance robustness against this attack pattern. Building on this insight, we further leverage the retrieval-augmented generation (RAG) techniques and propose Safety Context Retrieval (SCR), a scalable and robust safeguarding paradigm for LLMs against jailbreaking. Our comprehensive experiments demonstrate how SCR achieves superior defensive performance against both established and emerging jailbreaking tactics, contributing a new paradigm to LLM safety. Our code will be available upon publication.
Authors: Mikhail Budnikov, Ivan Yamshchikov
Abstract: This work explores transfer learning from several synthetic languages to English. We investigate the structure of the embeddings in the fine-tuned models, the information they contain, and the capabilities of the fine-tuned models on simple linguistic tasks. We also introduce a new synthetic language that leads to better transfer to English than the languages used in previous research. Finally, we introduce Tiny-Cloze Benchmark - a new synthetic benchmark for natural language understanding that is more informative for less powerful models. We use Tiny-Cloze Benchmark to evaluate fine-tuned models in several domains demonstrating that fine-tuning on a new synthetic language allows for better performance on a variety of tasks.
Authors: Huanxuan Liao, Wen Hu, Yao Xu, Shizhu He, Jun Zhao, Kang Liu
Abstract: Large Language Models (LLMs) encounter significant challenges in long-sequence inference due to computational inefficiency and redundant processing, driving interest in context compression techniques. Existing methods often rely on token importance to perform hard local compression or encode context into latent representations for soft global compression. However, the uneven distribution of textual content relevance and the diversity of demands for user instructions mean these approaches frequently lead to the loss of potentially valuable information. To address this, we propose $\textbf{Hy}$brid $\textbf{Co}$ntext $\textbf{Co}$mpression (HyCo$_2$) for LLMs, which integrates both global and local perspectives to guide context compression while retaining both the essential semantics and critical details for task completion. Specifically, we employ a hybrid adapter to refine global semantics with the global view, based on the observation that different adapters excel at different tasks. Then we incorporate a classification layer that assigns a retention probability to each context token based on the local view, determining whether it should be retained or discarded. To foster a balanced integration of global and local compression, we introduce auxiliary paraphrasing and completion pretraining before instruction tuning. This promotes a synergistic integration that emphasizes instruction-relevant information while preserving essential local details, ultimately balancing local and global information retention in context compression. Experiments show that our HyCo$_2$ method significantly enhances long-text reasoning while reducing token usage. It improves the performance of various LLM series by an average of 13.1\% across seven knowledge-intensive QA benchmarks. Moreover, HyCo$_2$ matches the performance of uncompressed methods while reducing token consumption by 88.8\%.
Authors: Fengyuan Dai, Zifeng Zhuang, Yufei Huang, Siteng Huang, Bangyan Liao, Donglin Wang, Fajie Yuan
Abstract: Diffusion models have emerged as powerful generative tools across various domains, yet tailoring pre-trained models to exhibit specific desirable properties remains challenging. While reinforcement learning (RL) offers a promising solution,current methods struggle to simultaneously achieve stable, efficient fine-tuning and support non-differentiable rewards. Furthermore, their reliance on sparse rewards provides inadequate supervision during intermediate steps, often resulting in suboptimal generation quality. To address these limitations, dense and differentiable signals are required throughout the diffusion process. Hence, we propose VAlue-based Reinforced Diffusion (VARD): a novel approach that first learns a value function predicting expection of rewards from intermediate states, and subsequently uses this value function with KL regularization to provide dense supervision throughout the generation process. Our method maintains proximity to the pretrained model while enabling effective and stable training via backpropagation. Experimental results demonstrate that our approach facilitates better trajectory guidance, improves training efficiency and extends the applicability of RL to diffusion models optimized for complex, non-differentiable reward functions.
Authors: Danna Zheng, Mirella Lapata, Jeff Z. Pan
Abstract: Information alignment evaluators are vital for various NLG evaluation tasks and trustworthy LLM deployment, reducing hallucinations and enhancing user trust. Current fine-grained methods, like FactScore, verify facts individually but neglect inter-fact dependencies, enabling subtle vulnerabilities. In this work, we introduce MontageLie, a challenging benchmark that constructs deceptive narratives by "montaging" truthful statements without introducing explicit hallucinations. We demonstrate that both coarse-grained LLM-based evaluators and current fine-grained frameworks are susceptible to this attack, with AUC-ROC scores falling below 65%. To enable more robust fine-grained evaluation, we propose DoveScore, a novel framework that jointly verifies factual accuracy and event-order consistency. By modeling inter-fact relationships, DoveScore outperforms existing fine-grained methods by over 8%, providing a more robust solution for long-form text alignment evaluation. Our code and datasets are available at https://github.com/dannalily/DoveScore.
Authors: Zhiwen Chen, Bo Leng, Zhuoren Li, Hanming Deng, Guizhe Jin, Ran Yu, Huanxi Wen
Abstract: Integrating Large Language Models (LLMs) with Reinforcement Learning (RL) can enhance autonomous driving (AD) performance in complex scenarios. However, current LLM-Dominated RL methods over-rely on LLM outputs, which are prone to hallucinations.Evaluations show that state-of-the-art LLM indicates a non-hallucination rate of only approximately 57.95% when assessed on essential driving-related tasks. Thus, in these methods, hallucinations from the LLM can directly jeopardize the performance of driving policies. This paper argues that maintaining relative independence between the LLM and the RL is vital for solving the hallucinations problem. Consequently, this paper is devoted to propose a novel LLM-Hinted RL paradigm. The LLM is used to generate semantic hints for state augmentation and policy optimization to assist RL agent in motion planning, while the RL agent counteracts potential erroneous semantic indications through policy learning to achieve excellent driving performance. Based on this paradigm, we propose the HCRMP (LLM-Hinted Contextual Reinforcement Learning Motion Planner) architecture, which is designed that includes Augmented Semantic Representation Module to extend state space. Contextual Stability Anchor Module enhances the reliability of multi-critic weight hints by utilizing information from the knowledge base. Semantic Cache Module is employed to seamlessly integrate LLM low-frequency guidance with RL high-frequency control. Extensive experiments in CARLA validate HCRMP's strong overall driving performance. HCRMP achieves a task success rate of up to 80.3% under diverse driving conditions with different traffic densities. Under safety-critical driving conditions, HCRMP significantly reduces the collision rate by 11.4%, which effectively improves the driving performance in complex scenarios.
Authors: Patrick Kahardipraja, Reduan Achtibat, Thomas Wiegand, Wojciech Samek, Sebastian Lapuschkin
Abstract: Large language models are able to exploit in-context learning to access external knowledge beyond their training data through retrieval-augmentation. While promising, its inner workings remain unclear. In this work, we shed light on the mechanism of in-context retrieval augmentation for question answering by viewing a prompt as a composition of informational components. We propose an attribution-based method to identify specialized attention heads, revealing in-context heads that comprehend instructions and retrieve relevant contextual information, and parametric heads that store entities' relational knowledge. To better understand their roles, we extract function vectors and modify their attention weights to show how they can influence the answer generation process. Finally, we leverage the gained insights to trace the sources of knowledge used during inference, paving the way towards more safe and transparent language models.
Authors: Frida Marie Viset, Rudy Helmons, Manon Kok
Abstract: When an agent, person, vehicle or robot is moving through an unknown environment without GNSS signals, online mapping of nonlinear terrains can be used to improve position estimates when the agent returns to a previously mapped area. Mapping algorithms using online Gaussian process (GP) regression are commonly integrated in algorithms for simultaneous localisation and mapping (SLAM). However, GP mapping algorithms have increasing computational demands as the mapped area expands relative to spatial field variations. This is due to the need for estimating an increasing amount of map parameters as the area of the map grows. Contrary to this, we propose a recursive GP mapping estimation algorithm which uses local basis functions in an information filter to achieve spatial scalability. Our proposed approximation employs a global grid of finite support basis functions but restricts computations to a localized subset around each prediction point. As our proposed algorithm is recursive, it can naturally be incorporated into existing algorithms that uses Gaussian process maps for SLAM. Incorporating our proposed algorithm into an extended Kalman filter (EKF) for magnetic field SLAM reduces the overall computational complexity of the algorithm. We show experimentally that our algorithm is faster than existing methods when the mapped area is large and the map is based on many measurements, both for recursive mapping tasks and for magnetic field SLAM.
Authors: Yue Yao, Daniel Goehring, Joerg Reichardt
Abstract: Linear trajectory models provide mathematical advantages to autonomous driving applications such as motion prediction. However, linear models' expressive power and bias for real-world trajectories have not been thoroughly analyzed. We present an in-depth empirical analysis of the trade-off between model complexity and fit error in modelling object trajectories. We analyze vehicle, cyclist, and pedestrian trajectories. Our methodology estimates observation noise and prior distributions over model parameters from several large-scale datasets. Incorporating these priors can then regularize prediction models. Our results show that linear models do represent real-world trajectories with high fidelity at very moderate model complexity. This suggests the feasibility of using linear trajectory models in future motion prediction systems with inherent mathematical advantages.
Authors: Giuseppe Floris, Christian Scano, Biagio Montaruli, Luca Demetrio, Andrea Valenza, Luca Compagna, Davide Ariu, Luca Piras, Davide Balzarotti, Battista Biggio
Abstract: Many Web Application Firewalls (WAFs) leverage the OWASP CRS to block incoming malicious requests. The CRS consists of different sets of rules designed by domain experts to detect well-known web attack patterns. Both the set of rules and the weights used to combine them are manually defined, yielding four different default configurations of the CRS. In this work, we focus on the detection of SQLi attacks, and show that the manual configurations of the CRS typically yield a suboptimal trade-off between detection and false alarm rates. Furthermore, we show that these configurations are not robust to adversarial SQLi attacks, i.e., carefully-crafted attacks that iteratively refine the malicious SQLi payload by querying the target WAF to bypass detection. To overcome these limitations, we propose (i) using machine learning to automate the selection of the set of rules to be combined along with their weights, i.e., customizing the CRS configuration based on the monitored web services; and (ii) leveraging adversarial training to significantly improve its robustness to adversarial SQLi manipulations. Our experiments, conducted using the well-known open-source ModSecurity WAF equipped with the CRS rules, show that our approach, named ModSec-AdvLearn, can (i) increase the detection rate up to 30%, while retaining negligible false alarm rates and discarding up to 50% of the CRS rules; and (ii) improve robustness against adversarial SQLi attacks up to 85%, marking a significant stride toward designing more effective and robust WAFs. We release our open-source code at https://github.com/pralab/modsec-advlearn.
Authors: Shouwei Gao, Meiyan Gao, Yuepeng Li, Wenqian Dong
Abstract: The generation of synthetic tropical cyclone(TC) tracks for risk assessment is a critical application of preparedness for the impacts of climate change and disaster relief, particularly in North America. Insurance companies use these synthetic tracks to estimate the potential risks and financial impacts of future TCs. For governments and policymakers, understanding the potential impacts of TCs helps in developing effective emergency response strategies, updating building codes, and prioritizing investments in resilience and mitigation projects. In this study, many hypothetical but plausible TC scenarios are created based on historical TC data HURDAT2 (HURricane DATA 2nd generation). A hybrid methodology, combining the ARIMA and K-MEANS methods with Autoencoder, is employed to capture better historical TC behaviors and project future trajectories and intensities. It demonstrates an efficient and reliable in the field of climate modeling and risk assessment. By effectively capturing past hurricane patterns and providing detailed future projections, this approach not only validates the reliability of this method but also offers crucial insights for a range of applications, from disaster preparedness and emergency management to insurance risk analysis and policy formulation.
Authors: Oleksandr Balabanov, Hampus Linander
Abstract: Fine-tuning large language models can improve task specific performance, although a general understanding of what the fine-tuned model has learned, forgotten and how to trust its predictions is still missing. We derive principled uncertainty quantification for fine-tuned LLMs with posterior approximations using computationally efficient low-rank adaptation ensembles. We analyze three common multiple-choice datasets using low-rank adaptation ensembles based on Mistral-7b, and draw quantitative and qualitative conclusions on their perceived complexity and balance between retained prior knowledge and domain specific adaptation during and after fine-tuning. We identify unexpected retention of acquired knowledge during fine-tuning in the overfitting regime.
Authors: Benjamin Eysenbach, Vivek Myers, Ruslan Salakhutdinov, Sergey Levine
Abstract: Given time series data, how can we answer questions like "what will happen in the future?" and "how did we get here?" These sorts of probabilistic inference questions are challenging when observations are high-dimensional. In this paper, we show how these questions can have compact, closed form solutions in terms of learned representations. The key idea is to apply a variant of contrastive learning to time series data. Prior work already shows that the representations learned by contrastive learning encode a probability ratio. By extending prior work to show that the marginal distribution over representations is Gaussian, we can then prove that joint distribution of representations is also Gaussian. Taken together, these results show that representations learned via temporal contrastive learning follow a Gauss-Markov chain, a graphical model where inference (e.g., prediction, planning) over representations corresponds to inverting a low-dimensional matrix. In one special case, inferring intermediate representations will be equivalent to interpolating between the learned representations. We validate our theory using numerical simulations on tasks up to 46-dimensions.
Authors: Jo\~ao Manoel Herrera Pinheiro, Marcelo Becker
Abstract: Cancer is one of the diseases that kill the most women in the world, with breast cancer being responsible for the highest number of cancer cases and consequently deaths. However, it can be prevented by early detection and, consequently, early treatment. Any development for detection or perdition this kind of cancer is important for a better healthy life. Many studies focus on a model with high accuracy in cancer prediction, but sometimes accuracy alone may not always be a reliable metric. This study implies an investigative approach to studying the performance of different machine learning algorithms based on boosting to predict breast cancer focusing on the recall metric. Boosting machine learning algorithms has been proven to be an effective tool for detecting medical diseases. The dataset of the University of California, Irvine (UCI) repository has been utilized to train and test the model classifier that contains their attributes. The main objective of this study is to use state-of-the-art boosting algorithms such as AdaBoost, XGBoost, CatBoost and LightGBM to predict and diagnose breast cancer and to find the most effective metric regarding recall, ROC-AUC, and confusion matrix. Furthermore, our study is the first to use these four boosting algorithms with Optuna, a library for hyperparameter optimization, and the SHAP method to improve the interpretability of our model, which can be used as a support to identify and predict breast cancer. We were able to improve AUC or recall for all the models and reduce the False Negative for AdaBoost and LigthGBM the final AUC were more than 99.41\% for all models.
Authors: Andrew Holliday, Ahmed El-Geneidy, Gregory Dudek
Abstract: Planning a network of public transit routes is a challenging optimization problem. Metaheuristic algorithms search through the space of possible transit networks by applying heuristics that randomly alter routes in a network. The design of these heuristics has a major impact on the quality of the result. In this paper, we use deep reinforcement learning to train a graph neural net to provide heuristics for an evolutionary algorithm. These neural heuristics improve the algorithm's results on benchmark synthetic cities with 70 nodes or more, and achieve new state-of-the-art results on the challenging Mumford benchmark. They also improve upon a simulation of the real transit network in the city of Laval, Canada, by 52% and 25% on two key metrics, and offer cost savings of up to 19% over the city's existing transit network.
Authors: Minh Nguyen, Chandrajit Bajaj
Abstract: Reinforcement learning (RL) in continuous state-action spaces remains challenging in scientific computing due to poor sample efficiency and lack of pathwise physical consistency. We introduce Differential Reinforcement Learning (Differential RL), a novel framework that reformulates RL from a continuous-time control perspective via a differential dual formulation. This induces a Hamiltonian structure that embeds physics priors and ensures consistent trajectories without requiring explicit constraints. To implement Differential RL, we develop Differential Policy Optimization (DPO), a pointwise, stage-wise algorithm that refines local movement operators along the trajectory for improved sample efficiency and dynamic alignment. We establish pointwise convergence guarantees, a property not available in standard RL, and derive a competitive theoretical regret bound of $O(K^{5/6})$. Empirically, DPO outperforms standard RL baselines on representative scientific computing tasks, including surface modeling, grid control, and molecular dynamics, under low-data and physics-constrained conditions.
Authors: Han Zhong, Zikang Shan, Guhao Feng, Wei Xiong, Xinle Cheng, Li Zhao, Di He, Jiang Bian, Liwei Wang
Abstract: In the classical Reinforcement Learning from Human Feedback (RLHF) framework, Proximal Policy Optimization (PPO) is employed to learn from sparse, sentence-level rewards -- a challenging scenario in traditional deep reinforcement learning. Despite the great successes of PPO in the alignment of large language models, its open-source implementation is still largely sub-optimal. To address these issues, we introduce a framework that models RLHF problems as a Markov decision process (MDP), enabling the capture of fine-grained token-wise information. Under this framework, we introduce an algorithm Reinforced Token Optimization (\texttt{RTO}), which learns the token-wise reward function from preference data and performs policy optimization based on this learned token-wise reward signal. Theoretically, \texttt{RTO} is proven to have the capability of finding the near-optimal policy sample-efficiently. For its practical implementation, \texttt{RTO} innovatively integrates Direct Preference Optimization (DPO) and PPO. DPO, originally derived from sparse sentence rewards, surprisingly provides us with a token-wise characterization of response quality, which is seamlessly incorporated into our subsequent PPO training stage. Extensive experiments demonstrate that \texttt{RTO} performs better than PPO and other direct preference learning algorithms. In particular, RTO outperforms PPO by 7.5 points on the AlpacaEval 2 benchmark and by 4.1 points on Arena-Hard. Our code and models are available at \href{https://github.com/zkshan2002/RTO}{https://github.com/zkshan2002/RTO}.
URLs: https://github.com/zkshan2002/RTO, https://github.com/zkshan2002/RTO
Authors: Sander Dalm, Joshua Offergeld, Nasir Ahmad, Marcel van Gerven
Abstract: The backpropagation algorithm remains the dominant and most successful method for training deep neural networks (DNNs). At the same time, training DNNs at scale comes at a significant computational cost and therefore a high carbon footprint. Converging evidence suggests that input decorrelation may speed up deep learning. However, to date, this has not yet translated into substantial improvements in training efficiency in large-scale DNNs. This is mainly caused by the challenge of enforcing fast and stable network-wide decorrelation. Here, we show for the first time that much more efficient training of deep convolutional neural networks is feasible by embracing decorrelated backpropagation as a mechanism for learning. To achieve this goal we made use of a novel algorithm which induces network-wide input decorrelation using minimal computational overhead. By combining this algorithm with careful optimizations, we achieve a more than two-fold speed-up and higher test accuracy compared to backpropagation when training several deep networks up to a 50-layer ResNet model. This demonstrates that decorrelation provides exciting prospects for efficient deep learning at scale.
Authors: Christopher Amato
Abstract: Multi-agent reinforcement learning (MARL) has exploded in popularity in recent years. While numerous approaches have been developed, they can be broadly categorized into three main types: centralized training and execution (CTE), centralized training for decentralized execution (CTDE), and decentralized training and execution (DTE). CTE methods assume centralization during training and execution (e.g., with fast, free, and perfect communication) and have the most information during execution. CTDE methods are the most common, as they leverage centralized information during training while enabling decentralized execution -- using only information available to that agent during execution. Decentralized training and execution methods make the fewest assumptions and are often simple to implement. This text is an introduction to cooperative MARL -- MARL in which all agents share a single, joint reward. It is meant to explain the setting, basic concepts, and common methods for the CTE, CTDE, and DTE settings. It does not cover all work in cooperative MARL as the area is quite extensive. I have included work that I believe is important for understanding the main concepts in the area and apologize to those that I have omitted. Topics include simple applications of single-agent methods to CTE as well as some more scalable methods that exploit the multi-agent structure, independent Q-learning and policy gradient methods and their extensions, as well as value function factorization methods including the well-known VDN, QMIX, and QPLEX approaches, and centralized critic methods including MADDPG, COMA, and MAPPO. I also discuss common misconceptions, the relationship between different approaches, and some open questions.
Authors: Shinyoung Kang, Jihan Kim
Abstract: In this study, we explore the potential of using quantum natural language processing (QNLP) to inverse design metal-organic frameworks (MOFs) with targeted properties. Specifically, by analyzing 450 hypothetical MOF structures consisting of 3 topologies, 10 metal nodes and 15 organic ligands, we categorize these structures into four distinct classes for pore volume and $CO_{2}$ Henry's constant values. We then compare various QNLP models (i.e. the bag-of-words, DisCoCat (Distributional Compositional Categorical), and sequence-based models) to identify the most effective approach to process the MOF dataset. Using a classical simulator provided by the IBM Qiskit, the bag-of-words model is identified to be the optimum model, achieving validation accuracies of 88.6% and 78.0% for binary classification tasks on pore volume and $CO_{2}$ Henry's constant, respectively. Further, we developed multi-class classification models tailored to the probabilistic nature of quantum circuits, with average test accuracies of 92% and 80% across different classes for pore volume and $CO_{2}$ Henry's constant datasets. Finally, the performance of generating MOF with target properties showed accuracies of 93.5% for pore volume and 87% for $CO_{2}$ Henry's constant, respectively. Although our investigation covers only a fraction of the vast MOF search space, it marks a promising first step towards using quantum computing for materials design, offering a new perspective through which to explore the complex landscape of MOFs.
Authors: Peng Kuang, Zhibo Wang, Zhixuan Chu, Jingyi Wang, Kui Ren
Abstract: Spurious correlations in training data significantly hinder the generalization capability of machine learning models when faced with distribution shifts, leading to the proposition of numberous debiasing methods. However, it remains to be asked: \textit{Do existing benchmarks for debiasing really represent biases in the real world?} Recent works attempt to address such concerns by sampling from real-world data (instead of synthesizing) according to some predefined biased distributions to ensure the realism of individual samples. However, the realism of the biased distribution is more critical yet challenging and underexplored due to the complexity of real-world bias distributions. To tackle the problem, we propose a fine-grained framework for analyzing biased distributions, based on which we empirically and theoretically identify key characteristics of biased distributions in the real world that are poorly represented by existing benchmarks. Towards applicable debiasing in the real world, we further introduce two novel real-world-inspired biases to bridge this gap and build a systematic evaluation framework for real-world debiasing, RDBench\footnote{RDBench: Code to be released. Preliminary version in supplementary material for anonimized review.}. Furthermore, focusing on the practical setting of debiasing w/o bias label, we find real-world biases pose a novel \textit{Sparse bias capturing} challenge to the existing paradigm. We propose a simple yet effective approach named Debias in Destruction (DiD), to address the challenge, whose effectiveness is validated with extensive experiments on 8 datasets of various biased distributions.
