Authors: Jihao Andreas Lin, Shreyas Padhy, Bruno Mlodozeniec, Javier Antor\'an, Jos\'e Miguel Hern\'andez-Lobato
Abstract: Scaling hyperparameter optimisation to very large datasets remains an open problem in the Gaussian process community. This paper focuses on iterative methods, which use linear system solvers, like conjugate gradients, alternating projections or stochastic gradient descent, to construct an estimate of the marginal likelihood gradient. We discuss three key improvements which are applicable across solvers: (i) a pathwise gradient estimator, which reduces the required number of solver iterations and amortises the computational cost of making predictions, (ii) warm starting linear system solvers with the solution from the previous step, which leads to faster solver convergence at the cost of negligible bias, (iii) early stopping linear system solvers after a limited computational budget, which synergises with warm starting, allowing solver progress to accumulate over multiple marginal likelihood steps. These techniques provide speed-ups of up to $72\times$ when solving to tolerance, and decrease the average residual norm by up to $7\times$ when stopping early.
Authors: Yizhi Wang, Minjia Chen, Chunhui Yao, Jie Ma, Ting Yan, Richard Penty, Qixiang Cheng
Abstract: Physical neural networks (PNNs) are emerging paradigms for neural network acceleration due to their high-bandwidth, in-propagation analogue processing. Despite the advantages of PNN for inference, training remains a challenge. The imperfect information of the physical transformation means the failure of conventional gradient-based updates from backpropagation (BP). Here, we present the asymmetrical training (AT) method, which treats the PNN structure as a grey box. AT performs training while only knowing the last layer output and neuron topological connectivity of a deep neural network structure, not requiring information about the physical control-transformation mapping. We experimentally demonstrated the AT method on deep grey-box PNNs implemented by uncalibrated photonic integrated circuits (PICs), improving the classification accuracy of Iris flower and modified MNIST hand-written digits from random guessing to near theoretical maximum. We also showcased the consistently enhanced performance of AT over BP for different datasets, including MNIST, fashion-MNIST, and Kuzushiji-MNIST. The AT method demonstrated successful training with minimal hardware overhead and reduced computational overhead, serving as a robust light-weight training alternative to fully explore the advantages of physical computation.
Authors: Gongyue Zhang, Honghai Liu
Abstract: We have identified a potential method for unifying first-order optimizers through the use of variable Second-Moment Exponential Scaling(SMES). We begin with back propagation, addressing classic phenomena such as gradient vanishing and explosion, as well as issues related to dataset sparsity, and introduce the theory of balance in optimization. Through this theory, we suggest that SGD and adaptive optimizers can be unified under a broader inference, employing variable moving exponential scaling to achieve a balanced approach within a generalized formula for first-order optimizers. We conducted tests on some classic datasets and networks to confirm the impact of different balance coefficients on the overall training process.
Authors: Fatemeh Nassajian Mojarrad, Lorenzo Bini, Thomas Matthes, St\'ephane Marchand-Maillet
Abstract: In the complex landscape of hematologic samples such as peripheral blood or bone marrow derived from flow cytometry (FC) data, cell-level prediction presents profound challenges. This work explores injecting hierarchical prior knowledge into graph neural networks (GNNs) for single-cell multi-class classification of tabular cellular data. By representing the data as graphs and encoding hierarchical relationships between classes, we propose our hierarchical plug-in method to be applied to several GNN models, namely, FCHC-GNN, and effectively designed to capture neighborhood information crucial for single-cell FC domain. Extensive experiments on our cohort of 19 distinct patients, demonstrate that incorporating hierarchical biological constraints boosts performance significantly across multiple metrics compared to baseline GNNs without such priors. The proposed approach highlights the importance of structured inductive biases for gaining improved generalization in complex biological prediction tasks.
Authors: Clayton Sanford, Bahare Fatemi, Ethan Hall, Anton Tsitsulin, Mehran Kazemi, Jonathan Halcrow, Bryan Perozzi, Vahab Mirrokni
Abstract: Which transformer scaling regimes are able to perfectly solve different classes of algorithmic problems? While tremendous empirical advances have been attained by transformer-based neural networks, a theoretical understanding of their algorithmic reasoning capabilities in realistic parameter regimes is lacking. We investigate this question in terms of the network's depth, width, and number of extra tokens for algorithm execution. Our novel representational hierarchy separates 9 algorithmic reasoning problems into classes solvable by transformers in different realistic parameter scaling regimes. We prove that logarithmic depth is necessary and sufficient for tasks like graph connectivity, while single-layer transformers with small embedding dimensions can solve contextual retrieval tasks. We also support our theoretical analysis with ample empirical evidence using the GraphQA benchmark. These results show that transformers excel at many graph reasoning tasks, even outperforming specialized graph neural networks.
Authors: Yunuo Chen, Tianyi Xie, Zeshun Zong, Xuan Li, Feng Gao, Yin Yang, Ying Nian Wu, Chenfanfu Jiang
Abstract: Existing diffusion-based text-to-3D generation methods primarily focus on producing visually realistic shapes and appearances, often neglecting the physical constraints necessary for downstream tasks. Generated models frequently fail to maintain balance when placed in physics-based simulations or 3D printed. This balance is crucial for satisfying user design intentions in interactive gaming, embodied AI, and robotics, where stable models are needed for reliable interaction. Additionally, stable models ensure that 3D-printed objects, such as figurines for home decoration, can stand on their own without requiring additional supports. To fill this gap, we introduce Atlas3D, an automatic and easy-to-implement method that enhances existing Score Distillation Sampling (SDS)-based text-to-3D tools. Atlas3D ensures the generation of self-supporting 3D models that adhere to physical laws of stability under gravity, contact, and friction. Our approach combines a novel differentiable simulation-based loss function with physically inspired regularization, serving as either a refinement or a post-processing module for existing frameworks. We verify Atlas3D's efficacy through extensive generation tasks and validate the resulting 3D models in both simulated and real-world environments.
Authors: Zhang Runquan, Shi Xiaoping
Abstract: In the current field of clinical medicine, traditional methods for analyzing recurrent events have limitations when dealing with complex time-dependent data. This study combines Long Short-Term Memory networks (LSTM) with the Cox model to enhance the model's performance in analyzing recurrent events with dynamic temporal information. Compared to classical models, the LSTM-Cox model significantly improves the accuracy of extracting clinical risk features and exhibits lower Akaike Information Criterion (AIC) values, while maintaining good performance on simulated datasets. In an empirical analysis of bladder cancer recurrence data, the model successfully reduced the mean squared error during the training phase and achieved a Concordance index of up to 0.90 on the test set. Furthermore, the model effectively distinguished between high and low-risk patient groups, and the identified recurrence risk features such as the number of tumor recurrences and maximum size were consistent with other research and clinical trial results. This study not only provides a straightforward and efficient method for analyzing recurrent data and extracting features but also offers a convenient pathway for integrating deep learning techniques into clinical risk prediction systems.
Authors: Yu Luo, Tianying Ji, Fuchun Sun, Jianwei Zhang, Huazhe Xu, Xianyuan Zhan
Abstract: Off-policy reinforcement learning (RL) has achieved notable success in tackling many complex real-world tasks, by leveraging previously collected data for policy learning. However, most existing off-policy RL algorithms fail to maximally exploit the information in the replay buffer, limiting sample efficiency and policy performance. In this work, we discover that concurrently training an offline RL policy based on the shared online replay buffer can sometimes outperform the original online learning policy, though the occurrence of such performance gains remains uncertain. This motivates a new possibility of harnessing the emergent outperforming offline optimal policy to improve online policy learning. Based on this insight, we present Offline-Boosted Actor-Critic (OBAC), a model-free online RL framework that elegantly identifies the outperforming offline policy through value comparison, and uses it as an adaptive constraint to guarantee stronger policy learning performance. Our experiments demonstrate that OBAC outperforms other popular model-free RL baselines and rivals advanced model-based RL methods in terms of sample efficiency and asymptotic performance across 53 tasks spanning 6 task suites.
Authors: Sophia Sun, Wenyuan Chen, Zihao Zhou, Sonia Fereidooni, Elise Jortberg, Rose Yu
Abstract: Mechanical Circulatory Support (MCS) devices, implemented as a probabilistic deep sequence model. Existing mechanical simulators for MCS rely on oversimplifying assumptions and are insensitive to patient-specific behavior, limiting their applicability to real-world treatment scenarios. To address these shortcomings, our model Domain Adversarial Neural Process (DANP) employs a neural process architecture, allowing it to capture the probabilistic relationship between MCS pump levels and aortic pressure measurements with uncertainty. We use domain adversarial training to combine simulation data with real-world observations, resulting in a more realistic and diverse representation of potential outcomes. Empirical results with an improvement of 19% in non-stationary trend prediction establish DANP as an effective tool for clinicians to understand and make informed decisions regarding MCS patient treatment.