Authors: Wenyu Wang, Zheyi Fan, Szu Hui Ng
Abstract: Training machine learning models inherently involves a resource-intensive and noisy iterative learning procedure that allows epoch-wise monitoring of the model performance. However, the insights gained from the iterative learning procedure typically remain underutilized in multi-objective hyperparameter optimization scenarios. Despite the limited research in this area, existing methods commonly identify the trade-offs only at the end of model training, overlooking the fact that trade-offs can emerge at earlier epochs in cases such as overfitting. To bridge this gap, we propose an enhanced multi-objective hyperparameter optimization problem that treats the number of training epochs as a decision variable, rather than merely an auxiliary parameter, to account for trade-offs at an earlier training stage. To solve this problem and accommodate its iterative learning, we then present a trajectory-based multi-objective Bayesian optimization algorithm characterized by two features: 1) a novel acquisition function that captures the improvement along the predictive trajectory of model performances over epochs for any hyperparameter setting and 2) a multi-objective early stopping mechanism that determines when to terminate the training to maximize epoch efficiency. Experiments on synthetic simulations and hyperparameter tuning benchmarks demonstrate that our algorithm can effectively identify the desirable trade-offs while improving tuning efficiency.
Authors: Alexander Denker, Francisco Vargas, Shreyas Padhy, Kieran Didi, Simon Mathis, Vincent Dutordoir, Riccardo Barbano, Emile Mathieu, Urszula Julia Komorowska, Pietro Lio
Abstract: Generative modelling paradigms based on denoising diffusion processes have emerged as a leading candidate for conditional sampling in inverse problems. In many real-world applications, we often have access to large, expensively trained unconditional diffusion models, which we aim to exploit for improving conditional sampling. Most recent approaches are motivated heuristically and lack a unifying framework, obscuring connections between them. Further, they often suffer from issues such as being very sensitive to hyperparameters, being expensive to train or needing access to weights hidden behind a closed API. In this work, we unify conditional training and sampling using the mathematically well-understood Doob's h-transform. This new perspective allows us to unify many existing methods under a common umbrella. Under this framework, we propose DEFT (Doob's h-transform Efficient FineTuning), a new approach for conditional generation that simply fine-tunes a very small network to quickly learn the conditional $h$-transform, while keeping the larger unconditional network unchanged. DEFT is much faster than existing baselines while achieving state-of-the-art performance across a variety of linear and non-linear benchmarks. On image reconstruction tasks, we achieve speedups of up to 1.6$\times$, while having the best perceptual quality on natural images and reconstruction performance on medical images. Further, we also provide initial experiments on protein motif scaffolding and outperform reconstruction guidance methods.
Authors: Spencer Young, Porter Jenkins, Longchao Da, Jeff Dotson, Hua Wei
Abstract: Neural networks that can produce accurate, input-conditional uncertainty representations are critical for real-world applications. Recent progress on heteroscedastic continuous regression has shown great promise for calibrated uncertainty quantification on complex tasks, like image regression. However, when these methods are applied to discrete regression tasks, such as crowd counting, ratings prediction, or inventory estimation, they tend to produce predictive distributions with numerous pathologies. Moreover, discrete models based on the Generalized Linear Model (GLM) framework either cannot process complex input or are not fully heterosedastic. To address these issues we propose the Deep Double Poisson Network (DDPN). In contrast to networks trained to minimize Gaussian negative log likelihood (NLL), discrete network parameterizations (i.e., Poisson, Negative binomial), and GLMs, DDPN can produce discrete predictive distributions of arbitrary flexibility. Additionally, we propose a technique to tune the prioritization of mean fit and probabilistic calibration during training. We show DDPN 1) vastly outperforms existing discrete models; 2) meets or exceeds the accuracy and flexibility of networks trained with Gaussian NLL; 3) produces proper predictive distributions over discrete counts; and 4) exhibits superior out-of-distribution detection. DDPN can easily be applied to a variety of count regression datasets including tabular, image, point cloud, and text data.
Authors: Davide Carbone (Laboratoire de Physique de l'Ecole Normale Sup\'erieure, ENS Universit\'e PSL, CNRS, Sorbonne Universit\'e, Universit\'e de Paris, Paris)
Abstract: Energy-Based Models have emerged as a powerful framework in the realm of generative modeling, offering a unique perspective that aligns closely with principles of statistical mechanics. This review aims to provide physicists with a comprehensive understanding of EBMs, delineating their connection to other generative models such as Generative Adversarial Networks, Variational Autoencoders, and Normalizing Flows. We explore the sampling techniques crucial for EBMs, including Markov Chain Monte Carlo (MCMC) methods, and draw parallels between EBM concepts and statistical mechanics, highlighting the significance of energy functions and partition functions. Furthermore, we delve into recent training methodologies for EBMs, covering recent advancements and their implications for enhanced model performance and efficiency. This review is designed to clarify the often complex interconnections between these models, which can be challenging due to the diverse communities working on the topic.
Authors: Phillip Mueller, Lars Mikelsons
Abstract: The synthesis of product design concepts stands at the crux of early-phase development processes for technical products, traditionally posing an intricate interdisciplinary challenge. The application of deep learning methods, particularly Deep Generative Models (DGMs), holds the promise of automating and streamlining manual iterations and therefore introducing heightened levels of innovation and efficiency. However, DGMs have yet to be widely adopted into the synthesis of product design concepts. This paper aims to explore the reasons behind this limited application and derive the requirements for successful integration of these technologies. We systematically analyze DGM-families (VAE, GAN, Diffusion, Transformer, Radiance Field), assessing their strengths, weaknesses, and general applicability for product design conception. Our objective is to provide insights that simplify the decision-making process for engineers, helping them determine which method might be most effective for their specific challenges. Recognizing the rapid evolution of this field, we hope that our analysis contributes to a fundamental understanding and guides practitioners towards the most promising approaches. This work seeks not only to illuminate current challenges but also to propose potential solutions, thereby offering a clear roadmap for leveraging DGMs in the realm of product design conception.
Authors: Aochuan Chen, Jiashun Cheng, Zijing Liu, Ziqi Gao, Fugee Tsung, Yu Li, Jia Li
Abstract: Low-Rank Adaptation (LoRA) has gained popularity for fine-tuning large foundation models, leveraging low-rank matrices $\mathbf{A}$ and $\mathbf{B}$ to represent weight changes (i.e., $\Delta \mathbf{W} = \mathbf{B} \mathbf{A}$). This method reduces trainable parameters and mitigates heavy memory consumption associated with full delta matrices by sequentially multiplying $\mathbf{A}$ and $\mathbf{B}$ with the activation. Despite its success, the intrinsic low-rank characteristic may limit its performance. Although several variants have been proposed to address this issue, they often overlook the crucial computational and memory efficiency brought by LoRA. In this paper, we propose Circular Convolution Adaptation (C$^3$A), which not only achieves high-rank adaptation with enhanced performance but also excels in both computational power and memory utilization. Extensive experiments demonstrate that C$^3$A consistently outperforms LoRA and its variants across various fine-tuning tasks.
Authors: Akhil Premkumar
Abstract: We explore the connection between deep learning and information theory through the paradigm of diffusion models. A diffusion model converts noise into structured data by reinstating, imperfectly, information that is erased when data was diffused to noise. This information is stored in a neural network during training. We quantify this information by introducing a measure called neural entropy, which is related to the total entropy produced by diffusion. Neural entropy is a function of not just the data distribution, but also the diffusive process itself. Measurements of neural entropy on a few simple image diffusion models reveal that they are extremely efficient at compressing large ensembles of structured data.
Authors: Haochen Yuan, Yutong Wang, Yihong Chen, Yunbo Wang, Xiaokang Yang
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: Gustav Wagner Zakarias, Lars Kai Hansen, Zheng-Hua Tan
Abstract: Models initialized from self-supervised pretraining may suffer from poor alignment with downstream tasks, reducing the extent to which subsequent fine-tuning can adapt pretrained features toward downstream objectives. To mitigate this, we introduce BiSSL, a novel bilevel training framework that enhances the alignment of self-supervised pretrained models with downstream tasks prior to fine-tuning. BiSSL acts as an intermediate training stage conducted after conventional self-supervised pretraining and is tasked with solving a bilevel optimization problem that incorporates the pretext and downstream training objectives in its lower- and upper-level objectives, respectively. This approach explicitly models the interdependence between the pretraining and fine-tuning stages within the conventional self-supervised learning pipeline, facilitating enhanced information sharing between them that ultimately leads to a model initialization better aligned with the downstream task. We propose a general training algorithm for BiSSL that is compatible with a broad range of pretext and downstream tasks. Using SimCLR and Bootstrap Your Own Latent to pretrain ResNet-50 backbones on the ImageNet dataset, we demonstrate that our proposed framework significantly improves accuracy on the vast majority of 12 downstream image classification datasets, as well as on object detection. Exploratory analyses alongside investigative experiments further provide compelling evidence that BiSSL enhances downstream alignment.
Authors: Huayu Deng, Xiangming Zhu, Yunbo Wang, Xiaokang Yang
Abstract: Graph neural networks have been a powerful tool for mesh-based physical simulation. To efficiently model large-scale systems, existing methods mainly employ hierarchical graph structures to capture multi-scale node relations. However, these graph hierarchies are typically manually designed and fixed, limiting their ability to adapt to the evolving dynamics of complex physical systems. We propose EvoMesh, a fully differentiable framework that jointly learns graph hierarchies and physical dynamics, adaptively guided by physical inputs. EvoMesh introduces anisotropic message passing, which enables direction-specific aggregation of dynamic features between nodes within each hierarchy, while simultaneously learning node selection probabilities for the next hierarchical level based on physical context. This design creates more flexible message shortcuts and enhances the model's capacity to capture long-range dependencies. Extensive experiments on five benchmark physical simulation datasets show that EvoMesh outperforms recent fixed-hierarchy message passing networks by large margins. The project page is available at https://hbell99.github.io/evo-mesh/.
Authors: Yuhang Yao, Yuan Li, Xinyi Fan, Junhao Li, Kay Liu, Weizhao Jin, Yu Yang, Srivatsan Ravi, Philip S. Yu, Carlee Joe-Wong
Abstract: Federated graph learning is an emerging field with significant practical challenges. While algorithms have been proposed to improve the accuracy of training graph neural networks, such as node classification on federated graphs, the system performance is often overlooked, despite it is crucial for real-world deployment. To bridge this gap, we introduce FedGraph, a research library designed for practical distributed training and comprehensive benchmarking of FGL algorithms. FedGraph supports a range of state-of-the-art graph learning methods and includes a monitoring class that evaluates system performance, with a particular focus on communication and computation costs during training. Unlike existing federated learning platforms, FedGraph natively integrates homomorphic encryption to enhance privacy preservation and supports scalable deployment across multiple physical machines with system-level performance evaluation to guide the system design of future algorithms. To enhance efficiency and privacy, we propose a low-rank communication scheme for algorithms like FedGCN that require pre-training communication, accelerating both the pre-training and training phases. Extensive experiments benchmark different FGL algorithms on three major graph learning tasks and demonstrate FedGraph as the first efficient FGL framework to support encrypted low-rank communication and scale to graphs with 100 million nodes.
Authors: Michael Lan, Philip Torr, Austin Meek, Ashkan Khakzar, David Krueger, Fazl Barez
Abstract: The Universality Hypothesis in large language models (LLMs) claims that different models converge towards similar concept representations in their latent spaces. Providing evidence for this hypothesis would enable researchers to exploit universal properties, facilitating the generalization of mechanistic interpretability techniques across models. Previous works studied if LLMs learned the same features, which are internal representations that activate on specific concepts. Since comparing features across LLMs is challenging due to polysemanticity, in which LLM neurons often correspond to multiple unrelated features rather than to distinct concepts, sparse autoencoders (SAEs) have been employed to disentangle LLM neurons into SAE features corresponding to distinct concepts. In this paper, we introduce a new variation of the universality hypothesis called Analogous Feature Universality: we hypothesize that even if SAEs across different models learn different feature representations, the spaces spanned by SAE features are similar, such that one SAE space is similar to another SAE space under rotation-invariant transformations. Evidence for this hypothesis would imply that interpretability techniques related to latent spaces, such as steering vectors, may be transferred across models via certain transformations. To investigate this hypothesis, we first pair SAE features across different models via activation correlation, and then measure spatial relation similarities between paired features via representational similarity measures, which transform spaces into representations that reveal hidden relational similarities. Our experiments demonstrate high similarities for SAE feature spaces across various LLMs, providing evidence for feature space universality.
Authors: Fabian Paischer, Lukas Hauzenberger, Thomas Schmied, Benedikt Alkin, Marc Peter Deisenroth, Sepp Hochreiter
Abstract: Foundation models (FMs) are pre-trained on large-scale datasets and then fine-tuned for a specific downstream task. The most common fine-tuning method is to update pretrained weights via low-rank adaptation (LoRA). Existing initialization strategies for LoRA often rely on singular value decompositions (SVD) of gradients or weight matrices. However, they do not provably maximize the expected gradient signal, which is critical for fast adaptation. To this end, we introduce Explained Variance Adaptation (EVA), an initialization scheme that uses the directions capturing the most activation variance, provably maximizing the expected gradient signal and accelerating fine-tuning. EVA performs incremental SVD on minibatches of activation vectors and selects the right-singular vectors for initialization once they converged. Further, by selecting the directions that capture the most activation-variance for a given rank budget, EVA accommodates adaptive ranks that reduce the number of trainable parameters, while maintaining or improving downstream performance. We apply EVA to a variety of fine-tuning tasks as language generation and understanding, image classification, and reinforcement learning. EVA exhibits faster convergence than competitors and achieves the highest average score across a multitude of tasks per domain while reducing the number of trainable parameters through rank redistribution.
Authors: Satoki Ishikawa, Rio Yokota, Ryo Karakida
Abstract: Local learning, which trains a network through layer-wise local targets and losses, has been studied as an alternative to backpropagation (BP) in neural computation. However, its algorithms often become more complex or require additional hyperparameters because of the locality, making it challenging to identify desirable settings in which the algorithm progresses in a stable manner. To provide theoretical and quantitative insights, we introduce the maximal update parameterization ($\mu$P) in the infinite-width limit for two representative designs of local targets: predictive coding (PC) and target propagation (TP). We verified that $\mu$P enables hyperparameter transfer across models of different widths. Furthermore, our analysis revealed unique and intriguing properties of $\mu$P that are not present in conventional BP. By analyzing deep linear networks, we found that PC's gradients interpolate between first-order and Gauss-Newton-like gradients, depending on the parameterization. We demonstrate that, in specific standard settings, PC in the infinite-width limit behaves more similarly to the first-order gradient. For TP, even with the standard scaling of the last layer, which differs from classical $\mu$P, its local loss optimization favors the feature learning regime over the kernel regime.
Authors: Santiago Acevedo, Alex Rodriguez, Alessandro Laio
Abstract: In real-world data, information is stored in extremely large feature vectors. These variables are typically correlated due to complex interactions involving many features simultaneously. Such correlations qualitatively correspond to semantic roles and are naturally recognized by both the human brain and artificial neural networks. This recognition enables, for instance, the prediction of missing parts of an image or text based on their context. We present a method to detect these correlations in high-dimensional data represented as binary numbers. We estimate the binary intrinsic dimension of a dataset, which quantifies the minimum number of independent coordinates needed to describe the data, and is therefore a proxy of semantic complexity. The proposed algorithm is largely insensitive to the so-called curse of dimensionality, and can therefore be used in big data analysis. We test this approach identifying phase transitions in model magnetic systems and we then apply it to the detection of semantic correlations of images and text inside deep neural networks.
Authors: Gautham Vasan, Mohamed Elsayed, Alireza Azimi, Jiamin He, Fahim Shariar, Colin Bellinger, Martha White, A. Rupam Mahmood
Abstract: Modern deep policy gradient methods achieve effective performance on simulated robotic tasks, but they all require large replay buffers or expensive batch updates, or both, making them incompatible for real systems with resource-limited computers. We show that these methods fail catastrophically when limited to small replay buffers or during incremental learning, where updates only use the most recent sample without batch updates or a replay buffer. We propose a novel incremental deep policy gradient method -- Action Value Gradient (AVG) and a set of normalization and scaling techniques to address the challenges of instability in incremental learning. On robotic simulation benchmarks, we show that AVG is the only incremental method that learns effectively, often achieving final performance comparable to batch policy gradient methods. This advancement enabled us to show for the first time effective deep reinforcement learning with real robots using only incremental updates, employing a robotic manipulator and a mobile robot.
Authors: Emile Anand, Ishani Karmarkar, Guannan Qu
Abstract: Designing efficient algorithms for multi-agent reinforcement learning (MARL) is fundamentally challenging because the size of the joint state and action spaces grows exponentially in the number of agents. These difficulties are exacerbated when balancing sequential global decision-making with local agent interactions. In this work, we propose a new algorithm $\texttt{SUBSAMPLE-MFQ}$ ($\textbf{Subsample}$-$\textbf{M}$ean-$\textbf{F}$ield-$\textbf{Q}$-learning) and a decentralized randomized policy for a system with $n$ agents. For any $k\leq n$, our algorithm learns a policy for the system in time polynomial in $k$. We prove that this learned policy converges to the optimal policy on the order of $\tilde{O}(1/\sqrt{k})$ as the number of subsampled agents $k$ increases. In particular, this bound is independent of the number of agents $n$.
Authors: Yongchao Ye, Xinting Zhu, Xuejin Shen, Xiaoyu Chen, Lishuai Li, S. Joe Qin
Abstract: Satellite pattern-of-life (PoL) identification is crucial for space safety and satellite monitoring, involving the analysis of typical satellite behaviors such as station-keeping, drift, etc. However, existing PoL identification methods remain underdeveloped due to the complexity of aerospace systems, variability in satellite behaviors, and fluctuating observation sampling rates. In a first attempt, we developed a domain expertise-informed machine learning method (Expert-ML) to combine satellite orbital movement knowledge and machine learning models. The Expert-ML method achieved high accuracy results in simulation data and real-world data with normal sampling rate. However, this approach lacks of generality as it requires domain expertise and its performance degraded significantly when data sampling rate varied. To achieve generality, we propose a novel diffusion-based PoL identification method. Distinct from prior approaches, the proposed method leverages a diffusion model to achieve end-to-end identification without manual refinement or domain-specific knowledge. Specifically, we employ a multivariate time-series encoder to capture hidden representations of satellite positional data. The encoded features are subsequently incorporated as conditional information in the denoising process to generate PoL labels. Through experimentation across real-world satellite settings, our proposed diffusion-based method demonstrates its high identification quality and provides a robust solution even with reduced data sampling rates, indicating its great potential in practical satellite behavior pattern identification, tracking and related mission deployment.
Authors: Qi Liu, Wanjing Ma
Abstract: Data corruption, including missing and noisy data, poses significant challenges in real-world machine learning. This study investigates the effects of data corruption on model performance and explores strategies to mitigate these effects through two experimental setups: supervised learning with NLP tasks (NLP-SL) and deep reinforcement learning for traffic signal optimization (Signal-RL). We analyze the relationship between data corruption levels and model performance, evaluate the effectiveness of data imputation methods, and assess the utility of enlarging datasets to address data corruption. Our results show that model performance under data corruption follows a diminishing return curve, modeled by the exponential function. Missing data, while detrimental, is less harmful than noisy data, which causes severe performance degradation and training instability, particularly in sequential decision-making tasks like Signal-RL. Imputation strategies involve a trade-off: they recover missing information but may introduce noise. Their effectiveness depends on imputation accuracy and corruption ratio. We identify distinct regions in the imputation advantage heatmap, including an "imputation advantageous corner" and an "imputation disadvantageous edge" and classify tasks as "noise-sensitive" or "noise-insensitive" based on their decision boundaries. Furthermore, we find that increasing dataset size mitigates but cannot fully overcome the effects of data corruption. The marginal utility of additional data diminishes as corruption increases. An empirical rule emerges: approximately 30% of the data is critical for determining performance, while the remaining 70% has minimal impact. These findings provide actionable insights into data preprocessing, imputation strategies, and data collection practices, guiding the development of robust machine learning systems in noisy environments.
Authors: Chanon Puttanawarut, Romen Samuel Wabina, Nat Sirirutbunkajorn
Abstract: Background: Radiation pneumonitis is a side effect of thoracic radiation therapy. Recently, machine learning models with radiomic features have improved radiation pneumonitis prediction by capturing spatial information. To further support clinical decision-making, this study explores the role of post hoc uncertainty quantification methods in enhancing model uncertainty estimate. Methods: We retrospectively analyzed a cohort of 101 esophageal cancer patients. This study evaluated four machine learning models: logistic regression, support vector machines, extreme gradient boosting, and random forest, using 15 dosimetric, 79 dosiomic, and 237 radiomic features to predict radiation pneumonitis. We applied uncertainty quantification methods, including Platt scaling, isotonic regression, Venn-ABERS predictor, and conformal prediction, to quantify uncertainty. Model performance was assessed through an area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), and adaptive calibration error using leave-one-out cross-validation. Results: Highest AUROC is achieved by the logistic regression model with the conformal prediction method (AUROC 0.75+-0.01, AUPRC 0.74+-0.01) at a certainty cut point of 0.8. Highest AUPRC of 0.82+-0.02 (with AUROC of 0.67+-0.04) achieved by The extreme gradient boosting model with conformal prediction at the 0.9 certainty threshold. Radiomic and dosiomic features improve both discriminative and calibration performance. Conclusions: Integrating uncertainty quantification into machine learning models with radiomic and dosiomic features may improve both predictive accuracy and calibration, supporting more reliable clinical decision-making. The findings emphasize the value of uncertainty quantification methods in enhancing applicability of predictive models for radiation pneumonitis in healthcare settings.
Authors: Ming Li, Chenyi Zhang, Qin Li
Abstract: Monotone learning describes learning processes in which expected performance consistently improves as the amount of training data increases. However, recent studies challenge this conventional wisdom, revealing significant gaps in the understanding of generalization in machine learning. Addressing these gaps is crucial for advancing the theoretical foundations of the field. In this work, we utilize Probably Approximately Correct (PAC) learning theory to construct a theoretical risk distribution that approximates a learning algorithm's actual performance. We rigorously prove that this theoretical distribution exhibits monotonicity as sample sizes increase. We identify two scenarios under which deterministic algorithms based on Empirical Risk Minimization (ERM) are monotone: (1) the hypothesis space is finite, or (2) the hypothesis space has finite VC-dimension. Experiments on two classical learning problems validate our findings by demonstrating that the monotonicity of the algorithms' generalization error is guaranteed, as its theoretical risk upper bound monotonically converges to 0.
Authors: Xingsi Dong, Xiangyuan Peng, Si Wu
Abstract: Predictive models are highly advanced in understanding the mechanisms of brain function. Recent advances in machine learning further underscore the power of prediction for optimal representation in learning. However, there remains a gap in creating a biologically plausible model that explains how the neural system achieves prediction. In this paper, we introduce a framework that employs an energy-based model (EBM) to capture the nuanced processes of predicting observation after action within the neural system, encompassing prediction, learning, and inference. We implement the EBM with a hierarchical structure and integrate a continuous attractor neural network for memory, constructing a biologically plausible model. In experimental evaluations, our model demonstrates efficacy across diverse scenarios. The range of actions includes eye movement, motion in environments, head turning, and static observation while the environment changes. Our model not only makes accurate predictions for environments it was trained on, but also provides reasonable predictions for unseen environments, matching the performances of machine learning methods in multiple tasks. We hope that this study contributes to a deep understanding of how the neural system performs prediction.
Authors: Haifeng Wen, Hong Xing, Osvaldo Simeone
Abstract: Post-hoc calibration of pre-trained models is critical for ensuring reliable inference, especially in safety-critical domains such as healthcare. Conformal Prediction (CP) offers a robust post-hoc calibration framework, providing distribution-free statistical coverage guarantees for prediction sets by leveraging held-out datasets. In this work, we address a decentralized setting where each device has limited calibration data and can communicate only with its neighbors over an arbitrary graph topology. We propose two message-passing-based approaches for achieving reliable inference via CP: quantile-based distributed conformal prediction (Q-DCP) and histogram-based distributed conformal prediction (H-DCP). Q-DCP employs distributed quantile regression enhanced with tailored smoothing and regularization terms to accelerate convergence, while H-DCP uses a consensus-based histogram estimation approach. Through extensive experiments, we investigate the trade-offs between hyperparameter tuning requirements, communication overhead, coverage guarantees, and prediction set sizes across different network topologies. The code of our work is released on: https://github.com/HaifengWen/Distributed-Conformal-Prediction.
URLs: https://github.com/HaifengWen/Distributed-Conformal-Prediction.
Authors: Agathe Fernandes Machado, Arthur Charpentier, Ewen Gallic
Abstract: Recently, optimal transport-based approaches have gained attention for deriving counterfactuals, e.g., to quantify algorithmic discrimination. However, in the general multivariate setting, these methods are often opaque and difficult to interpret. To address this, alternative methodologies have been proposed, using causal graphs combined with iterative quantile regressions (Ple\v{c}ko and Meinshausen (2020)) or sequential transport (Fernandes Machado et al. (2025)) to examine fairness at the individual level, often referred to as ``counterfactual fairness.'' Despite these advancements, transporting categorical variables remains a significant challenge in practical applications with real datasets. In this paper, we propose a novel approach to address this issue. Our method involves (1) converting categorical variables into compositional data and (2) transporting these compositions within the probabilistic simplex of $\mathbb{R}^d$. We demonstrate the applicability and effectiveness of this approach through an illustration on real-world data, and discuss limitations.
Authors: Federico Errica, Henrik Christiansen, Viktor Zaverkin, Mathias Niepert, Francesco Alesiani
Abstract: For almost 70 years, researchers have mostly relied on hyper-parameter tuning to select the width of neural networks' layers. This paper challenges the status quo by introducing an easy-to-use technique to learn an unbounded width of a neural network's layer during training. The technique does not rely on alternate optimization nor hand-crafted gradient heuristics; rather, it jointly optimizes the width and the parameters of each layer via simple backpropagation. We apply the technique to a broad range of data domains such as tables, images, text, sequences, and graphs, showing how the width adapts to the task's difficulty. The method imposes a soft ordering of importance among neurons, by which it also is possible to truncate the trained network at virtually zero cost, achieving a smooth trade-off between performance and compute resources in a structured way. Alternatively, one can dynamically compress the network with no performance degradation. In light of recent foundation models trained on large datasets, believed to require billions of parameters and where hyper-parameter tuning is unfeasible due to humongous training costs, our approach stands as a viable alternative for width learning.
Authors: Hossein Zakerinia, Dorsa Ghobadi, Christoph H. Lampert
Abstract: Deep learning methods are known to generalize well from training to future data, even in an overparametrized regime, where they could easily overfit. One explanation for this phenomenon is that even when their *ambient dimensionality*, (i.e. the number of parameters) is large, the models' *intrinsic dimensionality* is small; specifically, their learning takes place in a small subspace of all possible weight configurations. In this work, we confirm this phenomenon in the setting of *deep multi-task learning*. We introduce a method to parametrize multi-task network directly in the low-dimensional space, facilitated by the use of *random expansions* techniques. We then show that high-accuracy multi-task solutions can be found with much smaller intrinsic dimensionality (fewer free parameters) than what single-task learning requires. Subsequently, we show that the low-dimensional representations in combination with *weight compression* and *PAC-Bayesian* reasoning lead to the *first non-vacuous generalization bounds* for deep multi-task networks.
Authors: Hyeong Kyu Choi, Maxim Khanov, Hongxin Wei, Yixuan Li
Abstract: Dataset contamination, where evaluation datasets overlap with pre-training corpora, inflates performance metrics and undermines the reliability of model evaluations. Measuring dataset contamination thus becomes essential to ensure that performance evaluations genuinely reflect a model's ability to generalize to unseen data, rather than relying on memorized examples. To address this problem, we propose Kernel Divergence Score (KDS), a novel method that evaluates dataset contamination by computing the divergence between the kernel similarity matrix of sample embeddings, before and after fine-tuning on the benchmark dataset. Leveraging the insight that fine-tuning affects unseen samples more significantly than seen ones, KDS provides a reliable measure of contamination. Through extensive experiments on controlled contamination scenarios, KDS demonstrates a near-perfect correlation with contamination levels and outperforms existing baselines. Additionally, we perform comprehensive ablation studies to analyze the impact of key design choices, providing deeper insights into the components and effectiveness of KDS. These ablations highlight the importance of leveraging fine-grained kernel-based information and confirm the reliability of the proposed framework across diverse datasets and settings. Code is released in https://github.com/deeplearning-wisc/kernel-divergence-score.
URLs: https://github.com/deeplearning-wisc/kernel-divergence-score.
Authors: Dongwon Jo, Jiwon Song, Yulhwa Kim, Jae-Joon Kim
Abstract: While large language models (LLMs) excel at handling long-context sequences, they require substantial key-value (KV) caches to store contextual information, which can heavily burden computational efficiency and memory usage. Previous efforts to compress these KV caches primarily focused on reducing memory demands but were limited in enhancing latency. To address this issue, we introduce FastKV, a KV cache compression method designed to reduce latency for long-context inference. FastKV improves processing speed while preserving accuracy by adopting Token-Selective Propagation (TSP). This approach preserves full-context information in early layers of LLMs and selectively propagates only a portion of this information in later layers. This design enables FastKV to minimize redundant computation without sacrificing contextual fidelity. Our experimental results show that FastKV achieves up to 1.97$\times$ and 4.82$\times$ improvements in time-to-first-token (TTFT) and throughput, respectively, compared to baseline without KV cache compression. Moreover, FastKV successfully maintains accuracy within 1\% of the baseline on long-context benchmarks. Our code is available at https://github.com/dongwonjo/FastKV.
Authors: Zeyu Wang, Yao-Hui Li, Xin Li, Hongyu Zang, Romain Laroche, Riashat Islam
Abstract: Multi-View Reinforcement Learning (MVRL) seeks to provide agents with multi-view observations, enabling them to perceive environment with greater effectiveness and precision. Recent advancements in MVRL focus on extracting latent representations from multiview observations and leveraging them in control tasks. However, it is not straightforward to learn compact and task-relevant representations, particularly in the presence of redundancy, distracting information, or missing views. In this paper, we propose Multi-view Fusion State for Control (MFSC), firstly incorporating bisimulation metric learning into MVRL to learn task-relevant representations. Furthermore, we propose a multiview-based mask and latent reconstruction auxiliary task that exploits shared information across views and improves MFSC's robustness in missing views by introducing a mask token. Extensive experimental results demonstrate that our method outperforms existing approaches in MVRL tasks. Even in more realistic scenarios with interference or missing views, MFSC consistently maintains high performance.
Authors: David Rodr\'iguez-V\'itores (UVa, IMUVA), Cl\'ement Lalanne (IMT, ANITI), Jean-Michel Loubes (IMT, ANITI)
Abstract: In this work, we introduce a novel framework for privately optimizing objectives that rely on Wasserstein distances between data-dependent empirical measures. Our main theoretical contribution is, based on an explicit formulation of the Wasserstein gradient in a fully discrete setting, a control on the sensitivity of this gradient to individual data points, allowing strong privacy guarantees at minimal utility cost. Building on these insights, we develop a deep learning approach that incorporates gradient and activations clipping, originally designed for DP training of problems with a finite-sum structure. We further demonstrate that privacy accounting methods extend to Wasserstein-based objectives, facilitating large-scale private training. Empirical results confirm that our framework effectively balances accuracy and privacy, offering a theoretically sound solution for privacy-preserving machine learning tasks relying on optimal transport distances such as Wasserstein distance or sliced-Wasserstein distance.
Authors: Phil Ostheimer, Mayank Nagda, Jean Radig, Carl Herrmann, Stephan Mandt, Marius Kloft, Sophie Fellenz
Abstract: Sparse data is ubiquitous, appearing in numerous domains, from economics and recommender systems to astronomy and biomedical sciences. However, efficiently generating high-fidelity synthetic sparse data remains a significant challenge. We introduce Sparse Data Diffusion (SDD), a novel method for generating sparse data. SDD extends continuous state-space diffusion models with an explicit representation of exact zeros by modeling sparsity through the introduction of Sparsity Bits. Empirical validation in various domains, including two scientific applications in physics and biology, demonstrates that SDD achieves high fidelity in representing data sparsity while preserving the quality of the generated data.
Authors: Tom Hoddes, Alex Bijamov, Saket Joshi, Daniel Roggen, Ali Etemad, Robert Harle, David Racz
Abstract: Many deep architectures and self-supervised pre-training techniques have been proposed for human activity recognition (HAR) from wearable multimodal sensors. Scaling laws have the potential to help move towards more principled design by linking model capacity with pre-training data volume. Yet, scaling laws have not been established for HAR to the same extent as in language and vision. By conducting an exhaustive grid search on both amount of pre-training data and Transformer architectures, we establish the first known scaling laws for HAR. We show that pre-training loss scales with a power law relationship to amount of data and parameter count and that increasing the number of users in a dataset results in a steeper improvement in performance than increasing data per user, indicating that diversity of pre-training data is important, which contrasts to some previously reported findings in self-supervised HAR. We show that these scaling laws translate to downstream performance improvements on three HAR benchmark datasets of postures, modes of locomotion and activities of daily living: UCI HAR and WISDM Phone and WISDM Watch. Finally, we suggest some previously published works should be revisited in light of these scaling laws with more adequate model capacities.
Authors: Fedor Borisyuk, Lars Hertel, Ganesh Parameswaran, Gaurav Srivastava, Sudarshan Srinivasa Ramanujam, Borja Ocejo, Peng Du, Andrei Akterskii, Neil Daftary, Shao Tang, Daqi Sun, Qiang Charles Xiao, Deepesh Nathani, Mohit Kothari, Yun Dai, Guoyao Li, Aman Gupta
Abstract: We present LiGR, a large-scale ranking framework developed at LinkedIn that brings state-of-the-art transformer-based modeling architectures into production. We introduce a modified transformer architecture that incorporates learned normalization and simultaneous set-wise attention to user history and ranked items. This architecture enables several breakthrough achievements, including: (1) the deprecation of most manually designed feature engineering, outperforming the prior state-of-the-art system using only few features (compared to hundreds in the baseline), (2) validation of the scaling law for ranking systems, showing improved performance with larger models, more training data, and longer context sequences, and (3) simultaneous joint scoring of items in a set-wise manner, leading to automated improvements in diversity. To enable efficient serving of large ranking models, we describe techniques to scale inference effectively using single-pass processing of user history and set-wise attention. We also summarize key insights from various ablation studies and A/B tests, highlighting the most impactful technical approaches.
Authors: Indrashis Das, Mahmoud Safari, Steven Adriaensen, Frank Hutter
Abstract: Activation functions are fundamental elements of deep learning architectures as they significantly influence training dynamics. ReLU, while widely used, is prone to the dying neuron problem, which has been mitigated by variants such as LeakyReLU, PReLU, and ELU that better handle negative neuron outputs. Recently, self-gated activations like GELU and Swish have emerged as state-of-the-art alternatives, leveraging their smoothness to ensure stable gradient flow and prevent neuron inactivity. In this work, we introduce the Gompertz Linear Unit (GoLU), a novel self-gated activation function defined as $\mathrm{GoLU}(x) = x \, \mathrm{Gompertz}(x)$, where $\mathrm{Gompertz}(x) = e^{-e^{-x}}$. The GoLU activation leverages the right-skewed asymmetry in the Gompertz function to reduce variance in the latent space more effectively compared to GELU and Swish, while preserving robust gradient flow. Extensive experiments across diverse tasks, including Image Classification, Language Modeling, Semantic Segmentation, Object Detection, Instance Segmentation, and Diffusion, highlight GoLU's superior performance relative to state-of-the-art activation functions, establishing GoLU as a robust alternative to existing activation functions.
Authors: Kaicheng Zhang, Piero Deidda, Desmond Higham, Francesco Tudisco
Abstract: Oversmoothing is a fundamental challenge in graph neural networks (GNNs): as the number of layers increases, node embeddings become increasingly similar, and model performance drops sharply. Traditionally, oversmoothing has been quantified using metrics that measure the similarity of neighbouring node features, such as the Dirichlet energy. While these metrics are related to oversmoothing, we argue they have critical limitations and fail to reliably capture oversmoothing in realistic scenarios. For instance, they provide meaningful insights only for very deep networks and under somewhat strict conditions on the norm of network weights and feature representations. As an alternative, we propose measuring oversmoothing by examining the numerical or effective rank of the feature representations. We provide theoretical support for this approach, demonstrating that the numerical rank of feature representations converges to one for a broad family of nonlinear activation functions under the assumption of nonnegative trained weights. To the best of our knowledge, this is the first result that proves the occurrence of oversmoothing in the nonlinear setting without assumptions on the boundedness of the weight matrices. Along with the theoretical findings, we provide extensive numerical evaluation across diverse graph architectures. Our results show that rank-based metrics consistently capture oversmoothing, whereas energy-based metrics often fail. Notably, we reveal that a significant drop in the rank aligns closely with performance degradation, even in scenarios where energy metrics remain unchanged.
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 their expressiveness and suffer from effective 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 directed link prediction performance. To address limitations in current benchmark setups, we introduce DirLinkBench, a robust new benchmark with comprehensive coverage, standardized evaluation, and modular extensibility. The results on DirLinkBench show that current methods struggle to achieve strong performance, while DiGAE outperforms other baselines 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 state-of-the-art average performance on DirLinkBench. Finally, we analyze key factors influencing directed link prediction and highlight open challenges in this field.
Authors: Muhammad Umair Haider, Hammad Rizwan, Hassan Sajjad, Peizhong Ju, A. B. Siddique
Abstract: Interpreting the internal mechanisms of large language models (LLMs) is crucial for improving their trustworthiness and utility. Prior work has primarily focused on mapping individual neurons to discrete semantic concepts. However, such mappings struggle to handle the inherent polysemanticity in LLMs, where individual neurons encode multiple, distinct concepts. Through a comprehensive analysis of both encoder and decoder-based LLMs across diverse datasets, we observe that even highly salient neurons, identified via various attribution techniques for specific semantic concepts, consistently exhibit polysemantic behavior. Importantly, activation magnitudes for fine-grained concepts follow distinct, often Gaussian-like distributions with minimal overlap. This observation motivates a shift from neuron attribution to range-based interpretation. We hypothesize that interpreting and manipulating neuron activation ranges would enable more precise interpretability and targeted interventions in LLMs. To validate our hypothesis, we introduce NeuronLens, a novel range-based interpretation and manipulation framework that provides a finer view of neuron activation distributions to localize concept attribution within a neuron. Extensive empirical evaluations demonstrate that NeuronLens significantly reduces unintended interference, while maintaining precise manipulation of targeted concepts, outperforming neuron attribution.
Authors: Ziping Xu, Hinal Jajal, Sung Won Choi, Inbal Nahum-Shani, Guy Shani, Alexandra M. Psihogios, Pei-Yao Hung, Susan Murphy
Abstract: Medication adherence is critical for the recovery of adolescents and young adults (AYAs) who have undergone hematopoietic cell transplantation (HCT). However, maintaining adherence is challenging for AYAs after hospital discharge, who experience both individual (e.g. physical and emotional symptoms) and interpersonal barriers (e.g., relational difficulties with their care partner, who is often involved in medication management). To optimize the effectiveness of a three-component digital intervention targeting both members of the dyad as well as their relationship, we propose a novel Multi-Agent Reinforcement Learning (MARL) approach to personalize the delivery of interventions. By incorporating the domain knowledge, the MARL framework, where each agent is responsible for the delivery of one intervention component, allows for faster learning compared with a flattened agent. Evaluation using a dyadic simulator environment, based on real clinical data, shows a significant improvement in medication adherence (approximately 3%) compared to purely random intervention delivery. The effectiveness of this approach will be further evaluated in an upcoming trial.
Authors: Dongkyu Cho, Taesup Moon, Rumi Chunara, Kyunghyun Cho, Sungmin Cha
Abstract: Continual learning (CL) has traditionally emphasized minimizing exemplar memory usage, assuming that memory is the primary bottleneck. However, in modern computing environments-particularly those involving large foundation models-memory is inexpensive and abundant, while GPU time constitutes the main cost. This paper re-examines CL under a more realistic setting with sufficient exemplar memory, where the system can retain a representative portion of past data. We find that, under this regime, stability improves due to reduced forgetting, but plasticity diminishes as the model becomes biased toward prior tasks and struggles to adapt to new ones. Notably, even simple baselines like naive replay can match or exceed the performance of state-of-the-art methods at a fraction of the computational cost. Building on this insight, we propose a lightweight yet effective method called Weight Space Consolidation, which directly operates in the model's weight space via two core mechanisms: (1) rank-based parameter resets to recover plasticity, and (2) weight averaging to enhance stability. Our approach outperforms strong baselines across class-incremental learning with image classifiers and continual instruction tuning with large language models, while requiring only one-third to one-fourth of the training cost. These findings challenge long-standing CL assumptions and establish a new, cost-efficient baseline for real-world continual learning systems where exemplar memory is no longer the limiting factor.
Authors: Qingyuan Wu, Jianheng Liu, Jianye Hao, Jun Wang, Kun Shao
Abstract: State-of-the-art (SOTA) reinforcement learning (RL) methods have enabled vision-language model (VLM) agents to learn from interaction with online environments without human supervision. However, these methods often struggle with learning inefficiencies when applied to complex, real-world decision-making tasks with sparse rewards and long-horizon dependencies. We propose a novel framework, Variational Subgoal-Conditioned Reinforcement Learning (VSC-RL), advancing the VLM agents in resolving challenging decision-making tasks. Fundamentally distinct from existing methods, VSC-RL reformulates the decision-making problem as a variational subgoal-conditioned RL problem with the newly derived optimization objective, Subgoal Evidence Lower BOund (SGC-ELBO), which comprises two key components: (a) maximizing the subgoal-conditioned return, and (b) minimizing the divergence from a reference goal-conditioned policy. We theoretically and empirically demonstrate that the VSC-RL can efficiently improve the learning efficiency without compromising performance guarantees. Across a diverse set of challenging benchmarks, including mobile device and web control tasks, VSC-RL consistently outperforms existing SOTA methods, achieving superior learning efficiency and performance.
Authors: Abdelkrim Alahyane (LAAS-SARA, LAAS), C\'eline Comte (CNRS, LAAS-SARA, LAAS), Matthieu Jonckheere (CNRS, LAAS-SARA, LAAS), \'Eric Moulines (X)
Abstract: Synchronous federated learning (FL) scales poorly with the number of clients due to the straggler effect. Algorithms like FedAsync and GeneralizedFedAsync address this limitation by enabling asynchronous communication between clients and the central server. In this work, we rely on stochastic modeling and analysis to better understand the impact of design choices in asynchronous FL algorithms, such as the concurrency level and routing probabilities, and we leverage this knowledge to optimize loss. Compared to most existing studies, we account for the joint impact of heterogeneous and variable service speeds and heterogeneous datasets at the clients. We characterize in particular a fundamental trade-off for optimizing asynchronous FL: minimizing gradient estimation errors by avoiding model parameter staleness, while also speeding up the system by increasing the throughput of model updates. Our two main contributions can be summarized as follows. First, we prove a discrete variant of Little's law to derive a closed-form expression for relative delay, a metric that quantifies staleness. This allows us to efficiently minimize the average loss per model update, which has been the gold standard in literature to date. Second, we observe that naively optimizing this metric leads us to slow down the system drastically by overemphazing staleness at the detriment of throughput. This motivates us to introduce an alternative metric that also takes system speed into account, for which we derive a tractable upper-bound that can be minimized numerically. Extensive numerical results show that these optimizations enhance accuracy by 10% to 30%.