Authors: Jiongli Zhu, Su Feng, Boris Glavic, Babak Salimi
Abstract: We introduce an efficient method for learning linear models from uncertain data, where uncertainty is represented as a set of possible variations in the data, leading to predictive multiplicity. Our approach leverages abstract interpretation and zonotopes, a type of convex polytope, to compactly represent these dataset variations, enabling the symbolic execution of gradient descent on all possible worlds simultaneously. We develop techniques to ensure that this process converges to a fixed point and derive closed-form solutions for this fixed point. Our method provides sound over-approximations of all possible optimal models and viable prediction ranges. We demonstrate the effectiveness of our approach through theoretical and empirical analysis, highlighting its potential to reason about model and prediction uncertainty due to data quality issues in training data.
Authors: Trong-Tuan Nguyen, Van-Dat Thang, Nguyen Van Thin, Phuong T. Nguyen
Abstract: The entropy error function has been widely used in neural networks. Nevertheless, the network training based on this error function generally leads to a slow convergence rate, and can easily be trapped in a local minimum or even with the incorrect saturation problem in practice. In fact, there are many results based on entropy error function in neural network and its applications. However, the theory of such an algorithm and its convergence have not been fully studied so far. To tackle the issue, we propose a novel entropy function with smoothing l0 regularization for feed-forward neural networks. Using real-world datasets, we performed an empirical evaluation to demonstrate that the newly conceived algorithm allows us to substantially improve the prediction performance of the considered neural networks. More importantly, the experimental results also show that our proposed function brings in more precise classifications, compared to well-founded baselines. Our work is novel as it enables neural networks to learn effectively, producing more accurate predictions compared to state-of-the-art algorithms. In this respect, we expect that the algorithm will contribute to existing studies in the field, advancing research in Machine Learning and Deep Learning.
Authors: Feiyang Cai, Chuchu Fan, Stanley Bak
Abstract: Verifying safety of neural network control systems that use images as input is a difficult problem because, from a given system state, there is no known way to mathematically model what images are possible in the real-world. We build on recent work that considers a surrogate verification approach, training a conditional generative adversarial network (cGAN) as an image generator in place of the real world. This enables set-based formal analysis of the closed-loop system, providing analysis beyond simulation and testing. While existing work is effective on small examples, excessive overapproximation both within a single control period and across multiple control periods limits its scalability. We propose approaches to overcome these two sources of error. First, we overcome one-step error by composing the system's dynamics along with the cGAN and neural network controller, without losing the dependencies between input states and the control outputs as in the monotonic analysis of the system dynamics. Second, we reduce multi-step error by repeating the single-step composition, essentially unrolling multiple steps of the control loop into a large neural network. We then leverage existing network verification tools to compute accurate reachable sets for multiple steps, avoiding the accumulation of abstraction error at each step. We demonstrate the effectiveness of our approach in terms of both accuracy and scalability using two case studies: an autonomous aircraft taxiing system and an advanced emergency braking system. On the aircraft taxiing system, the converged reachable set is 175% larger using the prior baseline method compared with our proposed approach. On the emergency braking system, with 24x the number of image output variables from the cGAN, the baseline method fails to prove any states are safe, whereas our improvements enable set-based safety analysis.
Authors: Zhiyao Luo, Yangchen Pan, Peter Watkinson, Tingting Zhu
Abstract: In the rapidly changing healthcare landscape, the implementation of offline reinforcement learning (RL) in dynamic treatment regimes (DTRs) presents a mix of unprecedented opportunities and challenges. This position paper offers a critical examination of the current status of offline RL in the context of DTRs. We argue for a reassessment of applying RL in DTRs, citing concerns such as inconsistent and potentially inconclusive evaluation metrics, the absence of naive and supervised learning baselines, and the diverse choice of RL formulation in existing research. Through a case study with more than 17,000 evaluation experiments using a publicly available Sepsis dataset, we demonstrate that the performance of RL algorithms can significantly vary with changes in evaluation metrics and Markov Decision Process (MDP) formulations. Surprisingly, it is observed that in some instances, RL algorithms can be surpassed by random baselines subjected to policy evaluation methods and reward design. This calls for more careful policy evaluation and algorithm development in future DTR works. Additionally, we discussed potential enhancements toward more reliable development of RL-based dynamic treatment regimes and invited further discussion within the community. Code is available at https://github.com/GilesLuo/ReassessDTR.
Authors: Xiangyu Sun, Raquel Aoki, Kevin H. Wilson
Abstract: Machine learning (ML) methods have experienced significant growth in the past decade, yet their practical application in high-impact real-world domains has been hindered by their opacity. When ML methods are responsible for making critical decisions, stakeholders often require insights into how to alter these decisions. Counterfactual explanations (CFEs) have emerged as a solution, offering interpretations of opaque ML models and providing a pathway to transition from one decision to another. However, most existing CFE methods require access to the model's training dataset, few methods can handle multivariate time-series, and none can handle multivariate time-series without training datasets. These limitations can be formidable in many scenarios. In this paper, we present CFWoT, a novel reinforcement-learning-based CFE method that generates CFEs when training datasets are unavailable. CFWoT is model-agnostic and suitable for both static and multivariate time-series datasets with continuous and discrete features. Users have the flexibility to specify non-actionable, immutable, and preferred features, as well as causal constraints which CFWoT guarantees will be respected. We demonstrate the performance of CFWoT against four baselines on several datasets and find that, despite not having access to a training dataset, CFWoT finds CFEs that make significantly fewer and significantly smaller changes to the input time-series. These properties make CFEs more actionable, as the magnitude of change required to alter an outcome is vastly reduced.
Authors: Saswat Das, Marco Romanelli, Cuong Tran, Zarreen Reza, Bhavya Kailkhura, Ferdinando Fioretto
Abstract: Low-rank approximation techniques have become the de facto standard for fine-tuning Large Language Models (LLMs) due to their reduced computational and memory requirements. This paper investigates the effectiveness of these methods in capturing the shift of fine-tuning datasets from the initial pre-trained data distribution. Our findings reveal that there are cases in which low-rank fine-tuning falls short in learning such shifts. This, in turn, produces non-negligible side effects, especially when fine-tuning is adopted for toxicity mitigation in pre-trained models, or in scenarios where it is important to provide fair models. Through comprehensive empirical evidence on several models, datasets, and tasks, we show that low-rank fine-tuning inadvertently preserves undesirable biases and toxic behaviors. We also show that this extends to sequential decision-making tasks, emphasizing the need for careful evaluation to promote responsible LLMs development.
Authors: Zhiyao Luo, Mingcheng Zhu, Fenglin Liu, Jiali Li, Yangchen Pan, Jiandong Zhou, Tingting Zhu
Abstract: Reinforcement learning (RL) has garnered increasing recognition for its potential to optimise dynamic treatment regimes (DTRs) in personalised medicine, particularly for drug dosage prescriptions and medication recommendations. However, a significant challenge persists: the absence of a unified framework for simulating diverse healthcare scenarios and a comprehensive analysis to benchmark the effectiveness of RL algorithms within these contexts. To address this gap, we introduce \textit{DTR-Bench}, a benchmarking platform comprising four distinct simulation environments tailored to common DTR applications, including cancer chemotherapy, radiotherapy, glucose management in diabetes, and sepsis treatment. We evaluate various state-of-the-art RL algorithms across these settings, particularly highlighting their performance amidst real-world challenges such as pharmacokinetic/pharmacodynamic (PK/PD) variability, noise, and missing data. Our experiments reveal varying degrees of performance degradation among RL algorithms in the presence of noise and patient variability, with some algorithms failing to converge. Additionally, we observe that using temporal observation representations does not consistently lead to improved performance in DTR settings. Our findings underscore the necessity of developing robust, adaptive RL algorithms capable of effectively managing these complexities to enhance patient-specific healthcare. We have open-sourced our benchmark and code at https://github.com/GilesLuo/DTR-Bench.
Authors: Abhineet Agarwal, Anish Agarwal, Lorenzo Masoero, Justin Whitehouse
Abstract: Online experimentation with interference is a common challenge in modern applications such as e-commerce and adaptive clinical trials in medicine. For example, in online marketplaces, the revenue of a good depends on discounts applied to competing goods. Statistical inference with interference is widely studied in the offline setting, but far less is known about how to adaptively assign treatments to minimize regret. We address this gap by studying a multi-armed bandit (MAB) problem where a learner (e-commerce platform) sequentially assigns one of possible $\mathcal{A}$ actions (discounts) to $N$ units (goods) over $T$ rounds to minimize regret (maximize revenue). Unlike traditional MAB problems, the reward of each unit depends on the treatments assigned to other units, i.e., there is interference across the underlying network of units. With $\mathcal{A}$ actions and $N$ units, minimizing regret is combinatorially difficult since the action space grows as $\mathcal{A}^N$. To overcome this issue, we study a sparse network interference model, where the reward of a unit is only affected by the treatments assigned to $s$ neighboring units. We use tools from discrete Fourier analysis to develop a sparse linear representation of the unit-specific reward $r_n: [\mathcal{A}]^N \rightarrow \mathbb{R} $, and propose simple, linear regression-based algorithms to minimize regret. Importantly, our algorithms achieve provably low regret both when the learner observes the interference neighborhood for all units and when it is unknown. This significantly generalizes other works on this topic which impose strict conditions on the strength of interference on a known network, and also compare regret to a markedly weaker optimal action. Empirically, we corroborate our theoretical findings via numerical simulations.