Authors: Harsh Poonia, Felix Divo, Kristian Kersting, Devendra Singh Dhami
Abstract: Causality in time series can be difficult to determine, especially in the presence of non-linear dependencies. The concept of Granger causality helps analyze potential relationships between variables, thereby offering a method to determine whether one time series can predict - Granger cause - future values of another. Although successful, Granger causal methods still struggle with capturing long-range relations between variables. To this end, we leverage the recently successful Extended Long Short-Term Memory (xLSTM) architecture and propose Granger causal xLSTMs (GC-xLSTM). It first enforces sparsity between the time series components by using a novel dynamic loss penalty on the initial projection. Specifically, we adaptively improve the model and identify sparsity candidates. Our joint optimization procedure then ensures that the Granger causal relations are recovered robustly. Our experimental evaluation on six diverse datasets demonstrates the overall efficacy of our proposed GC-xLSTM model.
Authors: Matthew Lyle Olson, Neale Ratzlaff, Musashi Hinck, Man Luo, Sungduk Yu, Chendi Xue, Vasudev Lal
Abstract: In the past year, large (>100B parameter) mixture-of-expert (MoE) models have become increasingly common in the open domain. While their advantages are often framed in terms of efficiency, prior work has also explored functional differentiation through routing behavior. We investigate whether expert routing in large MoE models is influenced by the semantics of the inputs. To test this, we design two controlled experiments. First, we compare activations on sentence pairs with a shared target word used in the same or different senses. Second, we fix context and substitute the target word with semantically similar or dissimilar alternatives. Comparing expert overlap across these conditions reveals clear, statistically significant evidence of semantic routing in large MoE models.
Authors: Haochen Li, Wanjin Feng, Xin Zhou, Zhiqi Shen
Abstract: Training Large Language Models (LLMs) with synthetic data is a prevalent practice in code generation. A key approach is self-training, where LLMs are iteratively trained on self-generated correct code snippets. In this case, the self-generated codes are drawn from a conditional distribution, conditioned on a specific seed description. However, the seed description is not the only valid representation that aligns with its intended meaning. With all valid descriptions and codes forming a joint space, codes drawn from the conditional distribution would lead to an underrepresentation of the full description-code space. As such, we propose Gibbs Fine-Tuning (GiFT), a novel self-training method inspired by Gibbs sampling. GiFT allows self-generated data to be drawn from the marginal distribution of the joint space, thereby mitigating the biases inherent in conditional sampling. We provide a theoretical analysis demonstrating the potential benefits of fine-tuning LLMs with code derived from the marginal distribution. Furthermore, we propose a perplexity-based code selection method to mitigate the imbalanced long-tail distribution of the self-generated codes. Empirical evaluation of two LLMs across four datasets demonstrates that GiFT achieves superior performance, particularly on more challenging benchmarks. Source code is available at https://github.com/Alex-HaochenLi/GiFT.
Authors: Florian Sestak, Artur Toshev, Andreas F\"urst, G\"unter Klambauer, Andreas Mayr, Johannes Brandstetter
Abstract: Generative models are spearheading recent progress in deep learning, showcasing strong promise for trajectory sampling in dynamical systems as well. However, whereas latent space modeling paradigms have transformed image and video generation, similar approaches are more difficult for most dynamical systems. Such systems -- from chemical molecule structures to collective human behavior -- are described by interactions of entities, making them inherently linked to connectivity patterns, entity conservation, and the traceability of entities over time. Our approach, LaM-SLidE (Latent Space Modeling of Spatial Dynamical Systems via Linked Entities), bridges the gap between: (1) keeping the traceability of individual entities in a latent system representation, and (2) leveraging the efficiency and scalability of recent advances in image and video generation, where pre-trained encoder and decoder enable generative modeling directly in latent space. The core idea of LaM-SLidE is the introduction of identifier representations (IDs) that enable the retrieval of entity properties and entity composition from latent system representations, thus fostering traceability. Experimentally, across different domains, we show that LaM-SLidE performs favorably in terms of speed, accuracy, and generalizability. Code is available at https://github.com/ml-jku/LaM-SLidE .
Authors: Jiaqi Zhao, Ming Wang, Miao Zhang, Yuzhang Shang, Xuebo Liu, Yaowei Wang, Min Zhang, Liqiang Nie
Abstract: Post-training Quantization (PTQ) technique has been extensively adopted for large language models (LLMs) compression owing to its efficiency and low resource requirement. However, current research lacks a in-depth analysis of the superior and applicable scenarios of each PTQ strategy. In addition, existing algorithms focus primarily on performance, overlooking the trade-off among model size, performance, and quantization bitwidth. To mitigate these confusions, we provide a novel benchmark for LLMs PTQ in this paper. Firstly, in order to support our benchmark, we propose a comprehensive taxonomy for existing mainstream methods by scrutinizing their computational strategies (e.g., optimization-based, compensation-based, etc.). Then, we conduct extensive experiments with the baseline within each class, covering models with various sizes (7B-70B), bitwidths, training levels (LLaMA1/2/3/3.1), architectures (Mixtral, DeepSeekMoE and Mamba) and modality (LLaVA1.5 and VILA1.5) on a wide range of evaluation metrics.Through comparative analysis on the results, we summarize the superior of each PTQ strategy and modelsize-bitwidth trade-off considering the performance. For example, our benchmark reveals that compensation-based technique demonstrates outstanding cross-architecture robustness and extremely low-bit PTQ for ultra large models should be reexamined. Finally, we further accordingly claim that a practical combination of compensation and other PTQ strategy can achieve SOTA various robustness. We believe that our benchmark will provide valuable recommendations for the deployment of LLMs and future research on PTQ approaches.We conduct an repository for our benchmark at https://github.com/zjq0455/PTQ_Benchmark.
Authors: Arshia Kermani, Ehsan Zeraatkar, Habib Irani
Abstract: The increasing computational demands of transformer models in time series classification necessitate effective optimization strategies for energy-efficient deployment. Our study presents a systematic investigation of optimization techniques, focusing on structured pruning and quantization methods for transformer architectures. Through extensive experimentation on three distinct datasets (RefrigerationDevices, ElectricDevices, and PLAID), we quantitatively evaluate model performance and energy efficiency across different transformer configurations. Our experimental results demonstrate that static quantization reduces energy consumption by 29.14% while maintaining classification performance, and L1 pruning achieves a 63% improvement in inference speed with minimal accuracy degradation. Our findings provide valuable insights into the effectiveness of optimization strategies for transformer-based time series classification, establishing a foundation for efficient model deployment in resource-constrained environments.
Authors: Wenzhe Yin, Zehao Xiao, Pan Zhou, Shujian Yu, Jiayi Shen, Jan-Jakob Sonke, Efstratios Gavves
Abstract: Multimodal alignment is crucial for various downstream tasks such as cross-modal generation and retrieval. Previous multimodal approaches like CLIP utilize InfoNCE to maximize mutual information, primarily aligning pairwise samples across modalities while overlooking distributional differences. In addition, InfoNCE has inherent conflict in terms of alignment and uniformity in multimodality, leading to suboptimal alignment with modality gaps. To overcome the limitations, we propose CS-Aligner, a novel framework that performs distributional vision-language alignment by integrating Cauchy-Schwarz (CS) divergence with mutual information. CS-Aligner captures both the global distribution information of each modality and the pairwise semantic relationships. We find that the CS divergence seamlessly addresses the InfoNCE's alignment-uniformity conflict and serves complementary roles with InfoNCE, yielding tighter and more precise alignment. Moreover, by introducing distributional alignment, CS-Aligner enables incorporating additional information from unpaired data and token-level representations, enhancing flexible and fine-grained alignment in practice. Experiments on text-to-image generation and cross-modality retrieval tasks demonstrate the effectiveness of our method on vision-language alignment.
Authors: Stefan Hegselmann, Georg von Arnim, Tillmann Rheude, Noel Kronenberg, David Sontag, Gerhard Hindricks, Roland Eils, Benjamin Wild
Abstract: Electronic Health Records (EHRs) offer considerable potential for clinical prediction, but their complexity and heterogeneity present significant challenges for traditional machine learning methods. Recently, domain-specific EHR foundation models trained on large volumes of unlabeled EHR data have shown improved predictive accuracy and generalization. However, their development is constrained by limited access to diverse, high-quality datasets, and by inconsistencies in coding standards and clinical practices. In this study, we explore the use of general-purpose Large Language Models (LLMs) to encode EHR into high-dimensional representations for downstream clinical prediction tasks. We convert structured EHR data into markdown-formatted plain text documents by replacing medical codes with natural language descriptions. This enables the use of LLMs and their extensive semantic understanding and generalization capabilities as effective encoders of EHRs without requiring access to private medical training data. We show that LLM-based embeddings can often match or even surpass the performance of a specialized EHR foundation model, CLMBR-T-Base, across 15 diverse clinical tasks from the EHRSHOT benchmark. To demonstrate generalizability, we further evaluate the approach on the UK Biobank (UKB) cohort, a population distinct from that used to train CLMBR-T-Base. Notably, one of the tested LLM-based models achieves superior performance for disease onset, hospitalization, and mortality prediction, highlighting robustness to shifts in patient populations. Our findings suggest that repurposed general-purpose LLMs for EHR encoding provide a scalable and generalizable alternative to domain-specific models for clinical prediction.
Authors: Xiao Shao, Guoqiang Wu
Abstract: In multi-task learning (MTL) with each task involving graph-dependent data, existing generalization analyses yield a \emph{sub-optimal} risk bound of $O(\frac{1}{\sqrt{n}})$, where $n$ is the number of training samples of each task. However, to improve the risk bound is technically challenging, which is attributed to the lack of a foundational sharper concentration inequality for multi-graph dependent random variables. To fill up this gap, this paper proposes a new Bennett-type inequality, enabling the derivation of a sharper risk bound of $O(\frac{\log n}{n})$. Technically, building on the proposed Bennett-type inequality, we propose a new Talagrand-type inequality for the empirical process, and further develop a new analytical framework of the local fractional Rademacher complexity to enhance generalization analyses in MTL with multi-graph dependent data. Finally, we apply the theoretical advancements to applications such as Macro-AUC optimization, illustrating the superiority of our theoretical results over prior work, which is also verified by experimental results.
Authors: Jiwan Chung, Saejin Kim, Yongrae Jo, Jaewoo Park, Dongjun Min, Youngjae Yu
Abstract: As large language models expand beyond natural language to domains such as mathematics, multimodal understanding, and embodied agents, tokens increasingly reflect metric relationships rather than purely linguistic meaning. We introduce DIST2Loss, a distance-aware framework designed to train autoregressive discrete models by leveraging predefined distance relationships among output tokens. At its core, DIST2Loss transforms continuous exponential family distributions derived from inherent distance metrics into discrete, categorical optimization targets compatible with the models' architectures. This approach enables the models to learn and preserve meaningful distance relationships during token generation while maintaining compatibility with existing architectures. Empirical evaluations show consistent performance gains in diverse multimodal applications, including visual grounding, robotic manipulation, generative reward modeling, and image generation using vector-quantized features. These improvements are most notable in low-data regimes, demonstrating DIST2Loss's strength under resource constraints.
Authors: Houyi Li, Wenzhen Zheng, Qiufeng Wang, Hanshan Zhang, Zili Wang, Shijie Xuyang, Yuantao Fan, Shuigeng Zhou, Xiangyu Zhang, Daxin Jiang
Abstract: The impressive capabilities of Large Language Models (LLMs) across diverse tasks are now well-established, yet their effective deployment necessitates careful hyperparameter optimization. Through extensive empirical studies involving grid searches across diverse configurations, we discover universal scaling laws governing these hyperparameters: optimal learning rate follows a power-law relationship with both model parameters and data sizes, while optimal batch size scales primarily with data sizes. Our analysis reveals a convex optimization landscape for hyperparameters under fixed models and data size conditions. This convexity implies an optimal hyperparameter plateau. We contribute a universal, plug-and-play optimal hyperparameter tool for the community. Its estimated values on the test set are merely 0.09% away from the globally optimal LLM performance found via an exhaustive search. These laws demonstrate remarkable robustness across variations in model sparsity, training data distribution, and model shape. To our best known, this is the first work that unifies different model shapes and structures, such as Mixture-of-Experts models and dense transformers, as well as establishes optimal hyperparameter scaling laws across diverse data distributions. This exhaustive optimization process demands substantial computational resources, utilizing nearly one million NVIDIA H800 GPU hours to train 3,700 LLMs of varying sizes and hyperparameters from scratch and consuming approximately 100 trillion tokens in total. To facilitate reproducibility and further research, we will progressively release all loss measurements and model checkpoints through our designated repository https://step-law.github.io/
Authors: Tianyu Chen, Yasi Zhang, Zhendong Wang, Ying Nian Wu, Oscar Leong, Mingyuan Zhou
Abstract: Diffusion models have achieved remarkable success in generating high-resolution, realistic images across diverse natural distributions. However, their performance heavily relies on high-quality training data, making it challenging to learn meaningful distributions from corrupted samples. This limitation restricts their applicability in scientific domains where clean data is scarce or costly to obtain. In this work, we introduce denoising score distillation (DSD), a surprisingly effective and novel approach for training high-quality generative models from low-quality data. DSD first pretrains a diffusion model exclusively on noisy, corrupted samples and then distills it into a one-step generator capable of producing refined, clean outputs. While score distillation is traditionally viewed as a method to accelerate diffusion models, we show that it can also significantly enhance sample quality, particularly when starting from a degraded teacher model. Across varying noise levels and datasets, DSD consistently improves generative performancewe summarize our empirical evidence in Fig. 1. Furthermore, we provide theoretical insights showing that, in a linear model setting, DSD identifies the eigenspace of the clean data distributions covariance matrix, implicitly regularizing the generator. This perspective reframes score distillation as not only a tool for efficiency but also a mechanism for improving generative models, particularly in low-quality data settings.
Authors: Yuze Li, Wei Zhu
Abstract: We propose an efficient fine-tuning method for time series foundation models, termed TRACE: Time Series Parameter Efficient Fine-tuning. While pretrained time series foundation models are gaining popularity, they face the following challenges: (1) Unlike natural language tasks, time series data vary in frequency, channel numbers, historical/prediction lengths. For long-term forecasting tasks in particular, tailored fine-tuning can significantly enhance performance.(2) Existing parameter-efficient tuning methods like LoRA remain applicable but require adaptation to temporal characteristics. To address these challenges, our TRACE framework introduces two key innovations: (1) Gated DSIC (Gated Dynamic Simulation Importance Calculation), an unbiased LoRA module importance selection mechanism that ensures conditional parameter consistency before and after masking. Experiments demonstrate that Gated DSIC outperforms common fine-tuning. (2) Reconstructed prediction heads for long-term forecasting tasks, which achieve comparable or superior performance to linear probing heads while drastically reducing parameter counts. Extensive experiments on long-/short-term forecasting, anomaly detection and natural language tasks across diverse datasets, coupled with ablation studies, validate the effectiveness of our method.
Authors: Tong Wu, Chong Xiang, Jiachen T. Wang, G. Edward Suh, Prateek Mittal
Abstract: Reasoning-enhanced large language models (LLMs) explicitly generate intermediate reasoning steps prior to generating final answers, helping the model excel in complex problem-solving. In this paper, we demonstrate that this emerging generation framework offers a unique opportunity for more fine-grained control over model behavior. We propose Thinking Intervention, a novel paradigm designed to explicitly guide the internal reasoning processes of LLMs by strategically inserting or revising specific thinking tokens. We find that the Thinking Intervention paradigm enhances the capabilities of reasoning models across a wide range of tasks, including instruction following on IFEval and Overthinking, instruction hierarchy on SEP, and safety alignment on XSTest and SorryBench. Our results demonstrate that Thinking Intervention significantly outperforms baseline prompting approaches, achieving up to 6.7% accuracy gains in instruction-following scenarios, 15.4% improvements in reasoning about instruction hierarchies, and a 40.0% increase in refusal rates for unsafe prompts using open-source DeepSeek R1 models. Overall, our work opens a promising new research avenue for controlling reasoning LLMs.
Authors: Liheng Ma, Soumyasundar Pal, Yingxue Zhang, Philip H. S. Torr, Mark Coates
Abstract: Transformers have attained outstanding performance across various modalities, owing to their simple but powerful scaled-dot-product (SDP) attention mechanisms. Researchers have attempted to migrate Transformers to graph learning, but most advanced Graph Transformers (GTs) have strayed far from plain Transformers, exhibiting major architectural differences either by integrating message-passing or incorporating sophisticated attention mechanisms. These divergences hinder the easy adoption of training advances for Transformers developed in other domains. Contrary to previous GTs, this work demonstrates that the plain Transformer architecture can be a powerful graph learner. To achieve this, we propose to incorporate three simple, minimal, and easy-to-implement modifications to the plain Transformer architecture to construct our Powerful Plain Graph Transformers (PPGT): (1) simplified $L_2$ attention for measuring the magnitude closeness among tokens; (2) adaptive root-mean-square normalization to preserve token magnitude information; and (3) a simple MLP-based stem for graph positional encoding. Consistent with its theoretical expressivity, PPGT demonstrates noteworthy realized expressivity on the empirical graph expressivity benchmark, comparing favorably to more complicated competitors such as subgraph GNNs and higher-order GNNs. Its outstanding empirical performance across various graph datasets also justifies the practical effectiveness of PPGT.
Authors: Garrett Baker, George Wang, Jesse Hoogland, Daniel Murfet
Abstract: We develop a linear response framework for interpretability that treats a neural network as a Bayesian statistical mechanical system. A small perturbation of the data distribution, for example shifting the Pile toward GitHub or legal text, induces a first-order change in the posterior expectation of an observable localized on a chosen component of the network. The resulting susceptibility can be estimated efficiently with local SGLD samples and factorizes into signed, per-token contributions that serve as attribution scores. We combine these susceptibilities into a response matrix whose low-rank structure separates functional modules such as multigram and induction heads in a 3M-parameter transformer.
Authors: Yize Jiang, Xinze Li, Yuanyuan Zhang, Jin Han, Youjun Xu, Ayush Pandit, Zaixi Zhang, Mengdi Wang, Mengyang Wang, Chong Liu, Guang Yang, Yejin Choi, Wu-Jun Li, Tianfan Fu, Fang Wu, Junhong Liu
Abstract: Existing protein-ligand docking studies typically focus on the self-docking scenario, which is less practical in real applications. Moreover, some studies involve heavy frameworks requiring extensive training, posing challenges for convenient and efficient assessment of docking methods. To fill these gaps, we design PoseX, an open-source benchmark to evaluate both self-docking and cross-docking, enabling a practical and comprehensive assessment of algorithmic advances. Specifically, we curated a novel dataset comprising 718 entries for self-docking and 1,312 entries for cross-docking; second, we incorporated 23 docking methods in three methodological categories, including physics-based methods (e.g., Schr\"odinger Glide), AI docking methods (e.g., DiffDock) and AI co-folding methods (e.g., AlphaFold3); third, we developed a relaxation method for post-processing to minimize conformational energy and refine binding poses; fourth, we built a leaderboard to rank submitted models in real-time. We derived some key insights and conclusions from extensive experiments: (1) AI approaches have consistently outperformed physics-based methods in overall docking success rate. (2) Most intra- and intermolecular clashes of AI approaches can be greatly alleviated with relaxation, which means combining AI modeling with physics-based post-processing could achieve excellent performance. (3) AI co-folding methods exhibit ligand chirality issues, except for Boltz-1x, which introduced physics-inspired potentials to fix hallucinations, suggesting modeling on stereochemistry improves the structural plausibility markedly. (4) Specifying binding pockets significantly promotes docking performance, indicating that pocket information can be leveraged adequately, particularly for AI co-folding methods, in future modeling efforts. The code, dataset, and leaderboard are released at https://github.com/CataAI/PoseX.
Authors: Yue Chen, Hui Kang, Jiahui Li, Geng Sun, Boxiong Wang, Jiacheng Wang, Cong Liang, Shuang Liang, Dusit Niyato
Abstract: The integration of simultaneous wireless information and power transfer (SWIPT) technology in 6G Internet of Things (IoT) networks faces significant challenges in remote areas and disaster scenarios where ground infrastructure is unavailable. This paper proposes a novel unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) system enhanced by directional antennas to provide both computational resources and energy support for ground IoT terminals. However, such systems require multiple trade-off policies to balance UAV energy consumption, terminal battery levels, and computational resource allocation under various constraints, including limited UAV battery capacity, non-linear energy harvesting characteristics, and dynamic task arrivals. To address these challenges comprehensively, we formulate a bi-objective optimization problem that simultaneously considers system energy efficiency and terminal battery sustainability. We then reformulate this non-convex problem with a hybrid solution space as a Markov decision process (MDP) and propose an improved soft actor-critic (SAC) algorithm with an action simplification mechanism to enhance its convergence and generalization capabilities. Simulation results have demonstrated that our proposed approach outperforms various baselines in different scenarios, achieving efficient energy management while maintaining high computational performance. Furthermore, our method shows strong generalization ability across different scenarios, particularly in complex environments, validating the effectiveness of our designed boundary penalty and charging reward mechanisms.
Authors: Xuyang Wang, Siyuan Duan, Qizhi Li, Guiduo Duan, Yuan Sun, Dezhong Peng
Abstract: Trustworthy multi-view learning has attracted extensive attention because evidence learning can provide reliable uncertainty estimation to enhance the credibility of multi-view predictions. Existing trusted multi-view learning methods implicitly assume that multi-view data is secure. However, in safety-sensitive applications such as autonomous driving and security monitoring, multi-view data often faces threats from adversarial perturbations, thereby deceiving or disrupting multi-view models. This inevitably leads to the adversarial unreliability problem (AUP) in trusted multi-view learning. To overcome this tricky problem, we propose a novel multi-view learning framework, namely Reliable Disentanglement Multi-view Learning (RDML). Specifically, we first propose evidential disentanglement learning to decompose each view into clean and adversarial parts under the guidance of corresponding evidences, which is extracted by a pretrained evidence extractor. Then, we employ the feature recalibration module to mitigate the negative impact of adversarial perturbations and extract potential informative features from them. Finally, to further ignore the irreparable adversarial interferences, a view-level evidential attention mechanism is designed. Extensive experiments on multi-view classification tasks with adversarial attacks show that RDML outperforms the state-of-the-art methods by a relatively large margin. Our code is available at https://github.com/Willy1005/2025-IJCAI-RDML.
Authors: Yuhui Xu, Hanze Dong, Lei Wang, Doyen Sahoo, Junnan Li, Caiming Xiong
Abstract: Large reasoning models (LRMs) have achieved remarkable progress on complex tasks by generating extended chains of thought (CoT). However, their uncontrolled output lengths pose significant challenges for real-world deployment, where inference-time budgets on tokens, latency, or compute are strictly constrained. We propose Elastic Reasoning, a novel framework for scalable chain of thoughts that explicitly separates reasoning into two phases--thinking and solution--with independently allocated budgets. At test time, Elastic Reasoning prioritizes the completeness of solution segments, significantly improving reliability under tight resource constraints. To train models that are robust to truncated thinking, we introduce a lightweight budget-constrained rollout strategy, integrated into GRPO, which teaches the model to reason adaptively when the thinking process is cut short and generalizes effectively to unseen budget constraints without additional training. Empirical results on mathematical (AIME, MATH500) and programming (LiveCodeBench, Codeforces) benchmarks demonstrate that Elastic Reasoning performs robustly under strict budget constraints, while incurring significantly lower training cost than baseline methods. Remarkably, our approach also produces more concise and efficient reasoning even in unconstrained settings. Our code has been made available at https://github.com/SalesforceAIResearch/Elastic-Reasoning.
URLs: https://github.com/SalesforceAIResearch/Elastic-Reasoning.
Authors: Xuran Li, Jingyi Wang, Xiaohan Yuan, Peixin Zhang, Zhan Qin, Zhibo Wang, Kui Ren
Abstract: It is often desirable to remove (a.k.a. unlearn) a specific part of the training data from a trained neural network model. A typical application scenario is to protect the data holder's right to be forgotten, which has been promoted by many recent regulation rules. Existing unlearning methods involve training alternative models with remaining data, which may be costly and challenging to verify from the data holder or a thirdparty auditor's perspective. In this work, we provide a new angle and propose a novel unlearning approach by imposing carefully crafted "patch" on the original neural network to achieve targeted "forgetting" of the requested data to delete. Specifically, inspired by the research line of neural network repair, we propose to strategically seek a lightweight minimum "patch" for unlearning a given data point with certifiable guarantee. Furthermore, to unlearn a considerable amount of data points (or an entire class), we propose to iteratively select a small subset of representative data points to unlearn, which achieves the effect of unlearning the whole set. Extensive experiments on multiple categorical datasets demonstrates our approach's effectiveness, achieving measurable unlearning while preserving the model's performance and being competitive in efficiency and memory consumption compared to various baseline methods.
Authors: Tianyu Chen, Haoyi Zhou, Ying Li, Hao Wang, Zhenzhe Zhang, Tianchen Zhu, Shanghang Zhang, Jianxin Li
Abstract: Fourier Neural Operators (FNO) have emerged as promising solutions for efficiently solving partial differential equations (PDEs) by learning infinite-dimensional function mappings through frequency domain transformations. However, the sparsity of high-frequency signals limits computational efficiency for high-dimensional inputs, and fixed-pattern truncation often causes high-frequency signal loss, reducing performance in scenarios such as high-resolution inputs or long-term predictions. To address these challenges, we propose FreqMoE, an efficient and progressive training framework that exploits the dependency of high-frequency signals on low-frequency components. The model first learns low-frequency weights and then applies a sparse upward-cycling strategy to construct a mixture of experts (MoE) in the frequency domain, effectively extending the learned weights to high-frequency regions. Experiments on both regular and irregular grid PDEs demonstrate that FreqMoE achieves up to 16.6% accuracy improvement while using merely 2.1% parameters (47.32x reduction) compared to dense FNO. Furthermore, the approach demonstrates remarkable stability in long-term predictions and generalizes seamlessly to various FNO variants and grid structures, establishing a new ``Low frequency Pretraining, High frequency Fine-tuning'' paradigm for solving PDEs.
Authors: Hu Wang, Congbo Ma, Ian Reid, Mohammad Yaqub
Abstract: Reward baseline is important for Reinforcement Learning (RL) algorithms to reduce variance in policy gradient estimates. Recently, for language modeling, Group Relative Policy Optimization (GRPO) is proposed to compute the advantage for each output by subtracting the mean reward, as the baseline, for all outputs in the group. However, it can lead to inaccurate advantage estimates in environments with highly noisy rewards, potentially introducing bias. In this work, we propose a model, called Kalman Filter Enhanced Group Relative Policy Optimization (KRPO), by using lightweight Kalman filtering to dynamically estimate the latent reward mean and variance. This filtering technique replaces the naive batch mean baseline, enabling more adaptive advantage normalization. Our method does not require additional learned parameters over GRPO. This approach offers a simple yet effective way to incorporate multiple outputs of GRPO into advantage estimation, improving policy optimization in settings where highly dynamic reward signals are difficult to model for language models. Through accuracy and rewards obtained from math question answering and reasoning, we show that using a more adaptive advantage estimation model, KRPO can improve the stability and performance of GRPO. The code is available at https://github.com/billhhh/KRPO_LLMs_RL.
Authors: Zhikun Tao, Gang Xiong, He Fang, Zhen Shen, Yunjun Han, Qing-Shan Jia
Abstract: Offline safe reinforcement learning(OSRL) derives constraint-satisfying policies from pre-collected datasets, offers a promising avenue for deploying RL in safety-critical real-world domains such as robotics. However, the majority of existing approaches emphasize only short-term safety, neglecting long-horizon considerations. Consequently, they may violate safety constraints and fail to ensure sustained protection during online deployment. Moreover, the learned policies often struggle to handle states and actions that are not present or out-of-distribution(OOD) from the offline dataset, and exhibit limited sample efficiency. To address these challenges, we propose a novel framework Feasibility-Aware offline Safe Reinforcement Learning with CVAE-based Pessimism (FASP). First, we employ Hamilton-Jacobi (H-J) reachability analysis to generate reliable safety labels, which serve as supervisory signals for training both a conditional variational autoencoder (CVAE) and a safety classifier. This approach not only ensures high sampling efficiency but also provides rigorous long-horizon safety guarantees. Furthermore, we utilize pessimistic estimation methods to estimate the Q-value of reward and cost, which mitigates the extrapolation errors induces by OOD actions, and penalize unsafe actions to enabled the agent to proactively avoid high-risk behaviors. Moreover, we theoretically prove the validity of this pessimistic estimation. Extensive experiments on DSRL benchmarks demonstrate that FASP algorithm achieves competitive performance across multiple experimental tasks, particularly outperforming state-of-the-art algorithms in terms of safety.
Authors: Liyang Zhao, Olurotimi Seton, Himadeep Reddy Reddivari, Suvendu Jena, Shadow Zhao, Rachit Kumar, Changshuai Wei
Abstract: The sales process involves sales functions converting leads or opportunities to customers and selling more products to existing customers. The optimization of the sales process thus is key to success of any B2B business. In this work, we introduce a principled approach to sales optimization and business AI, namely the Causal Predictive Optimization and Generation, which includes three layers: 1) prediction layer with causal ML 2) optimization layer with constraint optimization and contextual bandit 3) serving layer with Generative AI and feedback-loop for system enhancement. We detail the implementation and deployment of the system in LinkedIn, showcasing significant wins over legacy systems and sharing learning and insight broadly applicable to this field.
Authors: Xiangwen Zhuge, Xu Shen, Zeyu Wang, Fan Dang, Xuan Ding, Danyang Li, Yahui Han, Tianxiang Hao, Zheng Yang
Abstract: Efficient LLM inference on resource-constrained devices presents significant challenges in compute and memory utilization. Due to limited GPU memory, existing systems offload model weights to CPU memory, incurring substantial I/O overhead between the CPU and GPU. This leads to two major inefficiencies: (1) GPU cores are underutilized, often remaining idle while waiting for data to be loaded; and (2) GPU memory has low impact on performance, as reducing its capacity has minimal effect on overall throughput.In this paper, we propose SpecOffload, a high-throughput inference engine that embeds speculative decoding into offloading. Our key idea is to unlock latent GPU resources for storing and executing a draft model used for speculative decoding, thus accelerating inference at near-zero additional cost. To support this, we carefully orchestrate the interleaved execution of target and draft models in speculative decoding within the offloading pipeline, and propose a planner to manage tensor placement and select optimal parameters. Compared to the best baseline, SpecOffload improves GPU core utilization by 4.49x and boosts inference throughput by 2.54x. Our code is available at https://github.com/MobiSense/SpecOffload-public .
Authors: Johannes Schwab, Bryan Kelly, Semyon Malamud, Teng Andrea Xu
Abstract: The performance of the data-dependent neural tangent kernel (NTK; Jacot et al. (2018)) associated with a trained deep neural network (DNN) often matches or exceeds that of the full network. This implies that DNN training via gradient descent implicitly performs kernel learning by optimizing the NTK. In this paper, we propose instead to optimize the NTK explicitly. Rather than minimizing empirical risk, we train the NTK to minimize its generalization error using the recently developed Kernel Alignment Risk Estimator (KARE; Jacot et al. (2020)). Our simulations and real data experiments show that NTKs trained with KARE consistently match or significantly outperform the original DNN and the DNN- induced NTK (the after-kernel). These results suggest that explicitly trained kernels can outperform traditional end-to-end DNN optimization in certain settings, challenging the conventional dominance of DNNs. We argue that explicit training of NTK is a form of over-parametrized feature learning.
Authors: Yinsicheng Jiang, Yao Fu, Yeqi Huang, Ping Nie, Zhan Lu, Leyang Xue, Congjie He, Man-Kit Sit, Jilong Xue, Li Dong, Ziming Miao, Dayou Du, Tairan Xu, Kai Zou, Edoardo Ponti, Luo Mai
Abstract: The sparse Mixture-of-Experts (MoE) architecture is increasingly favored for scaling Large Language Models (LLMs) efficiently, but it depends on heterogeneous compute and memory resources. These factors jointly affect system Cost, Accuracy, and Performance (CAP), making trade-offs inevitable. Existing benchmarks often fail to capture these trade-offs accurately, complicating practical deployment decisions. To address this, we introduce MoE-CAP, a benchmark specifically designed for MoE systems. Our analysis reveals that achieving an optimal balance across CAP is difficult with current hardware; MoE systems typically optimize two of the three dimensions at the expense of the third-a dynamic we term the MoE-CAP trade-off. To visualize this, we propose the CAP Radar Diagram. We further introduce sparsity-aware performance metrics-Sparse Memory Bandwidth Utilization (S-MBU) and Sparse Model FLOPS Utilization (S-MFU)-to enable accurate performance benchmarking of MoE systems across diverse hardware platforms and deployment scenarios.
Authors: Peimian Du, Jiabin Liu, Xiaowei Jin, Mengwang Zuo, Hui Li
Abstract: This research confronts the challenge of substantial physical equation discrepancies encountered in the generation of spatiotemporal physical fields through data-driven trained models. A spatiotemporal physical field generation model, named HMT-PF, is developed based on the hybrid Mamba-Transformer architecture, incorporating unstructured grid information as input. A fine-tuning block, enhanced with physical information, is introduced to effectively reduce the physical equation discrepancies. The physical equation residuals are computed through a point query mechanism for efficient gradient evaluation, then encoded into latent space for refinement. The fine-tuning process employs a self-supervised learning approach to achieve physical consistency while maintaining essential field characteristics. Results show that the hybrid Mamba-Transformer model achieves good performance in generating spatiotemporal fields, while the physics-informed fine-tuning mechanism further reduces significant physical errors effectively. A MSE-R evaluation method is developed to assess the accuracy and realism of physical field generation.
Authors: Yuhang Liu, Yingxue Zhang, Xin Zhang, Ling Tian, Yanhua Li, Jun Luo
Abstract: Understanding and predicting urban dynamics is crucial for managing transportation systems, optimizing urban planning, and enhancing public services. While neural network-based approaches have achieved success, they often rely on task-specific architectures and large volumes of data, limiting their ability to generalize across diverse urban scenarios. Meanwhile, Large Language Models (LLMs) offer strong reasoning and generalization capabilities, yet their application to spatial-temporal urban dynamics remains underexplored. Existing LLM-based methods struggle to effectively integrate multifaceted spatial-temporal data and fail to address distributional shifts between training and testing data, limiting their predictive reliability in real-world applications. To bridge this gap, we propose UrbanMind, a novel spatial-temporal LLM framework for multifaceted urban dynamics prediction that ensures both accurate forecasting and robust generalization. At its core, UrbanMind introduces Muffin-MAE, a multifaceted fusion masked autoencoder with specialized masking strategies that capture intricate spatial-temporal dependencies and intercorrelations among multifaceted urban dynamics. Additionally, we design a semantic-aware prompting and fine-tuning strategy that encodes spatial-temporal contextual details into prompts, enhancing LLMs' ability to reason over spatial-temporal patterns. To further improve generalization, we introduce a test time adaptation mechanism with a test data reconstructor, enabling UrbanMind to dynamically adjust to unseen test data by reconstructing LLM-generated embeddings. Extensive experiments on real-world urban datasets across multiple cities demonstrate that UrbanMind consistently outperforms state-of-the-art baselines, achieving high accuracy and robust generalization, even in zero-shot settings.
Authors: Ryan Chen, Youngmin Ko, Zeyu Zhang, Catherine Cho, Sunny Chung, Mauro Giuffr\'e, Dennis L. Shung, Bradly C. Stadie
Abstract: We introduce LAMP (Linear Attribution Mapping Probe), a method that shines light onto a black-box language model's decision surface and studies how reliably a model maps its stated reasons to its predictions through a locally linear model approximating the decision surface. LAMP treats the model's own self-reported explanations as a coordinate system and fits a locally linear surrogate that links those weights to the model's output. By doing so, it reveals which stated factors steer the model's decisions, and by how much. We apply LAMP to three tasks: sentiment analysis, controversial-topic detection, and safety-prompt auditing. Across these tasks, LAMP reveals that many LLMs exhibit locally linear decision landscapes. In addition, these surfaces correlate with human judgments on explanation quality and, on a clinical case-file data set, aligns with expert assessments. Since LAMP operates without requiring access to model gradients, logits, or internal activations, it serves as a practical and lightweight framework for auditing proprietary language models, and enabling assessment of whether a model behaves consistently with the explanations it provides.
Authors: Jiahan Zhang, Yaoyu Zhang, Tao Luo
Abstract: Understanding the convergence points and optimization landscape of neural networks is crucial, particularly for homogeneous networks where Karush-Kuhn-Tucker (KKT) points of the associated maximum-margin problem often characterize solutions. This paper investigates the relationship between such KKT points across networks of different widths generated via neuron splitting. We introduce and formalize the \textbf{KKT point embedding principle}, establishing that KKT points of a homogeneous network's max-margin problem ($P_{\Phi}$) can be embedded into the KKT points of a larger network's problem ($P_{\tilde{\Phi}}$) via specific linear isometric transformations corresponding to neuron splitting. We rigorously prove this principle holds for neuron splitting in both two-layer and deep homogeneous networks. Furthermore, we connect this static embedding to the dynamics of gradient flow training with smooth losses. We demonstrate that trajectories initiated from appropriately mapped points remain mapped throughout training and that the resulting $\omega$-limit sets of directions are correspondingly mapped ($T(L(\theta(0))) = L(\boldsymbol{\eta}(0))$), thereby preserving the alignment with KKT directions dynamically when directional convergence occurs. Our findings offer insights into the effects of network width, parameter redundancy, and the structural connections between solutions found via optimization in homogeneous networks of varying sizes.
Authors: Gabriel B\'ena, Maxence Faldor, Dan F. M. Goodman, Antoine Cully
Abstract: Cellular automata have long been celebrated for their ability to generate complex behaviors from simple, local rules, with well-known discrete models like Conway's Game of Life proven capable of universal computation. Recent advancements have extended cellular automata into continuous domains, raising the question of whether these systems retain the capacity for universal computation. In parallel, neural cellular automata have emerged as a powerful paradigm where rules are learned via gradient descent rather than manually designed. This work explores the potential of neural cellular automata to develop a continuous Universal Cellular Automaton through training by gradient descent. We introduce a cellular automaton model, objective functions and training strategies to guide neural cellular automata toward universal computation in a continuous setting. Our experiments demonstrate the successful training of fundamental computational primitives - such as matrix multiplication and transposition - culminating in the emulation of a neural network solving the MNIST digit classification task directly within the cellular automata state. These results represent a foundational step toward realizing analog general-purpose computers, with implications for understanding universal computation in continuous dynamics and advancing the automated discovery of complex cellular automata behaviors via machine learning.
Authors: Song-Lin Lv, Rui Zhu, Yu-Feng Li, Lan-Zhe Guo
Abstract: Semi-supervised learning (SSL) alleviates the cost of data labeling process by exploiting unlabeled data, and has achieved promising results on various tasks such as image classification. Meanwhile, the Pretrain-Finetuning paradigm has garnered significant attention in recent years, and exploiting pre-trained models could also reduce the requirement of labeled data in downstream tasks. Therefore, a question naturally occurs: \emph{When the labeled data is scarce in the target tasks, should we exploit unlabeled data or pre-trained models?} To answer this question, we select pre-trained Vision-Language Models (VLMs) as representative pretrain-finetuning instances and propose \textit{Few-shot SSL} -- a framework that enables fair comparison between these two paradigms by controlling the amount of labeled data used. Extensive experiments across various settings demonstrate that pre-trained VLMs generally outperform SSL methods in nearly all cases, except when the data has low resolution or lacks clear semantic structure. Therefore, we encourage future SSL research to compare with pre-trained models and explore deeper integration, such as using pre-trained knowledge to enhance pseudo-labeling. To support future research, we release our unified reproduction and evaluation framework. Codes are available \href{https://anonymous.4open.science/r/Rethinking-SSL-and-Pretrain-Finetuning-5566 }{here}.
URLs: https://anonymous.4open.science/r/Rethinking-SSL-and-Pretrain-Finetuning-5566
Authors: Keqi Deng, Philip C. Woodland
Abstract: While Transformer self-attention offers strong parallelism, the Key-Value (KV) cache grows linearly with sequence length and becomes a bottleneck for inference efficiency. Multi-head latent attention was recently developed to compress the KV cache into a low-rank latent space. This paper proposes Multi-head Temporal Latent Attention (MTLA), which further reduces the KV cache size along the temporal dimension, greatly lowering the memory footprint of self-attention inference. MTLA employs a hyper-network to dynamically merge temporally adjacent KV cache vectors. To address the mismatch between the compressed KV cache and processed sequence lengths, a stride-aware causal mask is proposed to ensure efficient parallel training and consistency with inference behaviour. Experiments across tasks, including speech translation, speech recognition, speech understanding and text summarisation, demonstrate that MTLA achieves competitive performance compared to standard Multi-Head Attention (MHA), while greatly improving inference speed and GPU memory usage. For example, on a English-German speech translation task, MTLA achieves a 5.3x speedup and a reduction in GPU memory usage by a factor of 8.3 compared to MHA, while maintaining translation quality.
Authors: Amitayush Thakur, Jasper Lee, George Tsoukalas, Meghana Sistla, Matthew Zhao, Stefan Zetzsche, Greg Durrett, Yisong Yue, Swarat Chaudhuri
Abstract: We introduce ${\rm C{\small LEVER}}$, a high-quality, curated benchmark of 161 problems for end-to-end verified code generation in Lean. Each problem consists of (1) the task of generating a specification that matches a held-out ground-truth specification, and (2) the task of generating a Lean implementation that provably satisfies this specification. Unlike prior benchmarks, ${\rm C{\small LEVER}}$ avoids test-case supervision, LLM-generated annotations, and specifications that leak implementation logic or allow vacuous solutions. All outputs are verified post-hoc using Lean's type checker to ensure machine-checkable correctness. We use ${\rm C{\small LEVER}}$ to evaluate several few-shot and agentic approaches based on state-of-the-art language models. These methods all struggle to achieve full verification, establishing it as a challenging frontier benchmark for program synthesis and formal reasoning. Our benchmark can be found on GitHub(https://github.com/trishullab/clever) as well as HuggingFace(https://huggingface.co/datasets/amitayusht/clever). All our evaluation code is also available online(https://github.com/trishullab/clever-prover).
URLs: https://github.com/trishullab/clever), https://huggingface.co/datasets/amitayusht/clever)., https://github.com/trishullab/clever-prover).
Authors: Yanzhe Wen, Xunkai Li, Qi Zhang, Zhu Lei, Guang Zeng, Rong-Hua Li, Guoren Wang
Abstract: Recently, large language models (LLMs) have significantly advanced text-attributed graph (TAG) learning. However, existing methods inadequately handle data uncertainty in open-world scenarios, especially concerning limited labeling and unknown-class nodes. Prior solutions typically rely on isolated semantic or structural approaches for unknown-class rejection, lacking effective annotation pipelines. To address these limitations, we propose Open-world Graph Assistant (OGA), an LLM-based framework that combines adaptive label traceability, which integrates semantics and topology for unknown-class rejection, and a graph label annotator to enable model updates using newly annotated nodes. Comprehensive experiments demonstrate OGA's effectiveness and practicality.