Authors: Rahul Madhavan, Aurghya Maiti, Gaurav Sinha, Siddharth Barman
Abstract: We study a variant of causal contextual bandits where the context is chosen based on an initial intervention chosen by the learner. At the beginning of each round, the learner selects an initial action, depending on which a stochastic context is revealed by the environment. Following this, the learner then selects a final action and receives a reward. Given $T$ rounds of interactions with the environment, the objective of the learner is to learn a policy (of selecting the initial and the final action) with maximum expected reward. In this paper we study the specific situation where every action corresponds to intervening on a node in some known causal graph. We extend prior work from the deterministic context setting to obtain simple regret minimization guarantees. This is achieved through an instance-dependent causal parameter, $\lambda$, which characterizes our upper bound. Furthermore, we prove that our simple regret is essentially tight for a large class of instances. A key feature of our work is that we use convex optimization to address the bandit exploration problem. We also conduct experiments to validate our theoretical results, and release our code at our project GitHub repository: https://github.com/adaptiveContextualCausalBandits/aCCB.
URLs: https://github.com/adaptiveContextualCausalBandits/aCCB.
Authors: Sunay Bhat, Jeffrey Jiang, Omead Pooladzandi, Alexander Branch, Gregory Pottie
Abstract: Train-time data poisoning attacks threaten machine learning models by introducing adversarial examples during training, leading to misclassification. Current defense methods often reduce generalization performance, are attack-specific, and impose significant training overhead. To address this, we introduce a set of universal data purification methods using a stochastic transform, $\Psi(x)$, realized via iterative Langevin dynamics of Energy-Based Models (EBMs), Denoising Diffusion Probabilistic Models (DDPMs), or both. These approaches purify poisoned data with minimal impact on classifier generalization. Our specially trained EBMs and DDPMs provide state-of-the-art defense against various attacks (including Narcissus, Bullseye Polytope, Gradient Matching) on CIFAR-10, Tiny-ImageNet, and CINIC-10, without needing attack or classifier-specific information. We discuss performance trade-offs and show that our methods remain highly effective even with poisoned or distributionally shifted generative model training data.
Authors: Hao (Mark), Chen, Wayne Luk, Ka Fai Cedric Yiu, Rui Li, Konstantin Mishchenko, Stylianos I. Venieris, Hongxiang Fan
Abstract: The auto-regressive decoding of Large Language Models (LLMs) results in significant overheads in their hardware performance. While recent research has investigated various speculative decoding techniques for multi-token generation, these efforts have primarily focused on improving processing speed such as throughput. Crucially, they often neglect other metrics essential for real-life deployments, such as memory consumption and training cost. To overcome these limitations, we propose a novel parallel prompt decoding that requires only $0.0002$% trainable parameters, enabling efficient training on a single A100-40GB GPU in just 16 hours. Inspired by the human natural language generation process, $PPD$ approximates outputs generated at future timesteps in parallel by using multiple prompt tokens. This approach partially recovers the missing conditional dependency information necessary for multi-token generation, resulting in up to a 28% higher acceptance rate for long-range predictions. Furthermore, we present a hardware-aware dynamic sparse tree technique that adaptively optimizes this decoding scheme to fully leverage the computational capacities on different GPUs. Through extensive experiments across LLMs ranging from MobileLlama to Vicuna-13B on a wide range of benchmarks, our approach demonstrates up to 2.49$\times$ speedup and maintains a minimal runtime memory overhead of just $0.0004$%. More importantly, our parallel prompt decoding can serve as an orthogonal optimization for synergistic integration with existing speculative decoding, showing up to $1.22\times$ further speed improvement. Our code is available at https://github.com/hmarkc/parallel-prompt-decoding.
Authors: Yifei Wang, Yuyang Wu, Zeming Wei, Stefanie Jegelka, Yisen Wang
Abstract: Going beyond mimicking limited human experiences, recent studies show initial evidence that, like humans, large language models (LLMs) are capable of improving their abilities purely by self-correction, i.e., correcting previous responses through self-examination, in certain circumstances. Nevertheless, little is known about how such capabilities arise. In this work, based on a simplified setup akin to an alignment task, we theoretically analyze self-correction from an in-context learning perspective, showing that when LLMs give relatively accurate self-examinations as rewards, they are capable of refining responses in an in-context way. Notably, going beyond previous theories on over-simplified linear transformers, our theoretical construction underpins the roles of several key designs of realistic transformers for self-correction: softmax attention, multi-head attention, and the MLP block. We validate these findings extensively on synthetic datasets. Inspired by these findings, we also illustrate novel applications of self-correction, such as defending against LLM jailbreaks, where a simple self-correction step does make a large difference. We believe that these findings will inspire further research on understanding, exploiting, and enhancing self-correction for building better foundation models.
Authors: Xuefeng Du, Yiyou Sun, Yixuan Li
Abstract: Detecting data points deviating from the training distribution is pivotal for ensuring reliable machine learning. Extensive research has been dedicated to the challenge, spanning classical anomaly detection techniques to contemporary out-of-distribution (OOD) detection approaches. While OOD detection commonly relies on supervised learning from a labeled in-distribution (ID) dataset, anomaly detection may treat the entire ID data as a single class and disregard ID labels. This fundamental distinction raises a significant question that has yet to be rigorously explored: when and how does ID label help OOD detection? This paper bridges this gap by offering a formal understanding to theoretically delineate the impact of ID labels on OOD detection. We employ a graph-theoretic approach, rigorously analyzing the separability of ID data from OOD data in a closed-form manner. Key to our approach is the characterization of data representations through spectral decomposition on the graph. Leveraging these representations, we establish a provable error bound that compares the OOD detection performance with and without ID labels, unveiling conditions for achieving enhanced OOD detection. Lastly, we present empirical results on both simulated and real datasets, validating theoretical guarantees and reinforcing our insights. Code is publicly available at https://github.com/deeplearning-wisc/id_label.
Authors: Tiansheng Huang, Sihao Hu, Fatih Ilhan, Selim Furkan Tekin, Ling Liu
Abstract: Recent studies show that Large Language Models (LLMs) with safety alignment can be jail-broken by fine-tuning on a dataset mixed with harmful data. First time in the literature, we show that the jail-broken effect can be mitigated by separating states in the finetuning stage to optimize the alignment and user datasets. Unfortunately, our subsequent study shows that this simple Bi-State Optimization (BSO) solution experiences convergence instability when steps invested in its alignment state is too small, leading to downgraded alignment performance. By statistical analysis, we show that the \textit{excess drift} towards consensus could be a probable reason for the instability. To remedy this issue, we propose \textbf{L}azy(\textbf{i}) \textbf{s}afety \textbf{a}lignment (\textbf{Lisa}), which introduces a proximal term to constraint the drift of each state. Theoretically, the benefit of the proximal term is supported by the convergence analysis, wherein we show that a sufficient large proximal factor is necessary to guarantee Lisa's convergence. Empirically, our results on four downstream finetuning tasks show that Lisa with a proximal term can significantly increase alignment performance while maintaining the LLM's accuracy on the user tasks. Code is available at \url{https://github.com/git-disl/Lisa}.
Authors: Ping-Han Hsieh, Ru-Xiu Hsiao, Katalin Ferenc, Anthony Mathelier, Rebekka Burkholz, Chien-Yu Chen, Geir Kjetil Sandve, Tatiana Belova, Marieke Lydia Kuijjer
Abstract: Paired single-cell sequencing technologies enable the simultaneous measurement of complementary modalities of molecular data at single-cell resolution. Along with the advances in these technologies, many methods based on variational autoencoders have been developed to integrate these data. However, these methods do not explicitly incorporate prior biological relationships between the data modalities, which could significantly enhance modeling and interpretation. We propose a novel probabilistic learning framework that explicitly incorporates conditional independence relationships between multi-modal data as a directed acyclic graph using a generalized hierarchical variational autoencoder. We demonstrate the versatility of our framework across various applications pertinent to single-cell multi-omics data integration. These include the isolation of common and distinct information from different modalities, modality-specific differential analysis, and integrated cell clustering. We anticipate that the proposed framework can facilitate the construction of highly flexible graphical models that can capture the complexities of biological hypotheses and unravel the connections between different biological data types, such as different modalities of paired single-cell multi-omics data. The implementation of the proposed framework can be found in the repository https://github.com/kuijjerlab/CAVACHON.