Authors: Chen Zhang, Weixin Bu, Zeyi Ren, Zhengwu Liu, Yik-Chung Wu, Ngai Wong
Abstract: Inferring properties of graph-structured data, e.g., the solubility of molecules, essentially involves learning the implicit mapping from graphs to their properties. This learning process is often costly for graph property learners like Graph Convolutional Networks (GCNs). To address this, we propose a paradigm called Graph Neural Teaching (GraNT) that reinterprets the learning process through a novel nonparametric teaching perspective. Specifically, the latter offers a theoretical framework for teaching implicitly defined (i.e., nonparametric) mappings via example selection. Such an implicit mapping is realized by a dense set of graph-property pairs, with the GraNT teacher selecting a subset of them to promote faster convergence in GCN training. By analytically examining the impact of graph structure on parameter-based gradient descent during training, and recasting the evolution of GCNs--shaped by parameter updates--through functional gradient descent in nonparametric teaching, we show for the first time that teaching graph property learners (i.e., GCNs) is consistent with teaching structure-aware nonparametric learners. These new findings readily commit GraNT to enhancing learning efficiency of the graph property learner, showing significant reductions in training time for graph-level regression (-36.62%), graph-level classification (-38.19%), node-level regression (-30.97%) and node-level classification (-47.30%), all while maintaining its generalization performance.
Authors: Yikai Mao, Shaswot Shresthamali, Masaaki Kondo
Abstract: The fidelity of quantum circuits (QC) is influenced by several factors, including hardware characteristics, calibration status, and the transpilation process, all of which impact their susceptibility to noise. However, existing methods struggle to estimate and compare the noise performance of different circuit layouts due to fluctuating error rates and the absence of a standardized fidelity metric. In this work, Q-fid is introduced, a Long Short-Term Memory (LSTM) based fidelity prediction system accompanied by a novel metric designed to quantify the fidelity of quantum circuits. Q-fid provides an intuitive way to predict the noise performance of Noisy Intermediate-Scale Quantum (NISQ) circuits. This approach frames fidelity prediction as a Time Series Forecasting problem to analyze the tokenized circuits, capturing the causal dependence of the gate sequences and their impact on overall fidelity. Additionally, the model is capable of dynamically adapting to changes in hardware characteristics, ensuring accurate fidelity predictions under varying conditions. Q-fid achieves a high prediction accuracy with an average RMSE of 0.0515, up to 24.7x more accurate than the Qiskit transpile tool mapomatic. By offering a reliable method for fidelity prediction, Q-fid empowers developers to optimize transpilation strategies, leading to more efficient and noise-resilient quantum circuit implementations.
Authors: Changze Lv, Tianlong Li, Wenhao Liu, Yufei Gu, Jianhan Xu, Cenyuan Zhang, Muling Wu, Xiaoqing Zheng, Xuanjing Huang
Abstract: Spiking Neural Networks (SNNs) have emerged as a promising alternative to conventional Artificial Neural Networks (ANNs), demonstrating comparable performance in both visual and linguistic tasks while offering the advantage of improved energy efficiency. Despite these advancements, the integration of linguistic and visual features into a unified representation through spike trains poses a significant challenge, and the application of SNNs to multimodal scenarios remains largely unexplored. This paper presents SpikeCLIP, a novel framework designed to bridge the modality gap in spike-based computation. Our approach employs a two-step recipe: an ``alignment pre-training'' to align features across modalities, followed by a ``dual-loss fine-tuning'' to refine the model's performance. Extensive experiments reveal that SNNs achieve results on par with ANNs while substantially reducing energy consumption across various datasets commonly used for multimodal model evaluation. Furthermore, SpikeCLIP maintains robust image classification capabilities, even when dealing with classes that fall outside predefined categories. This study marks a significant advancement in the development of energy-efficient and biologically plausible multimodal learning systems. Our code is available at https://github.com/Lvchangze/SpikeCLIP.
Authors: Luigi Sigillo, Eleonora Grassucci, Aurelio Uncini, Danilo Comminiello
Abstract: Neural network generalizability is becoming a broad research field due to the increasing availability of datasets from different sources and for various tasks. This issue is even wider when processing medical data, where a lack of methodological standards causes large variations being provided by different imaging centers or acquired with various devices and cofactors. To overcome these limitations, we introduce a novel, generalizable, data- and task-agnostic framework able to extract salient features from medical images. The proposed quaternion wavelet network (QUAVE) can be easily integrated with any pre-existing medical image analysis or synthesis task, and it can be involved with real, quaternion, or hypercomplex-valued models, generalizing their adoption to single-channel data. QUAVE first extracts different sub-bands through the quaternion wavelet transform, resulting in both low-frequency/approximation bands and high-frequency/fine-grained features. Then, it weighs the most representative set of sub-bands to be involved as input to any other neural model for image processing, replacing standard data samples. We conduct an extensive experimental evaluation comprising different datasets, diverse image analysis, and synthesis tasks including reconstruction, segmentation, and modality translation. We also evaluate QUAVE in combination with both real and quaternion-valued models. Results demonstrate the effectiveness and the generalizability of the proposed framework that improves network performance while being flexible to be adopted in manifold scenarios and robust to domain shifts. The full code is available at: https://github.com/ispamm/QWT.
Authors: Yize Sun, Zixin Wu, Yunpu Ma, Volker Tresp
Abstract: Unsupervised representation learning presents new opportunities for advancing Quantum Architecture Search (QAS) on Noisy Intermediate-Scale Quantum (NISQ) devices. QAS is designed to optimize quantum circuits for Variational Quantum Algorithms (VQAs). Most QAS algorithms tightly couple the search space and search algorithm, typically requiring the evaluation of numerous quantum circuits, resulting in high computational costs and limiting scalability to larger quantum circuits. Predictor-based QAS algorithms mitigate this issue by estimating circuit performance based on structure or embedding. However, these methods often demand time-intensive labeling to optimize gate parameters across many circuits, which is crucial for training accurate predictors. Inspired by the classical neural architecture search algorithm Arch2vec, we investigate the potential of unsupervised representation learning for QAS without relying on predictors. Our framework decouples unsupervised architecture representation learning from the search process, enabling the learned representations to be applied across various downstream tasks. Additionally, it integrates an improved quantum circuit graph encoding scheme, addressing the limitations of existing representations and enhancing search efficiency. This predictor-free approach removes the need for large labeled datasets. During the search, we employ REINFORCE and Bayesian Optimization to explore the latent representation space and compare their performance against baseline methods. We further validate our approach by executing the best-discovered MaxCut circuits on IBM's ibm_sherbrooke quantum processor, confirming that the architectures retain optimal performance even under real hardware noise. Our results demonstrate that the framework efficiently identifies high-performing quantum circuits with fewer search iterations.
Authors: Martin Eigel, Charles Miranda
Abstract: A novel approach to approximate solutions of Stochastic Differential Equations (SDEs) by Deep Neural Networks is derived and analysed. The architecture is inspired by the notion of Deep Operator Networks (DeepONets), which is based on operator learning in function spaces in terms of a reduced basis also represented in the network. In our setting, we make use of a polynomial chaos expansion (PCE) of stochastic processes and call the corresponding architecture SDEONet. The PCE has been used extensively in the area of uncertainty quantification (UQ) with parametric partial differential equations. This however is not the case with SDE, where classical sampling methods dominate and functional approaches are seen rarely. A main challenge with truncated PCEs occurs due to the drastic growth of the number of components with respect to the maximum polynomial degree and the number of basis elements. The proposed SDEONet architecture aims to alleviate the issue of exponential complexity by learning an optimal sparse truncation of the Wiener chaos expansion. A complete convergence and complexity analysis is presented, making use of recent Neural Network approximation results. Numerical experiments illustrate the promising performance of the suggested approach in 1D and higher dimensions.
Authors: Quanjun Lang, Jianfeng Lu
Abstract: We introduce a novel approach for learning memory kernels in Generalized Langevin Equations. This approach initially utilizes a regularized Prony method to estimate correlation functions from trajectory data, followed by regression over a Sobolev norm-based loss function with RKHS regularization. Our method guarantees improved performance within an exponentially weighted L^2 space, with the kernel estimation error controlled by the error in estimated correlation functions. We demonstrate the superiority of our estimator compared to other regression estimators that rely on L^2 loss functions and also an estimator derived from the inverse Laplace transform, using numerical examples that highlight its consistent advantage across various weight parameter selections. Additionally, we provide examples that include the application of force and drift terms in the equation.
Authors: Jiping Luo, Nikolaos Pappas
Abstract: This paper studies the remote estimation of multiple Markov sources over a lossy and rate-constrained channel. Unlike most existing studies that treat all source states equally, we exploit the \emph{semantics of information} and consider that the remote actuator has different tolerances for the estimation errors. We aim to find an optimal scheduling policy that minimizes the long-term \textit{state-dependent} costs of estimation errors under a transmission frequency constraint. The optimal scheduling problem is formulated as a \emph{constrained Markov decision process} (CMDP). We show that the optimal Lagrangian cost follows a piece-wise linear and concave (PWLC) function, and the optimal policy is, at most, a randomized mixture of two simple deterministic policies. By exploiting the structural results, we develop a new \textit{intersection search} algorithm that finds the optimal policy using only a few iterations. We further propose a reinforcement learning (RL) algorithm to compute the optimal policy without knowing \textit{a priori} the channel and source statistics. To avoid the ``curse of dimensionality" in MDPs, we propose an online low-complexity \textit{drift-plus-penalty} (DPP) algorithm. Numerical results show that continuous transmission is inefficient, and remarkably, our semantic-aware policies can attain the optimum by strategically utilizing fewer transmissions by exploiting the timing of the important information.
Authors: Yebo Wu, Li Li, Chengzhong Xu
Abstract: This paper presents ProFL, a new framework that effectively addresses the memory constraints in FL. Rather than updating the full model during local training, ProFL partitions the model into blocks based on its original architecture and trains each block in a progressive fashion. It first trains the front blocks and safely freezes them after convergence. Training of the next block is then triggered. This process progressively grows the model to be trained until the training of the full model is completed. In this way, the peak memory footprint is effectively reduced for feasible deployment on heterogeneous devices. In order to preserve the feature representation of each block, the training process is divided into two stages: model shrinking and model growing. During the model shrinking stage, we meticulously design corresponding output modules to assist each block in learning the expected feature representation and obtain the initialization model parameters. Subsequently, the obtained output modules and initialization model parameters are utilized in the corresponding model growing stage, which progressively trains the full model. Additionally, a novel metric from the scalar perspective is proposed to assess the learning status of each block, enabling us to securely freeze it after convergence and initiate the training of the next one. Finally, we theoretically prove the convergence of ProFL and conduct extensive experiments on representative models and datasets to evaluate its effectiveness. The results demonstrate that ProFL effectively reduces the peak memory footprint by up to 57.4% and improves model accuracy by up to 82.4%.
Authors: Jackson Hamel, Ming-Jun Lai, Zhaiming Shen, Ye Tian
Abstract: In this work, we propose to use a local clustering approach based on the sparse solution technique to study the medical image, especially the lung cancer image classification task. We view images as the vertices in a weighted graph and the similarity between a pair of images as the edges in the graph. The vertices within the same cluster can be assumed to share similar features and properties, thus making the applications of graph clustering techniques very useful for image classification. Recently, the approach based on the sparse solutions of linear systems for graph clustering has been found to identify clusters more efficiently than traditional clustering methods such as spectral clustering. We propose to use the two newly developed local clustering methods based on sparse solution of linear system for image classification. In addition, we employ a box spline-based tight-wavelet-framelet method to clean these images and help build a better adjacency matrix before clustering. The performance of our methods is shown to be very effective in classifying images. Our approach is significantly more efficient and either favorable or equally effective compared with other state-of-the-art approaches. Finally, we shall make a remark by pointing out two image deformation methods to build up more artificial image data to increase the number of labeled images.
Authors: Sejoon Oh, Moumita Bhattacharya, Yesu Feng, Sudarshan Lamkhede
Abstract: Recommender systems have played a critical role in diverse digital services such as e-commerce, streaming media, social networks, etc. If we know what a user's intent is in a given session (e.g. do they want to watch short videos or a movie or play games; are they shopping for a camping trip), it becomes easier to provide high-quality recommendations. In this paper, we introduce IntentRec, a novel recommendation framework based on hierarchical multi-task neural network architecture that tries to estimate a user's latent intent using their short- and long-term implicit signals as proxies and uses the intent prediction to predict the next item user is likely to engage with. By directly leveraging the intent prediction, we can offer accurate and personalized recommendations to users. Our comprehensive experiments on Netflix user engagement data show that IntentRec outperforms the state-of-the-art next-item and next-intent predictors. We also share several findings and downstream applications of IntentRec.
Authors: Yuhao Wu, Evin Jaff, Ke Yang, Ning Zhang, Umar Iqbal
Abstract: LLM app (tool) ecosystems are rapidly evolving to support sophisticated use cases that often require extensive user data collection. Given that LLM apps are developed by third parties and anecdotal evidence indicating inconsistent enforcement of policies by LLM platforms, sharing user data with these apps presents significant privacy risks. In this paper, we aim to bring transparency in data practices of LLM app ecosystems. We examine OpenAI's GPT app ecosystem as a case study. We propose an LLM-based framework to analyze the natural language specifications of GPT Actions (custom tools) and assess their data collection practices. Our analysis reveals that Actions collect excessive data across 24 categories and 145 data types, with third-party Actions collecting 6.03% more data on average. We find that several Actions violate OpenAI's policies by collecting sensitive information, such as passwords, which is explicitly prohibited by OpenAI. Lastly, we develop an LLM-based privacy policy analysis framework to automatically check the consistency of data collection by Actions with disclosures in their privacy policies. Our measurements indicate that the disclosures for most of the collected data types are omitted, with only 5.8% of Actions clearly disclosing their data collection practices.
Authors: Zhizhuo Kou, Holam Yu, Junyu Luo, Jingshu Peng, Xujia Li, Chengzhong Liu, Juntao Dai, Lei Chen, Sirui Han, Yike Guo
Abstract: We present a novel three-stage framework leveraging Large Language Models (LLMs) within a risk-aware multi-agent system for automate strategy finding in quantitative finance. Our approach addresses the brittleness of traditional deep learning models in financial applications by: employing prompt-engineered LLMs to generate executable alpha factor candidates across diverse financial data, implementing multimodal agent-based evaluation that filters factors based on market status, predictive quality while maintaining category balance, and deploying dynamic weight optimization that adapts to market conditions. Experimental results demonstrate the robust performance of the strategy in Chinese & US market regimes compared to established benchmarks. Our work extends LLMs capabilities to quantitative trading, providing a scalable architecture for financial signal extraction and portfolio construction. The overall framework significantly outperforms all benchmarks with 53.17% cumulative return on SSE50 (Jan 2023 to Jan 2024), demonstrating superior risk-adjusted performance and downside protection on the market.
Authors: Aaron Steiner, Ralph Peeters, Christian Bizer
Abstract: Generative large language models (LLMs) are a promising alternative to pre-trained language models for entity matching due to their high zero-shot performance and ability to generalize to unseen entities. Existing research on using LLMs for entity matching has focused on prompt engineering and in-context learning. This paper explores the potential of fine-tuning LLMs for entity matching. We analyze fine-tuning along two dimensions: 1) the representation of training examples, where we experiment with adding different types of LLM-generated explanations to the training set, and 2) the selection and generation of training examples using LLMs. In addition to the matching performance on the source dataset, we investigate how fine-tuning affects the models ability to generalize to other in-domain datasets as well as across topical domains. Our experiments show that fine-tuning significantly improves the performance of the smaller models while the results for the larger models are mixed. Fine-tuning also improves the generalization to in-domain datasets while hurting cross-domain transfer. We show that adding structured explanations to the training set has a positive impact on the performance of three out of four LLMs, while the proposed example selection and generation methods, only improve the performance of Llama 3.1 8B while decreasing the performance of GPT-4o-mini.
Authors: Xinyi Li, Ti Zhou, Haoyu Wang, Man Lin
Abstract: Finding an optimal energy-efficient policy that is adaptable to underlying edge devices while meeting deadlines for tasks has always been challenging. This research studies generalized systems with multi-task, multi-deadline scenarios with reinforcement learning-based DVFS for energy saving for periodic soft real-time applications on edge devices. This work addresses the limitation of previous work that models a periodic system as a single task and single-deadline scenario, which is too simplified to cope with complex situations. The method encodes time series data in the Linux kernel into information that is easy to interpret for reinforcement learning, allowing the system to generate DVFS policies to adapt system patterns based on the general workload. For encoding, we present two different methods for comparison. Both methods use only one performance counter: system utilization, and the kernel only needs minimal information from the userspace. Our method is implemented on Jetson Nano Board (2GB) and is tested with three fixed multitask workloads, which are three, five, and eight tasks in the workload, respectively. For randomness and generalization, we also designed a random workload generator to build different multitask workloads to test. Based on the test results, our method could save 3%-10% power compared to Linux built-in governors.
Authors: Christopher Scarvelis, David Benhaim, Paul Zhang
Abstract: Orientation estimation is a fundamental task in 3D shape analysis which consists of estimating a shape's orientation axes: its side-, up-, and front-axes. Using this data, one can rotate a shape into canonical orientation, where its orientation axes are aligned with the coordinate axes. Developing an orientation algorithm that reliably estimates complete orientations of general shapes remains an open problem. We introduce a two-stage orientation pipeline that achieves state of the art performance on up-axis estimation and further demonstrate its efficacy on full-orientation estimation, where one seeks all three orientation axes. Unlike previous work, we train and evaluate our method on all of Shapenet rather than a subset of classes. We motivate our engineering contributions by theory describing fundamental obstacles to orientation estimation for rotationally-symmetric shapes, and show how our method avoids these obstacles.
Authors: Xiaojian Yuan, Tianyu Pang, Chao Du, Kejiang Chen, Weiming Zhang, Min Lin
Abstract: Large language models (LLMs) may memorize sensitive or copyrighted content, raising privacy and legal concerns. Due to the high cost of retraining from scratch, researchers attempt to employ machine unlearning to remove specific content from LLMs while preserving the overall performance. In this paper, we discuss several issues in machine unlearning for LLMs and provide our insights on possible approaches. To address the issue of inadequate evaluation of model outputs after unlearning, we introduce three additional metrics to evaluate token diversity, sentence semantics, and factual correctness. We then categorize unlearning methods into untargeted and targeted, and discuss their issues respectively. Specifically, the behavior that untargeted unlearning attempts to approximate is unpredictable and may involve hallucinations, and existing regularization is insufficient for targeted unlearning. To alleviate these issues, we propose using the objective of maximizing entropy (ME) for untargeted unlearning and incorporate answer preservation (AP) loss as regularization for targeted unlearning. Experimental results across three scenarios, i.e., fictitious unlearning, continual unlearning, and real-world unlearning, demonstrate the effectiveness of our approaches. The code is available at https://github.com/sail-sg/closer-look-LLM-unlearning.
URLs: https://github.com/sail-sg/closer-look-LLM-unlearning.
Authors: Zizhao Chen, Mustafa Omer Gul, Yiwei Chen, Gloria Geng, Anne Wu, Yoav Artzi
Abstract: Multi-turn interactions between large language models (LLMs) and users naturally include implicit feedback signals. If an LLM responds in an unexpected way to an instruction, the user is likely to signal it by rephrasing the request, expressing frustration, or pivoting to an alternative task. Such signals are task-independent and occupy a relatively constrained subspace of language, allowing the LLM to identify them even if it fails on the actual task. We introduce ReSpect, a method to learn from such signals in past interactions via retrospection without additional annotations. We deploy ReSpect in a new multimodal interaction scenario, where humans instruct a multimodal LLM to solve an abstract reasoning task with a combinatorial solution space. Through thousands of interactions with humans, we show how ReSpect gradually improves task completion rate from 31% to 82%, all without any external annotation.
Authors: Myeonghoon Ryu, Hongseok Oh, Suji Lee, Han Park
Abstract: We present Unified Microphone Conversion, a unified generative framework designed to bolster sound event classification (SEC) systems against device variability. While our prior CycleGAN-based methods effectively simulate device characteristics, they require separate models for each device pair, limiting scalability. Our approach overcomes this constraint by conditioning the generator on frequency response data, enabling many-to-many device mappings through unpaired training. We integrate frequency-response information via Feature-wise Linear Modulation, further enhancing scalability. Additionally, incorporating synthetic frequency response differences improves the applicability of our framework for real-world application. Experimental results show that our method outperforms the state-of-the-art by 2.6% and reduces variability by 0.8% in macro-average F1 score.
Authors: Vittorio Erba, Emanuele Troiani, Luca Biggio, Antoine Maillard, Lenka Zdeborov\'a
Abstract: Current progress in artificial intelligence is centered around so-called large language models that consist of neural networks processing long sequences of high-dimensional vectors called tokens. Statistical physics provides powerful tools to study the functioning of learning with neural networks and has played a recognized role in the development of modern machine learning. The statistical physics approach relies on simplified and analytically tractable models of data. However, simple tractable models for long sequences of high-dimensional tokens are largely underexplored. Inspired by the crucial role models such as the single-layer teacher-student perceptron (aka generalized linear regression) played in the theory of fully connected neural networks, in this paper, we introduce and study the bilinear sequence regression (BSR) as one of the most basic models for sequences of tokens. We note that modern architectures naturally subsume the BSR model due to the skip connections. Building on recent methodological progress, we compute the Bayes-optimal generalization error for the model in the limit of long sequences of high-dimensional tokens, and provide a message-passing algorithm that matches this performance. We quantify the improvement that optimal learning brings with respect to vectorizing the sequence of tokens and learning via simple linear regression. We also unveil surprising properties of the gradient descent algorithms in the BSR model.