Authors: Deng Pan, Nuno Moniz, Nitesh Chawla
Abstract: A long-standing dilemma prevents the broader application of explanation methods: general applicability and inference speed. On the one hand, existing model-agnostic explanation methods usually make minimal pre-assumptions about the prediction models to be explained. Still, they require additional queries to the model through propagation or back-propagation to approximate the models' behaviors, resulting in slow inference and hindering their use in time-sensitive tasks. On the other hand, various model-dependent explanations have been proposed that achieve low-cost, fast inference but at the expense of limiting their applicability to specific model structures. In this study, we bridge the gap between the universality of model-agnostic approaches and the efficiency of model-specific approaches by proposing a novel framework without assumptions on the prediction model's structures, achieving high efficiency during inference and allowing for real-time explanations. To achieve this, we first define explanations through a set of human-comprehensible concepts and propose a framework to elucidate model predictions via minimal feasible concept sets. Second, we show that a minimal feasible set generator can be learned as a companion explainer to the prediction model, generating explanations for predictions. Finally, we validate this framework by implementing a novel model-agnostic method that provides robust explanations while facilitating real-time inference. Our claims are substantiated by comprehensive experiments, highlighting the effectiveness and efficiency of our approach.
Authors: Vicky Zayats, Peter Chen, Melissa Merrari, Dirk Padfield
Abstract: Integrating multiple generative foundation models, especially those trained on different modalities, into something greater than the sum of its parts poses significant challenges. Two key hurdles are the availability of aligned data (concepts that contain similar meaning but is expressed differently in different modalities), and effectively leveraging unimodal representations in cross-domain generative tasks, without compromising their original unimodal capabilities. We propose Zipper, a multi-tower decoder architecture that addresses these concerns by using cross-attention to flexibly compose multimodal generative models from independently pre-trained unimodal decoders. In our experiments fusing speech and text modalities, we show the proposed architecture performs very competitively in scenarios with limited aligned text-speech data. We also showcase the flexibility of our model to selectively maintain unimodal (e.g., text-to-text generation) generation performance by freezing the corresponding modal tower (e.g. text). In cross-modal tasks such as automatic speech recognition (ASR) where the output modality is text, we show that freezing the text backbone results in negligible performance degradation. In cross-modal tasks such as text-to-speech generation (TTS) where the output modality is speech, we show that using a pre-trained speech backbone results in superior performance to the baseline.
Authors: Kaveh Alimohammadi, Hao Wang, Ojas Gulati, Akash Srivastava, Navid Azizan
Abstract: Existing differentially private (DP) synthetic data generation mechanisms typically assume a single-source table. In practice, data is often distributed across multiple tables with relationships across tables. In this paper, we introduce the first-of-its-kind algorithm that can be combined with any existing DP mechanisms to generate synthetic relational databases. Our algorithm iteratively refines the relationship between individual synthetic tables to minimize their approximation errors in terms of low-order marginal distributions while maintaining referential integrity. Finally, we provide both DP and theoretical utility guarantees for our algorithm.
Authors: Hangzhi Guo, Amulya Yadav
Abstract: The field of Explainable Artificial Intelligence (XAI) focuses on techniques for providing explanations to end-users about the decision-making processes that underlie modern-day machine learning (ML) models. Within the vast universe of XAI techniques, counterfactual (CF) explanations are often preferred by end-users as they help explain the predictions of ML models by providing an easy-to-understand & actionable recourse (or contrastive) case to individual end-users who are adversely impacted by predicted outcomes. However, recent studies have shown significant security concerns with using CF explanations in real-world applications; in particular, malicious adversaries can exploit CF explanations to perform query-efficient model extraction attacks on proprietary ML models. In this paper, we propose a model-agnostic watermarking framework (for adding watermarks to CF explanations) that can be leveraged to detect unauthorized model extraction attacks (which rely on the watermarked CF explanations). Our novel framework solves a bi-level optimization problem to embed an indistinguishable watermark into the generated CF explanation such that any future model extraction attacks that rely on these watermarked CF explanations can be detected using a null hypothesis significance testing (NHST) scheme, while ensuring that these embedded watermarks do not compromise the quality of the generated CF explanations. We evaluate this framework's performance across a diverse set of real-world datasets, CF explanation methods, and model extraction techniques, and show that our watermarking detection system can be used to accurately identify extracted ML models that are trained using the watermarked CF explanations. Our work paves the way for the secure adoption of CF explanations in real-world applications.
Authors: Yuta Tarumi, Keisuke Fukuda, Shin-ichi Maeda
Abstract: State estimation for nonlinear state space models is a challenging task. Existing assimilation methodologies predominantly assume Gaussian posteriors on physical space, where true posteriors become inevitably non-Gaussian. We propose Deep Bayesian Filtering (DBF) for data assimilation on nonlinear state space models (SSMs). DBF constructs new latent variables $h_t$ on a new latent (``fancy'') space and assimilates observations $o_t$. By (i) constraining the state transition on fancy space to be linear and (ii) learning a Gaussian inverse observation operator $q(h_t|o_t)$, posteriors always remain Gaussian for DBF. Quite distinctively, the structured design of posteriors provides an analytic formula for the recursive computation of posteriors without accumulating Monte-Carlo sampling errors over time steps. DBF seeks the Gaussian inverse observation operators $q(h_t|o_t)$ and other latent SSM parameters (e.g., dynamics matrix) by maximizing the evidence lower bound. Experiments show that DBF outperforms model-based approaches and latent assimilation methods in various tasks and conditions.
Authors: Fengshuo Bai, Rui Zhao, Hongming Zhang, Sijia Cui, Ying Wen, Yaodong Yang, Bo Xu, Lei Han
Abstract: Preference-based reinforcement learning (PbRL) has shown impressive capabilities in training agents without reward engineering. However, a notable limitation of PbRL is its dependency on substantial human feedback. This dependency stems from the learning loop, which entails accurate reward learning compounded with value/policy learning, necessitating a considerable number of samples. To boost the learning loop, we propose SEER, an efficient PbRL method that integrates label smoothing and policy regularization techniques. Label smoothing reduces overfitting of the reward model by smoothing human preference labels. Additionally, we bootstrap a conservative estimate $\widehat{Q}$ using well-supported state-action pairs from the current replay memory to mitigate overestimation bias and utilize it for policy learning regularization. Our experimental results across a variety of complex tasks, both in online and offline settings, demonstrate that our approach improves feedback efficiency, outperforming state-of-the-art methods by a large margin. Ablation studies further reveal that SEER achieves a more accurate Q-function compared to prior work.
Authors: Abishek Sriramulu, Nicolas Fourrier, Christoph Bergmeir
Abstract: Graph Neural Networks (GNN) have gained significant traction in the forecasting domain, especially for their capacity to simultaneously account for intra-series temporal correlations and inter-series relationships. This paper introduces a novel Hierarchical GNN (DeepHGNN) framework, explicitly designed for forecasting in complex hierarchical structures. The uniqueness of DeepHGNN lies in its innovative graph-based hierarchical interpolation and an end-to-end reconciliation mechanism. This approach ensures forecast accuracy and coherence across various hierarchical levels while sharing signals across them, addressing a key challenge in hierarchical forecasting. A critical insight in hierarchical time series is the variance in forecastability across levels, with upper levels typically presenting more predictable components. DeepHGNN capitalizes on this insight by pooling and leveraging knowledge from all hierarchy levels, thereby enhancing the overall forecast accuracy. Our comprehensive evaluation set against several state-of-the-art models confirm the superior performance of DeepHGNN. This research not only demonstrates DeepHGNN's effectiveness in achieving significantly improved forecast accuracy but also contributes to the understanding of graph-based methods in hierarchical time series forecasting.
Authors: Dohyeong Kim, Taehyun Cho, Seungyub Han, Hojun Chung, Kyungjae Lee, Songhwai Oh
Abstract: The field of risk-constrained reinforcement learning (RCRL) has been developed to effectively reduce the likelihood of worst-case scenarios by explicitly handling risk-measure-based constraints. However, the nonlinearity of risk measures makes it challenging to achieve convergence and optimality. To overcome the difficulties posed by the nonlinearity, we propose a spectral risk measure-constrained RL algorithm, spectral-risk-constrained policy optimization (SRCPO), a bilevel optimization approach that utilizes the duality of spectral risk measures. In the bilevel optimization structure, the outer problem involves optimizing dual variables derived from the risk measures, while the inner problem involves finding an optimal policy given these dual variables. The proposed method, to the best of our knowledge, is the first to guarantee convergence to an optimum in the tabular setting. Furthermore, the proposed method has been evaluated on continuous control tasks and showed the best performance among other RCRL algorithms satisfying the constraints.
Authors: Xianke Qiang, Zheng Chang, Yun Hu, Lei Liu, Timo Hamalainen
Abstract: Vehicular edge intelligence (VEI) is a promising paradigm for enabling future intelligent transportation systems by accommodating artificial intelligence (AI) at the vehicular edge computing (VEC) system. Federated learning (FL) stands as one of the fundamental technologies facilitating collaborative model training locally and aggregation, while safeguarding the privacy of vehicle data in VEI. However, traditional FL faces challenges in adapting to vehicle heterogeneity, training large models on resource-constrained vehicles, and remaining susceptible to model weight privacy leakage. Meanwhile, split learning (SL) is proposed as a promising collaborative learning framework which can mitigate the risk of model wights leakage, and release the training workload on vehicles. SL sequentially trains a model between a vehicle and an edge cloud (EC) by dividing the entire model into a vehicle-side model and an EC-side model at a given cut layer. In this work, we combine the advantages of SL and FL to develop an Adaptive Split Federated Learning scheme for Vehicular Edge Computing (ASFV). The ASFV scheme adaptively splits the model and parallelizes the training process, taking into account mobile vehicle selection and resource allocation. Our extensive simulations, conducted on non-independent and identically distributed data, demonstrate that the proposed ASFV solution significantly reduces training latency compared to existing benchmarks, while adapting to network dynamics and vehicles' mobility.
Authors: Joonhyung Lee, Jeongin Bae, Byeongwook Kim, Se Jung Kwon, Dongsoo Lee
Abstract: The massive computational costs associated with large language model (LLM) pretraining have spurred great interest in reduced-precision floating-point representations to accelerate the process. As a result, the BrainFloat16 (BF16) precision has become the de facto standard for LLM training, with hardware support included in recent accelerators. This trend has gone even further in the latest processors, where FP8 has recently been introduced. However, prior experience with FP16, which was found to be less stable than BF16, raises concerns as to whether FP8, with even fewer bits than FP16, can be a cost-effective option for LLM training. We argue that reduced-precision training schemes must have similar training stability and hyperparameter sensitivities to their higher-precision counterparts in order to be cost-effective. However, we find that currently available methods for FP8 training are not robust enough to allow their use as economical replacements. This prompts us to investigate the stability of reduced-precision LLM training in terms of robustness across random seeds and learning rates. To this end, we propose new evaluation techniques and a new metric for quantifying loss landscape sharpness in autoregressive language models. By simulating incremental bit reductions in floating-point representations, we analyze the relationship between representational power and training stability with the intent of aiding future research into the field.
Authors: Yonghong Li, Shan Qu, Xiuzhuang Zhou
Abstract: While existing depression recognition methods based on deep learning show promise, their practical application is hindered by the lack of trustworthiness, as these deep models are often deployed as \textit{black box} models, leaving us uncertain about the confidence of the model predictions. For high-risk clinical applications like depression recognition, uncertainty quantification is essential in decision-making. In this paper, we introduce conformal depression prediction (CDP), a depression recognition method with uncertainty quantification based on conformal prediction (CP), giving valid confidence intervals with theoretical coverage guarantees for the model predictions. CDP is a plug-and-play module that requires neither model retraining nor an assumption about the depression data distribution. As CDP provides only an average performance guarantee across all inputs rather than per-input performance guarantee, we propose CDP-ACC, an improved conformal prediction with approximate conditional coverage. CDP-ACC firstly estimates the prediction distribution through neighborhood relaxation, and then introduces a conformal score function by constructing nested sequences, so as to provide tighter prediction interval for each specific input. We empirically demonstrate the application of uncertainty quantification in depression recognition, and the effectiveness and superiority of CDP and CDP-ACC on the AVEC 2013 and AVEC 2014 datasets
Authors: Jiajie Li, Bo Gu, Shimin Gong, Zhou Su, Mohsen Guizani
Abstract: Mobile crowdsensing (MCS) has emerged as a prominent trend across various domains. However, ensuring the quality of the sensing data submitted by mobile users (MUs) remains a complex and challenging problem. To address this challenge, an advanced method is required to detect low-quality sensing data and identify malicious MUs that may disrupt the normal operations of an MCS system. Therefore, this article proposes a prediction- and reputation-based truth discovery (PRBTD) framework, which can separate low-quality data from high-quality data in sensing tasks. First, we apply a correlation-focused spatial-temporal transformer network to predict the ground truth of the input sensing data. Then, we extract the sensing errors of the data as features based on the prediction results to calculate the implications among the data. Finally, we design a reputation-based truth discovery (TD) module for identifying low-quality data with their implications. Given sensing data submitted by MUs, PRBTD can eliminate the data with heavy noise and identify malicious MUs with high accuracy. Extensive experimental results demonstrate that PRBTD outperforms the existing methods in terms of identification accuracy and data quality enhancement.
Authors: Tianle Zhang, Jiayi Guan, Lin Zhao, Yihang Li, Dongjiang Li, Zecui Zeng, Lei Sun, Yue Chen, Xuelong Wei, Lusong Li, Xiaodong He
Abstract: Offline reinforcement learning (RL) aims to learn optimal policies from previously collected datasets. Recently, due to their powerful representational capabilities, diffusion models have shown significant potential as policy models for offline RL issues. However, previous offline RL algorithms based on diffusion policies generally adopt weighted regression to improve the policy. This approach optimizes the policy only using the collected actions and is sensitive to Q-values, which limits the potential for further performance enhancement. To this end, we propose a novel preferred-action-optimized diffusion policy for offline RL. In particular, an expressive conditional diffusion model is utilized to represent the diverse distribution of a behavior policy. Meanwhile, based on the diffusion model, preferred actions within the same behavior distribution are automatically generated through the critic function. Moreover, an anti-noise preference optimization is designed to achieve policy improvement by using the preferred actions, which can adapt to noise-preferred actions for stable training. Extensive experiments demonstrate that the proposed method provides competitive or superior performance compared to previous state-of-the-art offline RL methods, particularly in sparse reward tasks such as Kitchen and AntMaze. Additionally, we empirically prove the effectiveness of anti-noise preference optimization.
Authors: Hirofumi Tsuruta, Hiroyuki Yamazaki, Ryota Maeda, Ryotaro Tamura, Akihiro Imura
Abstract: Antibodies are crucial proteins produced by the immune system to eliminate harmful foreign substances and have become pivotal therapeutic agents for treating human diseases. To accelerate the discovery of antibody therapeutics, there is growing interest in constructing language models using antibody sequences. However, the applicability of pre-trained language models for antibody discovery has not been thoroughly evaluated due to the scarcity of labeled datasets. To overcome these limitations, we introduce AVIDa-SARS-CoV-2, a dataset featuring the antigen-variable domain of heavy chain of heavy chain antibody (VHH) interactions obtained from two alpacas immunized with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike proteins. AVIDa-SARS-CoV-2 includes binary labels indicating the binding or non-binding of diverse VHH sequences to 12 SARS-CoV-2 mutants, such as the Delta and Omicron variants. Furthermore, we release VHHCorpus-2M, a pre-training dataset for antibody language models, containing over two million VHH sequences. We report benchmark results for predicting SARS-CoV-2-VHH binding using VHHBERT pre-trained on VHHCorpus-2M and existing general protein and antibody-specific pre-trained language models. These results confirm that AVIDa-SARS-CoV-2 provides valuable benchmarks for evaluating the representation capabilities of antibody language models for binding prediction, thereby facilitating the development of AI-driven antibody discovery. The datasets are available at https://datasets.cognanous.com.
Authors: Richard H. Moulton, Gary A. McCully, John D. Hastings
Abstract: Reproducibility is a cornerstone of scientific research, enabling validation, extension, and progress. However, the rapidly evolving nature of software and dependencies poses significant challenges to reproducing research results, particularly in fields like adversarial robustness for deep neural networks, where complex codebases and specialized toolkits are utilized. This paper presents a case study of attempting to validate the results on certified adversarial robustness in "SoK: Certified Robustness for Deep Neural Networks" using the VeriGauge toolkit. Despite following the documented methodology, numerous software and hardware compatibility issues were encountered, including outdated or unavailable dependencies, version conflicts, and driver incompatibilities. While a subset of the original results could be run, key findings related to the empirical robust accuracy of various verification methods proved elusive due to these technical obstacles, as well as slight discrepancies in the test results. This practical experience sheds light on the reproducibility crisis afflicting adversarial robustness research, where a lack of reproducibility threatens scientific integrity and hinders progress. The paper discusses the broader implications of this crisis, proposing potential solutions such as containerization, software preservation, and comprehensive documentation practices. Furthermore, it highlights the need for collaboration and standardization efforts within the research community to develop robust frameworks for reproducible research. By addressing the reproducibility crisis head-on, this work aims to contribute to the ongoing discourse on scientific reproducibility and advocate for best practices that ensure the reliability and validity of research findings within not only adversarial robustness, but security and technology research as a whole.
Authors: Yichen Wen, Zhiquan Tan, Kaipeng Zheng, Chuanlong Xie, Weiran Huang
Abstract: Continual learning requires learning incremental tasks with dynamic data distributions. So far, it has been observed that employing a combination of contrastive loss and distillation loss for training in continual learning yields strong performance. To the best of our knowledge, however, this contrastive continual learning framework lacks convincing theoretical explanations. In this work, we fill this gap by establishing theoretical performance guarantees, which reveal how the performance of the model is bounded by training losses of previous tasks in the contrastive continual learning framework. Our theoretical explanations further support the idea that pre-training can benefit continual learning. Inspired by our theoretical analysis of these guarantees, we propose a novel contrastive continual learning algorithm called CILA, which uses adaptive distillation coefficients for different tasks. These distillation coefficients are easily computed by the ratio between average distillation losses and average contrastive losses from previous tasks. Our method shows great improvement on standard benchmarks and achieves new state-of-the-art performance.