Authors: Tiange Liu, Nikola Surjanovic, Miguel Biron-Lattes, Alexandre Bouchard-C\^ot\'e, Trevor Campbell
Abstract: Many common Markov chain Monte Carlo (MCMC) kernels can be formulated using a deterministic involutive proposal with a step size parameter. Selecting an appropriate step size is often a challenging task in practice; and for complex multiscale targets, there may not be one choice of step size that works well globally. In this work, we address this problem with a novel class of involutive MCMC methods -- AutoStep MCMC -- that selects an appropriate step size at each iteration adapted to the local geometry of the target distribution. We prove that under mild conditions AutoStep MCMC is $\pi$-invariant, irreducible, and aperiodic, and obtain bounds on expected energy jump distance and cost per iteration. Empirical results examine the robustness and efficacy of our proposed step size selection procedure, and show that AutoStep MCMC is competitive with state-of-the-art methods in terms of effective sample size per unit cost on a range of challenging target distributions.
Authors: Yamin Arefeen, Brett Levac, Jonathan I. Tamir
Abstract: We present the first acquisition-agnostic diffusion generative model for Magnetic Resonance Imaging (MRI) in the neonatal intensive care unit (NICU) to solve a range of inverse problems for shortening scan time and improving motion robustness. In-NICU MRI scanners leverage permanent magnets at lower field-strengths (i.e., below 1.5 Tesla) for non-invasive assessment of potential brain abnormalities during the critical phase of early live development, but suffer from long scan times and motion artifacts. In this setting, training data sizes are small and intrinsically suffer from low signal-to-noise ratio (SNR). This work trains a diffusion probabilistic generative model using such a real-world training dataset of clinical neonatal MRI by applying several novel signal processing and machine learning methods to handle the low SNR and low quantity of data. The model is then used as a statistical image prior to solve various inverse problems at inference time without requiring any retraining. Experiments demonstrate the generative model's utility for three real-world applications of neonatal MRI: accelerated reconstruction, motion correction, and super-resolution.
Authors: Tyler Clark, Mark Towers, Christine Evers, Jonathon Hare
Abstract: Rainbow Deep Q-Network (DQN) demonstrated combining multiple independent enhancements could significantly boost a reinforcement learning (RL) agent's performance. In this paper, we present "Beyond The Rainbow" (BTR), a novel algorithm that integrates six improvements from across the RL literature to Rainbow DQN, establishing a new state-of-the-art for RL using a desktop PC, with a human-normalized interquartile mean (IQM) of 7.4 on Atari-60. Beyond Atari, we demonstrate BTR's capability to handle complex 3D games, successfully training agents to play Super Mario Galaxy, Mario Kart, and Mortal Kombat with minimal algorithmic changes. Designing BTR with computational efficiency in mind, agents can be trained using a high-end desktop PC on 200 million Atari frames within 12 hours. Additionally, we conduct detailed ablation studies of each component, analyzing the performance and impact using numerous measures. Code is available at https://github.com/VIPTankz/BTR.
Authors: Kangyu Zhu, Peng Xia, Yun Li, Hongtu Zhu, Sheng Wang, Huaxiu Yao
Abstract: The advancement of Large Vision-Language Models (LVLMs) has propelled their application in the medical field. However, Medical LVLMs (Med-LVLMs) encounter factuality challenges due to modality misalignment, where the models prioritize textual knowledge over visual input, leading to hallucinations that contradict information in medical images. Previous attempts to enhance modality alignment in Med-LVLMs through preference optimization have inadequately mitigated clinical relevance in preference data, making these samples easily distinguishable and reducing alignment effectiveness. To address this challenge, we propose MMedPO, a novel multimodal medical preference optimization approach that considers the clinical relevance of preference samples to enhance Med-LVLM alignment. MMedPO curates multimodal preference data by introducing two types of dispreference: (1) plausible hallucinations injected through target Med-LVLMs or GPT-4o to produce medically inaccurate responses, and (2) lesion region neglect achieved through local lesion-noising, disrupting visual understanding of critical areas. We then calculate clinical relevance for each sample based on scores from multiple Med-LLMs and visual tools, and integrate these scores into the preference optimization process as weights, enabling effective alignment. Our experiments demonstrate that MMedPO significantly enhances factual accuracy in Med-LVLMs, achieving substantial improvements over existing preference optimization methods by averaging 14.2% and 51.7% across the Med-VQA and report generation tasks. Our code are available in https://github.com/aiming-lab/MMedPO.
Authors: Nikolaos Manginas, George Paliouras, Luc De Raedt
Abstract: Neurosymbolic (NeSy) AI has emerged as a promising direction to integrate neural and symbolic reasoning. Unfortunately, little effort has been given to developing NeSy systems tailored to sequential/temporal problems. We identify symbolic automata (which combine the power of automata for temporal reasoning with that of propositional logic for static reasoning) as a suitable formalism for expressing knowledge in temporal domains. Focusing on the task of sequence classification and tagging we show that symbolic automata can be integrated with neural-based perception, under probabilistic semantics towards an end-to-end differentiable model. Our proposed hybrid model, termed NeSyA (Neuro Symbolic Automata) is shown to either scale or perform more accurately than previous NeSy systems in a synthetic benchmark and to provide benefits in terms of generalization compared to purely neural systems in a real-world event recognition task.
Authors: Mengze Hong, Wailing Ng, Chen Jason Zhang, Yuanfeng Song, Di Jiang
Abstract: Discovering customer intentions in dialogue conversations is crucial for automated service agents. However, existing intent clustering methods often fail to align with human perceptions due to a heavy reliance on embedding distance metrics and a tendency to overlook underlying semantic structures. This paper proposes an LLM-in-the-loop (LLM-ITL) intent clustering framework, integrating the semantic understanding capabilities of LLMs into conventional clustering algorithms. Specifically, this paper (1) investigates the effectiveness of fine-tuned LLMs in semantic coherence evaluation and intent cluster naming, achieving over 95% accuracy aligned with human judgments; (2) designs an LLM-ITL framework that facilitates the iterative discovery of coherent intent clusters and the optimal number of clusters; and (3) proposes context-aware techniques tailored for customer service dialogue. As existing English benchmarks offer limited semantic diversity and intent groups, we introduce a comprehensive Chinese dialogue intent dataset, comprising over 100k real customer service calls and 1,507 human-annotated intent clusters. The proposed approaches significantly outperform LLM-guided baselines, achieving notable enhancements in clustering quality and lower computational cost. Combined with several best practices, our findings highlight the potential of LLM-in-the-loop techniques for scalable and human-aligned intent clustering.
Authors: Ao-Qun Jin, Tian-Yu Xiang, Xiao-Hu Zhou, Mei-Jiang Gui, Xiao-Liang Xie, Shi-Qi Liu, Shuang-Yi Wang, Yue Cao, Sheng-Bin Duan, Fu-Chao Xie, Zeng-Guang Hou
Abstract: Robots are increasingly deployed across diverse domains to tackle tasks requiring novel skills. However, current robot learning algorithms for acquiring novel skills often rely on demonstration datasets or environment interactions, resulting in high labor costs and potential safety risks. To address these challenges, this study proposes DemoGen, a skill-learning framework that enables robots to acquire novel skills from natural language instructions. DemoGen leverages the vision-language model and the video diffusion model to generate demonstration videos of novel skills, which enabling robots to learn new skills effectively. Experimental evaluations in the MetaWorld simulation environments demonstrate the pipeline's capability to generate high-fidelity and reliable demonstrations. Using the generated demonstrations, various skill learning algorithms achieve an accomplishment rate three times the original on novel tasks. These results highlight a novel approach to robot learning, offering a foundation for the intuitive and intelligent acquisition of novel robotic skills. (Project website: https://aoqunjin.github.io/LNSLGD/)
Authors: Haoyang Li, Yuchen Hu, Chen Chen, Sabato Marco Siniscalchi, Songting Liu, Eng Siong Chng
Abstract: Deep neural network (DNN)-based speech enhancement (SE) usually uses conventional activation functions, which lack the expressiveness to capture complex multiscale structures needed for high-fidelity SE. Group-Rational KAN (GR-KAN), a variant of Kolmogorov-Arnold Networks (KAN), retains KAN's expressiveness while improving scalability on complex tasks. We adapt GR-KAN to existing DNN-based SE by replacing dense layers with GR-KAN layers in the time-frequency (T-F) domain MP-SENet and adapting GR-KAN's activations into the 1D CNN layers in the time-domain Demucs. Results on Voicebank-DEMAND show that GR-KAN requires up to 4x fewer parameters while improving PESQ by up to 0.1. In contrast, KAN, facing scalability issues, outperforms MLP on a small-scale signal modeling task but fails to improve MP-SENet. We demonstrate the first successful use of KAN-based methods for consistent improvement in both time- and SoTA TF-domain SE, establishing GR-KAN as a promising alternative for SE.
Authors: Marta Gentiloni-Silveri, Antonio Ocello
Abstract: Score-based Generative Models (SGMs) aim to sample from a target distribution by learning score functions using samples perturbed by Gaussian noise. Existing convergence bounds for SGMs in the $\mathcal{W}_2$-distance rely on stringent assumptions about the data distribution. In this work, we present a novel framework for analyzing $\mathcal{W}_2$-convergence in SGMs, significantly relaxing traditional assumptions such as log-concavity and score regularity. Leveraging the regularization properties of the Ornstein--Uhlenbeck (OU) process, we show that weak log-concavity of the data distribution evolves into log-concavity over time. This transition is rigorously quantified through a PDE-based analysis of the Hamilton--Jacobi--Bellman equation governing the log-density of the forward process. Moreover, we establish that the drift of the time-reversed OU process alternates between contractive and non-contractive regimes, reflecting the dynamics of concavity. Our approach circumvents the need for stringent regularity conditions on the score function and its estimators, relying instead on milder, more practical assumptions. We demonstrate the wide applicability of this framework through explicit computations on Gaussian mixture models, illustrating its versatility and potential for broader classes of data distributions.
Authors: Fanxing Li, Fangyu Sun, Tianbao Zhang, Danping Zou
Abstract: Quadrotor control policies can be trained with high performance using the exact gradients of the rewards to directly optimize policy parameters via backpropagation-through-time (BPTT). However, designing a fully differentiable reward architecture is often challenging. Partially differentiable rewards will result in biased gradient propagation that degrades training performance. To overcome this limitation, we propose Amended Backpropagation-through-Time (ABPT), a novel approach that mitigates gradient bias while preserving the training efficiency of BPTT. ABPT combines 0-step and N-step returns, effectively reducing the bias by leveraging value gradients from the learned Q-value function. Additionally, it adopts entropy regularization and state initialization mechanisms to encourage exploration during training. We evaluate ABPT on four representative quadrotor flight tasks \li{in both real world and simulation}. Experimental results demonstrate that ABPT converges significantly faster and achieves higher ultimate rewards than existing learning algorithms, particularly in tasks involving partially differentiable rewards. The code will be released at http://github.com/Fanxing-LI/ABPT.
Authors: Nico Meyer, Christian Ufrecht, George Yammine, Georgios Kontes, Christopher Mutschler, Daniel D. Scherer
Abstract: Benchmarking and establishing proper statistical validation metrics for reinforcement learning (RL) remain ongoing challenges, where no consensus has been established yet. The emergence of quantum computing and its potential applications in quantum reinforcement learning (QRL) further complicate benchmarking efforts. To enable valid performance comparisons and to streamline current research in this area, we propose a novel benchmarking methodology, which is based on a statistical estimator for sample complexity and a definition of statistical outperformance. Furthermore, considering QRL, our methodology casts doubt on some previous claims regarding its superiority. We conducted experiments on a novel benchmarking environment with flexible levels of complexity. While we still identify possible advantages, our findings are more nuanced overall. We discuss the potential limitations of these results and explore their implications for empirical research on quantum advantage in QRL.
Authors: Akshat Gupta, Phudish Prateepamornkul, Maochuan Lu, Ahmed Alaa, Thomas Hartvigsen, Gopala Anumanchipalli
Abstract: Knowledge editing is a promising way to improve factuality in large language models, but recent studies have shown significant model degradation during sequential editing. In this paper, we formalize the popular locate-then-edit methods as a two-step fine-tuning process, allowing us to precisely identify the root cause of this degradation. We show that model degradation occurs due to (1) over-optimization of internal activations and (2) continuous norm-growth of edited matrices. To mitigate these issues, we introduce two regularization techniques: (1) Most-Probable Early Stopping (MPES) and (2) explicit Frobenius norm-constraint. We demonstrate that applying these simple yet effective regularization techniques at key points in the editing process can substantially mitigate model degradation. Combining these regularization methods enables scaling locate-then-edit methods to 10,000 edits while reducing editing time by 42-61%. These results show that targeted regularization is essential for lifelong knowledge editing.
Authors: Alan Zhu, Parth Asawa, Jared Quincy Davis, Lingjiao Chen, Boris Hanin, Ion Stoica, Joseph E. Gonzalez, Matei Zaharia
Abstract: As the demand for high-quality data in model training grows, researchers and developers are increasingly generating synthetic data to tune and train LLMs. However, current data generation methods rely on seed sets containing tens of thousands of examples to prompt instruction-tuned models. This reliance can be especially problematic when the curation of high-quality examples is expensive or difficult. In this paper we explore the novel few-shot synthetic data generation setting -- generating a high-quality dataset from a few examples. We show that when working with only a few seed examples, instruction-tuned models used in current synthetic data methods produce insufficient diversity for downstream tasks. In contrast, we show that base models without post-training, largely untapped for synthetic data generation, offer substantially greater output diversity, albeit with lower instruction following abilities. Leveraging this insight, we propose Base-Refine (BARE), a novel two-stage method that combines the diversity of base models with the quality assurance of instruction-tuned models. BARE excels in few-shot synthetic data generation: using only 3 seed examples it generates diverse, high-quality datasets that significantly improve downstream task performance. We show that fine-tuning Llama 3.1 8B with 1,000 BARE-generated samples achieves performance comparable to state-of-the-art similarly sized models on LiveCodeBench tasks. Furthermore, data generated with BARE enables a 101% improvement for a fine-tuned Llama 3.2 1B on GSM8K over data generated by only instruction-models, and an 18.4% improvement for a fine-tuned Llama 3.1 8B over the state-of-the-art RAFT method for RAG data generation.
Authors: Junjie Yao, Zhongwang Zhang, Zhi-Qin John Xu
Abstract: Transformer-based Large Language Models (LLMs) have revolutionized Natural Language Processing by demonstrating exceptional performance across diverse tasks. This study investigates the impact of the parameter initialization scale on the training behavior and task preferences of LLMs. We discover that smaller initialization scales encourage models to favor reasoning tasks, whereas larger initialization scales lead to a preference for memorization tasks. We validate this reasoning bias via real datasets and meticulously designed anchor functions. Further analysis of initial training dynamics suggests that specific model components, particularly the embedding space and self-attention mechanisms, play pivotal roles in shaping these learning biases. We provide a theoretical framework from the perspective of model training dynamics to explain these phenomena. Additionally, experiments on real-world language tasks corroborate our theoretical insights. This work enhances our understanding of how initialization strategies influence LLM performance on reasoning tasks and offers valuable guidelines for training models.
Authors: Etienne Gauthier, Francis Bach, Michael I. Jordan
Abstract: As platforms increasingly rely on learning algorithms, collectives may form and seek ways to influence these platforms to align with their own interests. This can be achieved by coordinated submission of altered data. To evaluate the potential impact of such behavior, it is essential to understand the computations that collectives must perform to impact platforms in this way. In particular, collectives need to make a priori assessments of the effect of the collective before taking action, as they may face potential risks when modifying their data. Moreover they need to develop implementable coordination algorithms based on quantities that can be inferred from observed data. We develop a framework that provides a theoretical and algorithmic treatment of these issues and present experimental results in a product evaluation domain.
Authors: Anupama Sridhar, Alexander Johansen
Abstract: Temporal Difference Learning (TD(0)) is fundamental in reinforcement learning, yet its finite-sample behavior under non-i.i.d. data and nonlinear approximation remains unknown. We provide the first high-probability, finite-sample analysis of vanilla TD(0) on polynomially mixing Markov data, assuming only Holder continuity and bounded generalized gradients. This breaks with previous work, which often requires subsampling, projections, or instance-dependent step-sizes. Concretely, for mixing exponent $\beta > 1$, Holder continuity exponent $\gamma$, and step-size decay rate $\eta \in (1/2, 1]$, we show that, with high probability, \[ \| \theta_t - \theta^* \| \leq C(\beta, \gamma, \eta)\, t^{-\beta/2} + C'(\gamma, \eta)\, t^{-\eta\gamma} \] after $t = \mathcal{O}(1/\varepsilon^2)$ iterations. These bounds match the known i.i.d. rates and hold even when initialization is nonstationary. Central to our proof is a novel discrete-time coupling that bypasses geometric ergodicity, yielding the first such guarantee for nonlinear TD(0) under realistic mixing.
Authors: Sungnyun Kim, Kangwook Jang, Sangmin Bae, Sungwoo Cho, Se-Young Yun
Abstract: Audio-visual speech recognition (AVSR) has become critical for enhancing speech recognition in noisy environments by integrating both auditory and visual modalities. However, existing AVSR systems struggle to scale up without compromising computational efficiency. In this study, we introduce MoHAVE (Mixture of Hierarchical Audio-Visual Experts), a novel robust AVSR framework designed to address these scalability constraints. By leveraging a Mixture-of-Experts (MoE) architecture, MoHAVE activates modality-specific expert groups, ensuring dynamic adaptation to various audio-visual inputs with minimal computational overhead. Key contributions of MoHAVE include: (1) a sparse MoE framework that efficiently scales AVSR model capacity, (2) a hierarchical gating mechanism that dynamically utilizes the expert groups based on input context, enhancing adaptability and robustness, and (3) remarkable performance across robust AVSR benchmarks, including LRS3 and MuAViC transcription and translation tasks, setting a new standard for scalable speech recognition systems.
Authors: Ashley Ferreira, Mahip Singh, Yukiya Saito, Andrea Capra, Ina Carli, Daniel Duque Quiceno, Wojciech T. Fedorko, Makoto C. Fujiwara, Muyan Li, Lars Martin, Gareth Smith, Anqui Xu
Abstract: The ALPHA-g experiment at CERN aims to precisely measure the terrestrial gravitational acceleration of antihydrogen atoms. A radial Time Projection Chamber (rTPC), that surrounds the ALPHA-g magnetic trap, is employed to determine the annihilation location, called the vertex. The standard approach requires identifying the trajectories of the ionizing particles in the rTPC from the location of their interaction in the gas (spacepoints), and inferring the vertex positions by finding the point where those trajectories (helices) pass closest to one another. In this work, we present a novel approach to vertex reconstruction using an ensemble of models based on the PointNet deep learning architecture. The newly developed model, PointNet Ensemble for Annihilation Reconstruction (PEAR), directly learns the relation between the location of the vertices and the rTPC spacepoints, thus eliminating the need to identify and fit the particle tracks. PEAR shows strong performance in reconstructing vertical vertex positions from simulated data, that is superior to the standard approach for all metrics considered. Furthermore, the deep learning approach can reconstruct the vertical vertex position when the standard approach fails.
Authors: Wentao Wang, Guangyuan Jiang, Tal Linzen, Brenden M. Lake
Abstract: Humans can quickly learn a new word from a few illustrative examples, and then systematically and flexibly use it in novel contexts. Yet the abilities of current language models for few-shot word learning, and methods for improving these abilities, are underexplored. In this study, we introduce a novel method, Meta-training for IN-context learNing Of Words (Minnow). This method trains language models to generate new examples of a word's usage given a few in-context examples, using a special placeholder token to represent the new word. This training is repeated on many new words to develop a general word-learning ability. We find that training models from scratch with Minnow on human-scale child-directed language enables strong few-shot word learning, comparable to a large language model (LLM) pre-trained on orders of magnitude more data. Furthermore, through discriminative and generative evaluations, we demonstrate that finetuning pre-trained LLMs with Minnow improves their ability to discriminate between new words, identify syntactic categories of new words, and generate reasonable new usages and definitions for new words, based on one or a few in-context examples. These findings highlight the data efficiency of Minnow and its potential to improve language model performance in word learning tasks.
Authors: Han-Byul Kim, Duc Hoang, Arnav Kundu, Mohammad Samragh, Minsik Cho
Abstract: With the rapid expansion in the scale of large language models (LLMs), enabling efficient distributed inference across multiple computing units has become increasingly critical. However, communication overheads from popular distributed inference techniques such as Tensor Parallelism pose a significant challenge to achieve scalability and low latency. Therefore, we introduce a novel optimization technique, Sync-Point Drop (SPD), to reduce communication overheads in tensor parallelism by selectively dropping synchronization on attention outputs. In detail, we first propose a block design that allows execution to proceed without communication through SPD. Second, we apply different SPD strategies to attention blocks based on their sensitivity to the model accuracy. The proposed methods effectively alleviate communication bottlenecks while minimizing accuracy degradation during LLM inference, offering a scalable solution for diverse distributed environments: SPD offered about 20% overall inference latency reduction with < 1% accuracy regression for LLaMA2-70B inference over 8 GPUs.
Authors: Rui Hu, Delai Qiu, Shuyu Wei, Jiaming Zhang, Yining Wang, Shengping Liu, Jitao Sang
Abstract: Omnimodal Large Language Models (OLLMs) have shown significant progress in integrating vision and text, but still struggle with integrating vision and audio, often exhibiting suboptimal performance when processing audio queries compared to text queries. This disparity is primarily due to insufficient alignment between vision and audio modalities during training, leading to inadequate attention to visual information when using audio queries. To mitigate this issue, we propose a Self-Knowledge Distillation (Self-KD) training method where the vision-text component of the OLLM serves as the teacher and the vision-audio component as the student. This enables the model to process audio in a manner analogous to its text processing. Our experimental results demonstrate that Self-KD is an effective method for enhancing the vision-audio capabilities of OLLMs by learning from the vision-text components, which subsequently improves the interaction between audio and images and results in improved performance on multimodal tasks.