Authors: Soochan Lee, Hyeonseong Jeon, Jaehyeon Son, Gunhee Kim
Abstract: In the present era of deep learning, continual learning research is mainly focused on mitigating forgetting when training a neural network with stochastic gradient descent on a non-stationary stream of data. On the other hand, in the more classical literature of statistical machine learning, many models have sequential Bayesian update rules that yield the same learning outcome as the batch training, i.e., they are completely immune to catastrophic forgetting. However, they are often overly simple to model complex real-world data. In this work, we adopt the meta-learning paradigm to combine the strong representational power of neural networks and simple statistical models' robustness to forgetting. In our novel meta-continual learning framework, continual learning takes place only in statistical models via ideal sequential Bayesian update rules, while neural networks are meta-learned to bridge the raw data and the statistical models. Since the neural networks remain fixed during continual learning, they are protected from catastrophic forgetting. This approach not only achieves significantly improved performance but also exhibits excellent scalability. Since our approach is domain-agnostic and model-agnostic, it can be applied to a wide range of problems and easily integrated with existing model architectures.
Authors: Prabhath Reddy Gujavarthy
Abstract: In reinforcement learning, it is often difficult to automate high-dimensional, rapid decision-making in dynamic environments, especially when domains require real-time online interaction and adaptive strategies such as web-based games. This work proposes a state-of-the-art Flexible Deep Q-Network (FDQN) framework that can address this challenge with a selfadaptive approach that is processing high-dimensional sensory data in realtime using a CNN and dynamically adapting the model architecture to varying action spaces of different gaming environments and outperforming previous baseline models in various Atari games and the Chrome Dino game as baselines. Using the epsilon-greedy policy, it effectively balances the new learning and exploitation for improved performance, and it has been designed with a modular structure that it can be easily adapted to other HTML-based games without touching the core part of the framework. It is demonstrated that the FDQN framework can successfully solve a well-defined task in a laboratory condition, but more importantly it also discusses potential applications to more challenging real-world cases and serve as the starting point for future further exploration into automated game play and beyond.
Authors: Wei-Bang Jiang, Li-Ming Zhao, Bao-Liang Lu
Abstract: The current electroencephalogram (EEG) based deep learning models are typically designed for specific datasets and applications in brain-computer interaction (BCI), limiting the scale of the models and thus diminishing their perceptual capabilities and generalizability. Recently, Large Language Models (LLMs) have achieved unprecedented success in text processing, prompting us to explore the capabilities of Large EEG Models (LEMs). We hope that LEMs can break through the limitations of different task types of EEG datasets, and obtain universal perceptual capabilities of EEG signals through unsupervised pre-training. Then the models can be fine-tuned for different downstream tasks. However, compared to text data, the volume of EEG datasets is generally small and the format varies widely. For example, there can be mismatched numbers of electrodes, unequal length data samples, varied task designs, and low signal-to-noise ratio. To overcome these challenges, we propose a unified foundation model for EEG called Large Brain Model (LaBraM). LaBraM enables cross-dataset learning by segmenting the EEG signals into EEG channel patches. Vector-quantized neural spectrum prediction is used to train a semantically rich neural tokenizer that encodes continuous raw EEG channel patches into compact neural codes. We then pre-train neural Transformers by predicting the original neural codes for the masked EEG channel patches. The LaBraMs were pre-trained on about 2,500 hours of various types of EEG signals from around 20 datasets and validated on multiple different types of downstream tasks. Experiments on abnormal detection, event type classification, emotion recognition, and gait prediction show that our LaBraM outperforms all compared SOTA methods in their respective fields. Our code is available at https://github.com/935963004/LaBraM.
Authors: Xinyi Wu, Amir Ajorlou, Yifei Wang, Stefanie Jegelka, Ali Jadbabaie
Abstract: Self-attention is the key mechanism of transformers, which are the essential building blocks of modern foundation models. Recent studies have shown that pure self-attention suffers from an increasing degree of rank collapse as depth increases, limiting model expressivity and further utilization of model depth. The existing literature on rank collapse, however, has mostly overlooked other critical components in transformers that may alleviate the rank collapse issue. In this paper, we provide a general analysis of rank collapse under self-attention, taking into account the effects of attention masks and layer normalization (LayerNorm). In particular, we find that although pure masked attention still suffers from exponential collapse to a rank one subspace, local masked attention can provably slow down the collapse rate. In the case of self-attention with LayerNorm, we first show that for certain classes of value matrices, collapse to a rank one subspace still happens exponentially. However, through construction of nontrivial counterexamples, we then establish that with proper choice of value matrices, a general class of sequences may not converge to a rank one subspace, and the self-attention dynamics with LayerNorm can simultaneously possess a rich set of equilibria with any possible rank between one and full. Our result refutes the previous hypothesis that LayerNorm plays no role in the rank collapse of self-attention and suggests that self-attention with LayerNorm constitutes a much more expressive, versatile nonlinear dynamical system than what was originally thought.
Authors: Hongduan Tian, Feng Liu, Tongliang Liu, Bo Du, Yiu-ming Cheung, Bo Han
Abstract: In cross-domain few-shot classification, \emph{nearest centroid classifier} (NCC) aims to learn representations to construct a metric space where few-shot classification can be performed by measuring the similarities between samples and the prototype of each class. An intuition behind NCC is that each sample is pulled closer to the class centroid it belongs to while pushed away from those of other classes. However, in this paper, we find that there exist high similarities between NCC-learned representations of two samples from different classes. In order to address this problem, we propose a bi-level optimization framework, \emph{maximizing optimized kernel dependence} (MOKD) to learn a set of class-specific representations that match the cluster structures indicated by labeled data of the given task. Specifically, MOKD first optimizes the kernel adopted in \emph{Hilbert-Schmidt independence criterion} (HSIC) to obtain the optimized kernel HSIC (opt-HSIC) that can capture the dependence more precisely. Then, an optimization problem regarding the opt-HSIC is addressed to simultaneously maximize the dependence between representations and labels and minimize the dependence among all samples. Extensive experiments on Meta-Dataset demonstrate that MOKD can not only achieve better generalization performance on unseen domains in most cases but also learn better data representation clusters. The project repository of MOKD is available at: \href{https://github.com/tmlr-group/MOKD}{https://github.com/tmlr-group/MOKD}.
URLs: https://github.com/tmlr-group/MOKD, https://github.com/tmlr-group/MOKD
Authors: Haanvid Lee, Tri Wahyu Guntara, Jongmin Lee, Yung-Kyun Noh, Kee-Eung Kim
Abstract: We consider off-policy evaluation (OPE) of deterministic target policies for reinforcement learning (RL) in environments with continuous action spaces. While it is common to use importance sampling for OPE, it suffers from high variance when the behavior policy deviates significantly from the target policy. In order to address this issue, some recent works on OPE proposed in-sample learning with importance resampling. Yet, these approaches are not applicable to deterministic target policies for continuous action spaces. To address this limitation, we propose to relax the deterministic target policy using a kernel and learn the kernel metrics that minimize the overall mean squared error of the estimated temporal difference update vector of an action value function, where the action value function is used for policy evaluation. We derive the bias and variance of the estimation error due to this relaxation and provide analytic solutions for the optimal kernel metric. In empirical studies using various test domains, we show that the OPE with in-sample learning using the kernel with optimized metric achieves significantly improved accuracy than other baselines.
Authors: Avik Kar, Rahul Singh
Abstract: We study non-episodic Reinforcement Learning for Lipschitz MDPs in which state-action space is a metric space, and the transition kernel and rewards are Lipschitz functions. We develop computationally efficient UCB-based algorithm, $\textit{ZoRL-}\epsilon$ that adaptively discretizes the state-action space and show that their regret as compared with $\epsilon$-optimal policy is bounded as $\mathcal{O}(\epsilon^{-(2 d_\mathcal{S} + d^\epsilon_z + 1)}\log{(T)})$, where $d^\epsilon_z$ is the $\epsilon$-zooming dimension. In contrast, if one uses the vanilla $\textit{UCRL-}2$ on a fixed discretization of the MDP, the regret w.r.t. a $\epsilon$-optimal policy scales as $\mathcal{O}(\epsilon^{-(2 d_\mathcal{S} + d + 1)}\log{(T)})$ so that the adaptivity gains are huge when $d^\epsilon_z \ll d$. Note that the absolute regret of any 'uniformly good' algorithm for a large family of continuous MDPs asymptotically scales as at least $\Omega(\log{(T)})$. Though adaptive discretization has been shown to yield $\mathcal{\tilde{O}}(H^{2.5}K^\frac{d_z + 1}{d_z + 2})$ regret in episodic RL, an attempt to extend this to the non-episodic case by employing constant duration episodes whose duration increases with $T$, is futile since $d_z \to d$ as $T \to \infty$. The current work shows how to obtain adaptivity gains for non-episodic RL. The theoretical results are supported by simulations on two systems where the performance of $\textit{ZoRL-}\epsilon$ is compared with that of '$\textit{UCRL-C}$,' the fixed discretization-based extension of $\textit{UCRL-}2$ for systems with continuous state-action spaces.