Authors: Matteo Gasparin, Aaditya Ramdas
Abstract: Vovk (2015) introduced cross-conformal prediction, a modification of split conformal designed to improve the width of prediction sets. The method, when trained with a miscoverage rate equal to $\alpha$ and $n \gg K$, ensures a marginal coverage of at least $1 - 2\alpha - 2(1-\alpha)(K-1)/(n+K)$, where $n$ is the number of observations and $K$ denotes the number of folds. A simple modification of the method achieves coverage of at least $1-2\alpha$. In this work, we propose new variants of both methods that yield smaller prediction sets without compromising the latter theoretical guarantees. The proposed methods are based on recent results deriving more statistically efficient combination of p-values that leverage exchangeability and randomization. Simulations confirm the theoretical findings and bring out some important tradeoffs.
Authors: Davis Brown, Prithvi Balehannina, Helen Jin, Shreya Havaldar, Hamed Hassani, Eric Wong
Abstract: Language model evaluations often fail to characterize consequential failure modes, forcing experts to inspect outputs and build new benchmarks. We introduce task elicitation, a method that automatically builds new evaluations to profile model behavior. Task elicitation finds hundreds of natural-language tasks -- an order of magnitude more than prior work -- where frontier models exhibit systematic failures, in domains ranging from forecasting to online harassment. For example, we find that Sonnet 3.5 over-associates quantum computing and AGI and that o3-mini is prone to hallucination when fabrications are repeated in-context.
Authors: Zhengyao Gu, Henry Peng Zou, Yankai Chen, Aiwei Liu, Weizhi Zhang, Philip S. Yu
Abstract: The high cost of obtaining high-quality annotated data for in-context learning (ICL) has motivated the development of methods that use self-generated annotations in place of ground-truth labels. While these approaches have shown promising results in few-shot settings, they generally do not scale to many-shot scenarios. In this work, we study ICL with self-generated examples using a framework analogous to traditional semi-supervised learning, consisting of annotation generation, demonstration selection, and in-context inference. Within this framework, we propose a simple baseline that outperforms ground-truth ICL in zero-shot, few-shot, and many-shot settings. Notably, we observe a scaling law with this baseline, where optimal performance is achieved with more than 1,000 demonstrations. To fully exploit the many-shot capabilities of semi-supervised ICL, we introduce IterPSD, an iterative annotation approach that integrates iterative refinement and curriculum pseudo-labeling techniques from semi-supervised learning, yielding up to 6.8% additional gains on classification tasks.
Authors: Simon A. Aytes, Jinheon Baek, Sung Ju Hwang
Abstract: Recent advances in large language models (LLMs) have enabled strong reasoning capabilities through Chain-of-Thought (CoT) prompting, which elicits step-by-step problem solving, but often at the cost of excessive verbosity in intermediate outputs, leading to increased computational overhead. We propose Sketch-of-Thought (SoT), a prompting framework that integrates cognitively inspired reasoning paradigms with linguistic constraints to reduce token usage while preserving reasoning accuracy. SoT is designed as a flexible, modular approach and is instantiated with three paradigms--Conceptual Chaining, Chunked Symbolism, and Expert Lexicons--each tailored to distinct reasoning tasks and selected dynamically at test-time by a lightweight routing model. Across 15 reasoning datasets spanning multiple domains, languages, and modalities, SoT achieves token reductions of up to 78% with minimal accuracy loss. In tasks such as mathematical and multi-hop reasoning, it even improves accuracy while shortening outputs.
Authors: Abeda Sultana, Nabin Pakka, Fei Xu, Xu Yuan, Li Chen, Nian-Feng Tzeng
Abstract: Scheduling deep learning (DL) models to train on powerful clusters with accelerators like GPUs and TPUs, presently falls short, either lacking fine-grained heterogeneity awareness or leaving resources substantially under-utilized. To fill this gap, we propose a novel design of a task-level heterogeneity-aware scheduler, Hadar, based on an optimization framework that can boost resource utilization. Hadar leverages the performance traits of DL jobs on a heterogeneous DL cluster, characterizes the task-level performance heterogeneity in the optimization problem, and makes scheduling decisions across both spatial and temporal dimensions. It involves the primal-dual framework employing a dual subroutine, to solve the optimization problem and guide the scheduling design. Our trace-driven simulation with representative DL model training workloads demonstrates that Hadar accelerates the total time duration by 1.20x when compared with its state-of-the-art heterogeneity-aware counterpart, Gavel. Further, our Hadar scheduler is enhanced to HadarE by forking each job into multiple copies to let a job train concurrently on heterogeneous GPUs resided on separate available nodes (i.e., machines or servers) for resource utilization enhancement. HadarE is evaluated extensively on physical DL clusters for comparison with Hadar and Gavel. With substantial enhancement in cluster resource utilization (by 1.45x), HadarE exhibits considerable speed-ups in DL model training, reducing the total time duration by 50% (or 80%) on an Amazon's AWS (or our lab) cluster, while producing trained DL models with consistently better inference quality than those trained by Hadar.
Authors: Mohammad Al-Jarrah, Bamdad Hosseini, Amirhossein Taghvaei
Abstract: In this paper, we present the amortized optimal transport filter (A-OTF) designed to mitigate the computational burden associated with the real-time training of optimal transport filters (OTFs). OTFs can perform accurate non-Gaussian Bayesian updates in the filtering procedure, but they require training at every time step, which makes them expensive. The proposed A-OTF framework exploits the similarity between OTF maps during an initial/offline training stage in order to reduce the cost of inference during online calculations. More precisely, we use clustering algorithms to select relevant subsets of pre-trained maps whose weighted average is used to compute the A-OTF model akin to a mixture of experts. A series of numerical experiments validate that A-OTF achieves substantial computational savings during online inference while preserving the inherent flexibility and accuracy of OTF.
Authors: Martin Ritzert, Polina Turishcheva, Laura Hansel, Paul Wollenhaupt, Marissa A. Weis, Alexander S. Ecker
Abstract: Hierarchical clustering is an effective, interpretable method for analyzing structure in data. It reveals insights at multiple scales without requiring a predefined number of clusters and captures nested patterns and subtle relationships, which are often missed by flat clustering approaches. However, existing hierarchical clustering methods struggle with high-dimensional data, especially when there are no clear density gaps between modes. In this work, we introduce t-NEB, a probabilistically grounded hierarchical clustering method, which yields state-of-the-art clustering performance on naturalistic high-dimensional data. t-NEB consists of three steps: (1) density estimation via overclustering; (2) finding maximum density paths between clusters; (3) creating a hierarchical structure via bottom-up cluster merging. t-NEB uses a probabilistic parametric density model for both overclustering and cluster merging, which yields both high clustering performance and a meaningful hierarchy, making it a valuable tool for exploratory data analysis. Code is available at https://github.com/ecker-lab/tneb clustering.
Authors: Richie Li, Sicheng Chen
Abstract: Transformer-based large language models (LLMs) rely heavily on intensive matrix multiplications for attention and feed-forward layers, with the Q, K, and V linear projections in the Multi-Head Self-Attention (MHA) module constituting a decisive performance bottleneck. In this work, we introduce a highly optimized tiled matrix multiplication accelerator on a resource-constrained Xilinx KV260 FPGA that not only addresses this challenge but sets a new standard for efficiency and performance. Our design exploits persistent on-chip storage, a robust two-level tiling strategy for maximal data reuse, and a systolic-like unrolled compute engine that together deliver unparalleled speed and energy efficiency. Integrated with DistilBERT for Q, K, and V projections, our accelerator achieves an unequivocal 7x speedup over ARM CPU implementations (PyTorch) and an extraordinary 200x improvement over naive NumPy, reaching a throughput of up to 3.1~GFLOPs for matrix multiplications on (64,768) x (768,3072) matrices while operating at a conservative 100 MHz. These results decisively demonstrate the transformative potential of FPGA-based acceleration for critical Transformer operations, paving the way for scalable and energy-efficient deep learning inference on edge devices.
Authors: Luis Wyss, Vincent Mallet, Wissam Karroucha, Karsten Borgwardt, Carlos Oliver
Abstract: The relationship between RNA structure and function has recently attracted interest within the deep learning community, a trend expected to intensify as nucleic acid structure models advance. Despite this momentum, a lack of standardized, accessible benchmarks for applying deep learning to RNA 3D structures hinders progress. To this end, we introduce a collection of seven benchmarking datasets specifically designed to support RNA structure-function prediction. Built on top of the established Python package rnaglib, our library streamlines data distribution and encoding, provides tools for dataset splitting and evaluation, and offers a comprehensive, user-friendly environment for model comparison. The modular and reproducible design of our datasets encourages community contributions and enables rapid customization. To demonstrate the utility of our benchmarks, we report baseline results for all tasks using a relational graph neural network.
Authors: Flavia Esposito, Andersen Ang
Abstract: Low-rank optimization problems with sparse simplex constraints involve variables that must satisfy nonnegativity, sparsity, and sum-to-one conditions, making their optimization particularly challenging due to the interplay between low-rank structures and constraints. These problems arise in various applications, including machine learning, signal processing, environmental fields, and computational biology. In this paper, we propose a novel manifold optimization approach to tackle these problems efficiently. Our method leverages the geometry of oblique rotation manifolds to reformulate the problem and introduces a new Riemannian optimization method based on Riemannian gradient descent that strictly maintains the simplex constraints. By exploiting the underlying manifold structure, our approach improves optimization efficiency. Experiments on synthetic datasets compared to standard Euclidean and Riemannian methods show the effectiveness of the proposed method.
Authors: Junjie Yang, Ke Lin, Xing Yu
Abstract: Chain of Thought (CoT) reasoning enhances language models' performance but often leads to inefficient "overthinking" on simple problems. We identify that existing approaches directly penalizing reasoning length fail to account for varying problem complexity. Our approach constructs rewards through length and quality comparisons, guided by theoretical assumptions that jointly enhance solution correctness with conciseness. Moreover, we further demonstrate our method to fuzzy tasks where ground truth is unavailable. Experiments across multiple reasoning benchmarks demonstrate that our method maintains accuracy while generating significantly more concise explanations, effectively teaching models to "think when needed."
Authors: Conghao Xiong, Hao Chen, Joseph J. Y. Sung
Abstract: Computational pathology, which involves analyzing whole slide images for automated cancer diagnosis, relies on multiple instance learning, where performance depends heavily on the feature extractor and aggregator. Recent Pathology Foundation Models (PFMs), pretrained on large-scale histopathology data, have significantly enhanced both the extractor and aggregator, but they lack a systematic analysis framework. In this survey, we present a hierarchical taxonomy organizing PFMs through a top-down philosophy applicable to foundation model analysis in any domain: model scope, model pretraining, and model design. Additionally, we systematically categorize PFM evaluation tasks into slide-level, patch-level, multimodal, and biological tasks, providing comprehensive benchmarking criteria. Our analysis identifies critical challenges in both PFM development (pathology-specific methodology, end-to-end pretraining, data-model scalability) and utilization (effective adaptation, model maintenance), paving the way for future directions in this promising field. Resources referenced in this survey are available at https://github.com/BearCleverProud/AwesomeWSI.
Authors: Maximilian G. Schuh, Joshua Hesse, Stephan A. Sieber
Abstract: Antibiotic resistance presents a growing global health crisis, demanding new therapeutic strategies that target novel bacterial mechanisms. Recent advances in protein structure prediction and machine learning-driven molecule generation offer a promising opportunity to accelerate drug discovery. However, practical guidance on selecting and integrating these models into real-world pipelines remains limited. In this study, we develop an end-to-end, artificial intelligence-guided antibiotic discovery pipeline that spans target identification to compound realization. We leverage structure-based clustering across predicted proteomes of multiple pathogens to identify conserved, essential, and non-human-homologous targets. We then systematically evaluate six leading 3D-structure-aware generative models$\unicode{x2014}$spanning diffusion, autoregressive, graph neural network, and language model architectures$\unicode{x2014}$on their usability, chemical validity, and biological relevance. Rigorous post-processing filters and commercial analogue searches reduce over 100 000 generated compounds to a focused, synthesizable set. Our results highlight DeepBlock and TamGen as top performers across diverse criteria, while also revealing critical trade-offs between model complexity, usability, and output quality. This work provides a comparative benchmark and blueprint for deploying artificial intelligence in early-stage antibiotic development.
Authors: Shu Tamano, Masanori Nojima
Abstract: Off-policy evaluation (OPE) and off-policy learning (OPL) for contextual bandit policies leverage historical data to evaluate and optimize a target policy. Most existing OPE/OPL methods--based on importance weighting or imputation--assume common support between the target and logging policies. When this assumption is violated, these methods typically require unstable extrapolation, truncation, or conservative strategies for individuals outside the common support assumption. However, such approaches can be inadequate in settings where explicit evaluation or optimization for such individuals is required. To address this issue, we propose DOLCE: Decomposing Off-policy evaluation/learning into Lagged and Current Effects, a novel estimator that leverages contextual information from multiple time points to decompose rewards into lagged and current effects. By incorporating both past and present contexts, DOLCE effectively handles individuals who violate the common support assumption. We show that the proposed estimator is unbiased under two assumptions--local correctness and conditional independence. Our experiments demonstrate that DOLCE achieves substantial improvements in OPE and OPL, particularly as the proportion of individuals outside the common support assumption increases.
Authors: Sai Krishna Mendu, Harish Yenala, Aditi Gulati, Shanu Kumar, Parag Agrawal
Abstract: Large language models (LLMs) have become integral to various real-world applications, leveraging massive, web-sourced datasets like Common Crawl, C4, and FineWeb for pretraining. While these datasets provide linguistic data essential for high-quality natural language generation, they often contain harmful content, such as hate speech, misinformation, and biased narratives. Training LLMs on such unfiltered data risks perpetuating toxic behaviors, spreading misinformation, and amplifying societal biases which can undermine trust in LLM-driven applications and raise ethical concerns about their use. This paper presents a large-scale analysis of inappropriate content across these datasets, offering a comprehensive taxonomy that categorizes harmful webpages into Topical and Toxic based on their intent. We also introduce a prompt evaluation dataset, a high-accuracy Topical and Toxic Prompt (TTP), and a transformer-based model (HarmFormer) for harmful content filtering. Additionally, we create a new multi-harm open-ended toxicity benchmark (HAVOC) and provide crucial insights into how models respond to adversarial toxic inputs. We share TTP, TTP-Eval, HAVOC and a sample of C4 inferenced on HarmFormer. Our work offers insights into ensuring safer LLM pretraining and serves as a resource for Responsible AI (RAI) compliance.
Authors: Woosang Lim, Zekun Li, Gyuwan Kim, Sungyoung Ji, HyeonJung Kim, Kyuri Choi, Jin Hyuk Lim, Kyungpyo Park, William Yang Wang
Abstract: Long-context large language models (LC LLMs) combined with retrieval-augmented generation (RAG) hold strong potential for complex multi-hop and large-document tasks. However, existing RAG systems often suffer from imprecise retrieval, incomplete context coverage under constrained windows, and fragmented information from suboptimal context construction. We introduce Multi-scale Adaptive Context RAG (MacRAG), a hierarchical RAG framework that compresses and partitions documents into coarse-to-fine granularities, then adaptively merges relevant contexts through real-time chunk- and document-level expansions. By initiating with finest-level retrieval and progressively incorporating broader, higher-level context, MacRAG constructs effective query-specific long contexts, optimizing both precision and coverage. Evaluations on challenging LongBench expansions of HotpotQA, 2WikiMultihopQA, and Musique confirm MacRAG consistently surpasses baseline RAG pipelines in single- and multi-step generation using Llama-3.1-8B, Gemini-1.5-pro, and GPT-4o. Our results establish MacRAG as an efficient, scalable solution for real-world long-context, multi-hop reasoning. Our code is available at https://github.com/Leezekun/MacRAG.
Authors: Tingfeng Lan, Yusen Wu, Bin Ma, Zhaoyuan Su, Rui Yang, Tekin Bicer, Dong Li, Yue Cheng
Abstract: Fine-tuning large language models (LLMs) often exceeds GPU memory limits, prompting systems to offload model states to CPU memory. However, existing offloaded training frameworks like ZeRO-Offload treat all parameters equally and update the full model on the CPU, causing severe GPU stalls, where fast, expensive GPUs sit idle waiting for slow CPU updates and limited-bandwidth PCIe transfers. We present ZenFlow, a new offloading framework that prioritizes important parameters and decouples updates between GPU and CPU. ZenFlow performs in-place updates of important gradients on GPU, while asynchronously offloading and accumulating less important ones on CPU, fully overlapping CPU work with GPU computation. To scale across GPUs, ZenFlow introduces a lightweight gradient selection method that exploits a novel spatial and temporal locality property of important gradients, avoiding costly global synchronization. ZenFlow achieves up to 5x end-to-end speedup, 2x lower PCIe traffic, and reduces GPU stalls by over 85 percent, all while preserving accuracy.
Authors: Naoki Hayashi, Takuro Kutsuna, Sawa Takamuku
Abstract: In statistical learning, models are classified as regular or singular depending on whether the mapping from parameters to probability distributions is injective. Most models with hierarchical structures or latent variables are singular, for which conventional criteria such as the Akaike Information Criterion and the Bayesian Information Criterion are inapplicable due to the breakdown of normal approximations for the likelihood and posterior. To address this, the Widely Applicable Information Criterion (WAIC) and the Widely Applicable Bayesian Information Criterion (WBIC) have been proposed. Since WAIC and WBIC are computed using posterior distributions at different temperature settings, separate posterior sampling is generally required. In this paper, we theoretically derive an asymptotic equation that links WAIC and WBIC, despite their dependence on different posteriors. This equation yields an asymptotically unbiased expression of WAIC in terms of the posterior distribution used for WBIC. The result clarifies the structural relationship between these criteria within the framework of singular learning theory, and deepens understanding of their asymptotic behavior. This theoretical contribution provides a foundation for future developments in the computational efficiency of model selection in singular models.
Authors: Guangke Chen, Fu Song, Zhe Zhao, Xiaojun Jia, Yang Liu, Yanchen Qiao, Weizhe Zhang
Abstract: Jailbreak attacks to Large audio-language models (LALMs) are studied recently, but they achieve suboptimal effectiveness, applicability, and practicability, particularly, assuming that the adversary can fully manipulate user prompts. In this work, we first conduct an extensive experiment showing that advanced text jailbreak attacks cannot be easily ported to end-to-end LALMs via text-to speech (TTS) techniques. We then propose AudioJailbreak, a novel audio jailbreak attack, featuring (1) asynchrony: the jailbreak audio does not need to align with user prompts in the time axis by crafting suffixal jailbreak audios; (2) universality: a single jailbreak perturbation is effective for different prompts by incorporating multiple prompts into perturbation generation; (3) stealthiness: the malicious intent of jailbreak audios will not raise the awareness of victims by proposing various intent concealment strategies; and (4) over-the-air robustness: the jailbreak audios remain effective when being played over the air by incorporating the reverberation distortion effect with room impulse response into the generation of the perturbations. In contrast, all prior audio jailbreak attacks cannot offer asynchrony, universality, stealthiness, or over-the-air robustness. Moreover, AudioJailbreak is also applicable to the adversary who cannot fully manipulate user prompts, thus has a much broader attack scenario. Extensive experiments with thus far the most LALMs demonstrate the high effectiveness of AudioJailbreak. We highlight that our work peeks into the security implications of audio jailbreak attacks against LALMs, and realistically fosters improving their security robustness. The implementation and audio samples are available at our website https://audiojailbreak.github.io/AudioJailbreak.
Authors: Jiajun Shi, Jian Yang, Jiaheng Liu, Xingyuan Bu, Jiangjie Chen, Junting Zhou, Kaijing Ma, Zhoufutu Wen, Bingli Wang, Yancheng He, Liang Song, Hualei Zhu, Shilong Li, Xingjian Wang, Wei Zhang, Ruibin Yuan, Yifan Yao, Wenjun Yang, Yunli Wang, Siyuan Fang, Siyu Yuan, Qianyu He, Xiangru Tang, Yingshui Tan, Wangchunshu Zhou, Zhaoxiang Zhang, Zhoujun Li, Wenhao Huang, Ge Zhang
Abstract: Recent advancements in large language models (LLMs) underscore the need for more comprehensive evaluation methods to accurately assess their reasoning capabilities. Existing benchmarks are often domain-specific and thus cannot fully capture an LLM's general reasoning potential. To address this limitation, we introduce the Knowledge Orthogonal Reasoning Gymnasium (KORGym), a dynamic evaluation platform inspired by KOR-Bench and Gymnasium. KORGym offers over fifty games in either textual or visual formats and supports interactive, multi-turn assessments with reinforcement learning scenarios. Using KORGym, we conduct extensive experiments on 19 LLMs and 8 VLMs, revealing consistent reasoning patterns within model families and demonstrating the superior performance of closed-source models. Further analysis examines the effects of modality, reasoning strategies, reinforcement learning techniques, and response length on model performance. We expect KORGym to become a valuable resource for advancing LLM reasoning research and developing evaluation methodologies suited to complex, interactive environments.
Authors: Abhimanyu Talwar, Julien Laasri
Abstract: Certain pairs of languages suffer from lack of a parallel corpus which is large in size and diverse in domain. One of the ways this is overcome is via use of a pivot language. In this paper we use Hindi as a pivot language to translate Nepali into English. We describe what makes Hindi a good candidate for the pivot. We discuss ways in which a pivot language can be used, and use two such approaches - the Transfer Method (fully supervised) and Backtranslation (semi-supervised) - to translate Nepali into English. Using the former, we are able to achieve a devtest Set SacreBLEU score of 14.2, which improves the baseline fully supervised score reported by (Guzman et al., 2019) by 6.6 points. While we are slightly below the semi-supervised baseline score of 15.1, we discuss what may have caused this under-performance, and suggest scope for future work.