Authors: Bart M. N. Smets, Peter D. Donker, Jim W. Portegies, Remco Duits
Abstract: We introduce a class of trainable nonlinear operators based on semirings that are suitable for use in neural networks. These operators generalize the traditional alternation of linear operators with activation functions in neural networks. Semirings are algebraic structures that describe a generalised notation of linearity, greatly expanding the range of trainable operators that can be included in neural networks. In fact, max- or min-pooling operations are convolutions in the tropical semiring with a fixed kernel. We perform experiments where we replace the activation functions for trainable semiring-based operators to show that these are viable operations to include in fully connected as well as convolutional neural networks (ConvNeXt). We discuss some of the challenges of replacing traditional activation functions with trainable semiring activations and the trade-offs of doing so.
Authors: Taehyun Kim, Kwanseok Choi, Youngmock Cho, Jaehoon Cho, Hyuk-Jae Lee, Jaewoong Sim
Abstract: Mixture-of-Experts (MoE) large language models (LLM) have memory requirements that often exceed the GPU memory capacity, requiring costly parameter movement from secondary memories to the GPU for expert computation. In this work, we present Mixture of Near-Data Experts (MoNDE), a near-data computing solution that efficiently enables MoE LLM inference. MoNDE reduces the volume of MoE parameter movement by transferring only the $\textit{hot}$ experts to the GPU, while computing the remaining $\textit{cold}$ experts inside the host memory device. By replacing the transfers of massive expert parameters with the ones of small activations, MoNDE enables far more communication-efficient MoE inference, thereby resulting in substantial speedups over the existing parameter offloading frameworks for both encoder and decoder operations.
Authors: Alain Ryser, Thomas M. Sutter, Alexander Marx, Julia E. Vogt
Abstract: Anomaly Detection focuses on identifying samples that deviate from the norm. When working with high-dimensional data such as images, a crucial requirement for detecting anomalous patterns is learning lower-dimensional representations that capture normal concepts seen during training. Recent advances in self-supervised learning have shown great promise in this regard. However, many of the most successful self-supervised anomaly detection methods assume prior knowledge about the structure of anomalies and leverage synthetic anomalies during training. Yet, in many real-world applications, we do not know what to expect from unseen data, and we can solely leverage knowledge about normal data. In this work, we propose Con2, which addresses this problem by setting normal training data into distinct contexts while preserving its normal properties, letting us observe the data from different perspectives. Unseen normal data consequently adheres to learned context representations while anomalies fail to do so, letting us detect them without any knowledge about anomalies during training. Our experiments demonstrate that our approach achieves state-of-the-art performance on various benchmarks while exhibiting superior performance in a more realistic healthcare setting, where knowledge about potential anomalies is often scarce.
Authors: Vid Han\v{z}el, Bla\v{z} Bertalani\v{c}, Carolina Fortuna
Abstract: Due to growing population and technological advances, global electricity consumption, and consequently also CO2 emissions are increasing. The residential sector makes up 25% of global electricity consumption and has great potential to increase efficiency and reduce CO2 footprint without sacrificing comfort. However, a lack of uniform consumption data at the household level spanning multiple regions hinders large-scale studies and robust multi-region model development. This paper introduces a multi-region dataset compiled from publicly available sources and presented in a uniform format. This data enables machine learning tasks such as disaggregation, demand forecasting, appliance ON/OFF classification, etc. Furthermore, we develop an RDF knowledge graph that characterizes the electricity consumption of the households and contextualizes it with household related properties enabling semantic queries and interoperability with other open knowledge bases like Wikidata and DBpedia. This structured data can be utilized to inform various stakeholders towards data-driven policy and business development.
Authors: Aref Einizade, Fragkiskos D. Malliaros, Jhony H. Giraldo
Abstract: Processing multidomain data defined on multiple graphs holds significant potential in various practical applications in computer science. However, current methods are mostly limited to discrete graph filtering operations. Tensorial partial differential equations on graphs (TPDEGs) provide a principled framework for modeling structured data across multiple interacting graphs, addressing the limitations of the existing discrete methodologies. In this paper, we introduce Continuous Product Graph Neural Networks (CITRUS) that emerge as a natural solution to the TPDEG. CITRUS leverages the separability of continuous heat kernels from Cartesian graph products to efficiently implement graph spectral decomposition. We conduct thorough theoretical analyses of the stability and over-smoothing properties of CITRUS in response to domain-specific graph perturbations and graph spectra effects on the performance. We evaluate CITRUS on well-known traffic and weather spatiotemporal forecasting datasets, demonstrating superior performance over existing approaches.
Authors: Aref Einizade, Fragkiskos D. Malliaros, Jhony H. Giraldo
Abstract: Graph Neural Networks (GNNs) have advanced spatiotemporal forecasting by leveraging relational inductive biases among sensors (or any other measuring scheme) represented as nodes in a graph. However, current methods often rely on Recurrent Neural Networks (RNNs), leading to increased runtimes and memory use. Moreover, these methods typically operate within 1-hop neighborhoods, exacerbating the reduction of the receptive field. Causal Graph Processes (CGPs) offer an alternative, using graph filters instead of MLP layers to reduce parameters and minimize memory consumption. This paper introduces the Causal Graph Process Neural Network (CGProNet), a non-linear model combining CGPs and GNNs for spatiotemporal forecasting. CGProNet employs higher-order graph filters, optimizing the model with fewer parameters, reducing memory usage, and improving runtime efficiency. We present a comprehensive theoretical and experimental stability analysis, highlighting key aspects of CGProNet. Experiments on synthetic and real data demonstrate CGProNet's superior efficiency, minimizing memory and time requirements while maintaining competitive forecasting performance.
Authors: Zhiwei Tang, Jiangweizhi Peng, Jiasheng Tang, Mingyi Hong, Fan Wang, Tsung-Hui Chang
Abstract: In this work, we focus on the alignment problem of diffusion models with a continuous reward function, which represents specific objectives for downstream tasks, such as improving human preference. The central goal of the alignment problem is to adjust the distribution learned by diffusion models such that the generated samples maximize the target reward function. We propose a novel alignment approach, named Direct Noise Optimization (DNO), that optimizes the injected noise during the sampling process of diffusion models. By design, DNO is tuning-free and prompt-agnostic, as the alignment occurs in an online fashion during generation. We rigorously study the theoretical properties of DNO and also propose variants to deal with non-differentiable reward functions. Furthermore, we identify that naive implementation of DNO occasionally suffers from the out-of-distribution reward hacking problem, where optimized samples have high rewards but are no longer in the support of the pretrained distribution. To remedy this issue, we leverage classical high-dimensional statistics theory and propose to augment the DNO loss with certain probability regularization. We conduct extensive experiments on several popular reward functions trained on human feedback data and demonstrate that the proposed DNO approach achieves state-of-the-art reward scores as well as high image quality, all within a reasonable time budget for generation.
Authors: Rajarshi Saha, Naomi Sagan, Varun Srivastava, Andrea J. Goldsmith, Mert Pilanci
Abstract: The prohibitive sizes of Large Language Models (LLMs) today make it difficult to deploy them on memory-constrained edge devices. This work introduces $\rm CALDERA$ -- a new post-training LLM compression algorithm that harnesses the inherent low-rank structure of a weight matrix $\mathbf{W}$ by approximating it via a low-rank, low-precision decomposition as $\mathbf{W} \approx \mathbf{Q} + \mathbf{L}\mathbf{R}$. Here, $\mathbf{L}$ and $\mathbf{R}$ are low rank factors, and the entries of $\mathbf{Q}$, $\mathbf{L}$ and $\mathbf{R}$ are quantized. The model is compressed by substituting each layer with its $\mathbf{Q} + \mathbf{L}\mathbf{R}$ decomposition, and the zero-shot performance of the compressed model is evaluated. Additionally, $\mathbf{L}$ and $\mathbf{R}$ are readily amenable to low-rank adaptation, consequently enhancing the zero-shot performance. $\rm CALDERA$ obtains this decomposition by formulating it as an optimization problem $\min_{\mathbf{Q},\mathbf{L},\mathbf{R}}\lVert(\mathbf{Q} + \mathbf{L}\mathbf{R} - \mathbf{W})\mathbf{X}^\top\rVert_{\rm F}^2$, where $\mathbf{X}$ is the calibration data, and $\mathbf{Q}, \mathbf{L}, \mathbf{R}$ are constrained to be representable using low-precision formats. Theoretical upper bounds on the approximation error of $\rm CALDERA$ are established using a rank-constrained regression framework, and the tradeoff between compression ratio and model performance is studied by analyzing the impact of target rank and quantization bit budget. Results illustrate that compressing LlaMa-$2$ $7$B/$70$B and LlaMa-$3$ $8$B models obtained using $\rm CALDERA$ outperforms existing post-training LLM compression techniques in the regime of less than $2.5$ bits per parameter. The implementation is available at: \href{https://github.com/pilancilab/caldera}{https://github.com/pilancilab/caldera}.
URLs: https://github.com/pilancilab/caldera, https://github.com/pilancilab/caldera
Authors: Ziqing Fan, Shengchao Hu, Jiangchao Yao, Gang Niu, Ya Zhang, Masashi Sugiyama, Yanfeng Wang
Abstract: In federated learning (FL), the multi-step update and data heterogeneity among clients often lead to a loss landscape with sharper minima, degenerating the performance of the resulted global model. Prevalent federated approaches incorporate sharpness-aware minimization (SAM) into local training to mitigate this problem. However, the local loss landscapes may not accurately reflect the flatness of global loss landscape in heterogeneous environments; as a result, minimizing local sharpness and calculating perturbations on client data might not align the efficacy of SAM in FL with centralized training. To overcome this challenge, we propose FedLESAM, a novel algorithm that locally estimates the direction of global perturbation on client side as the difference between global models received in the previous active and current rounds. Besides the improved quality, FedLESAM also speed up federated SAM-based approaches since it only performs once backpropagation in each iteration. Theoretically, we prove a slightly tighter bound than its original FedSAM by ensuring consistent perturbation. Empirically, we conduct comprehensive experiments on four federated benchmark datasets under three partition strategies to demonstrate the superior performance and efficiency of FedLESAM.
Authors: Soyed Tuhin Ahmed, Mehdi Tahoori
Abstract: The performance of deep learning algorithms such as neural networks (NNs) has increased tremendously recently, and they can achieve state-of-the-art performance in many domains. However, due to memory and computation resource constraints, implementing NNs on edge devices is a challenging task. Therefore, hardware accelerators such as computation-in-memory (CIM) with memristive devices have been developed to accelerate the most common operations, i.e., matrix-vector multiplication. However, due to inherent device properties, external environmental factors such as temperature, and an immature fabrication process, memristors suffer from various non-idealities, including defects and variations occurring during manufacturing and runtime. Consequently, there is a lack of complete confidence in the predictions made by the model. To improve confidence in NN predictions made by hardware accelerators in the presence of device non-idealities, in this paper, we propose a Bayesian test vector generation framework that can estimate the model uncertainty of NNs implemented on memristor-based CIM hardware. Compared to the conventional point estimate test vector generation method, our method is more generalizable across different model dimensions and requires storing only one test Bayesian vector in the hardware. Our method is evaluated on different model dimensions, tasks, fault rates, and variation noise to show that it can consistently achieve $100\%$ coverage with only $0.024$ MB of memory overhead.
Authors: Julia Reuter, Viktor Martinek, Roland Herzog, Sanaz Mostaghim
Abstract: When developing empirical equations, domain experts require these to be accurate and adhere to physical laws. Often, constants with unknown units need to be discovered alongside the equations. Traditional unit-aware genetic programming (GP) approaches cannot be used when unknown constants with undetermined units are included. This paper presents a method for dimensional analysis that propagates unknown units as ''jokers'' and returns the magnitude of unit violations. We propose three methods, namely evolutive culling, a repair mechanism, and a multi-objective approach, to integrate the dimensional analysis in the GP algorithm. Experiments on datasets with ground truth demonstrate comparable performance of evolutive culling and the multi-objective approach to a baseline without dimensional analysis. Extensive analysis of the results on datasets without ground truth reveals that the unit-aware algorithms make only low sacrifices in accuracy, while producing unit-adherent solutions. Overall, we presented a promising novel approach for developing unit-adherent empirical equations.
Authors: Filippo Palomba, Andrea Pugnana, Jos\'e Manuel Alvarez, Salvatore Ruggieri
Abstract: Deferring systems extend supervised Machine Learning (ML) models with the possibility to defer predictions to human experts. However, evaluating the impact of a deferring strategy on system accuracy is still an overlooked area. This paper fills this gap by evaluating deferring systems through a causal lens. We link the potential outcomes framework for causal inference with deferring systems. This allows us to identify the causal impact of the deferring strategy on predictive accuracy. We distinguish two scenarios. In the first one, we can access both the human and the ML model predictions for the deferred instances. In such a case, we can identify the individual causal effects for deferred instances and aggregates of them. In the second scenario, only human predictions are available for the deferred instances. In this case, we can resort to regression discontinuity design to estimate a local causal effect. We empirically evaluate our approach on synthetic and real datasets for seven deferring systems from the literature.
Authors: Boje Deforce, Bart Baesens, Estefan\'ia Serral Asensio
Abstract: The recent surge in foundation models for natural language processing and computer vision has fueled innovation across various domains. Inspired by this progress, we explore the potential of foundation models for time-series forecasting in smart agriculture, a field often plagued by limited data availability. Specifically, this work presents a novel application of $\texttt{TimeGPT}$, a state-of-the-art (SOTA) time-series foundation model, to predict soil water potential ($\psi_\mathrm{soil}$), a key indicator of field water status that is typically used for irrigation advice. Traditionally, this task relies on a wide array of input variables. We explore $\psi_\mathrm{soil}$'s ability to forecast $\psi_\mathrm{soil}$ in: ($i$) a zero-shot setting, ($ii$) a fine-tuned setting relying solely on historic $\psi_\mathrm{soil}$ measurements, and ($iii$) a fine-tuned setting where we also add exogenous variables to the model. We compare $\texttt{TimeGPT}$'s performance to established SOTA baseline models for forecasting $\psi_\mathrm{soil}$. Our results demonstrate that $\texttt{TimeGPT}$ achieves competitive forecasting accuracy using only historical $\psi_\mathrm{soil}$ data, highlighting its remarkable potential for agricultural applications. This research paves the way for foundation time-series models for sustainable development in agriculture by enabling forecasting tasks that were traditionally reliant on extensive data collection and domain expertise.
Authors: N\'uria Armengol Urp\'i, Marco Bagatella, Marin Vlastelica, Georg Martius
Abstract: Offline data are both valuable and practical resources for teaching robots complex behaviors. Ideally, learning agents should not be constrained by the scarcity of available demonstrations, but rather generalize beyond the training distribution. However, the complexity of real-world scenarios typically requires huge amounts of data to prevent neural network policies from picking up on spurious correlations and learning non-causal relationships. We propose CAIAC, a data augmentation method that can create feasible synthetic transitions from a fixed dataset without having access to online environment interactions. By utilizing principled methods for quantifying causal influence, we are able to perform counterfactual reasoning by swapping $\it{action}$-unaffected parts of the state-space between independent trajectories in the dataset. We empirically show that this leads to a substantial increase in robustness of offline learning algorithms against distributional shift.
Authors: Ioannis Emiris, Dimitris Fotakis, Giorgos Giannopoulos, Dimitrios Gunopulos, Loukas Kavouras, Kleopatra Markou, Eleni Psaroudaki, Dimitrios Rontogiannis, Dimitris Sacharidis, Nikolaos Theologitis, Dimitrios Tomaras, Konstantinos Tsopelas
Abstract: Counterfactual explanations have emerged as an important tool to understand, debug, and audit complex machine learning models. To offer global counterfactual explainability, state-of-the-art methods construct summaries of local explanations, offering a trade-off among conciseness, counterfactual effectiveness, and counterfactual cost or burden imposed on instances. In this work, we provide a concise formulation of the problem of identifying global counterfactuals and establish principled criteria for comparing solutions, drawing inspiration from Pareto dominance. We introduce innovative algorithms designed to address the challenge of finding global counterfactuals for either the entire input space or specific partitions, employing clustering and decision trees as key components. Additionally, we conduct a comprehensive experimental evaluation, considering various instances of the problem and comparing our proposed algorithms with state-of-the-art methods. The results highlight the consistent capability of our algorithms to generate meaningful and interpretable global counterfactual explanations.
Authors: Giuseppe Serra, Florian Buettner
Abstract: Given the ability to model more realistic and dynamic problems, Federated Continual Learning (FCL) has been increasingly investigated recently. A well-known problem encountered in this setting is the so-called catastrophic forgetting, for which the learning model is inclined to focus on more recent tasks while forgetting the previously learned knowledge. The majority of the current approaches in FCL propose generative-based solutions to solve said problem. However, this setting requires multiple training epochs over the data, implying an offline setting where datasets are stored locally and remain unchanged over time. Furthermore, the proposed solutions are tailored for vision tasks solely. To overcome these limitations, we propose a new modality-agnostic approach to deal with the online scenario where new data arrive in streams of mini-batches that can only be processed once. To solve catastrophic forgetting, we propose an uncertainty-aware memory-based approach. In particular, we suggest using an estimator based on the Bregman Information (BI) to compute the model's variance at the sample level. Through measures of predictive uncertainty, we retrieve samples with specific characteristics, and - by retraining the model on such samples - we demonstrate the potential of this approach to reduce the forgetting effect in realistic settings.
Authors: Yufei Zhaoa, Shiduo Wanga, Hua Duana
Abstract: Existing heterogeneous graph neural network algorithms (HGNNs) mostly rely on meta-paths to capture the rich semantic information contained in heterogeneous graphs (also known as heterogeneous information networks (HINs)), but most of these HGNNs focus on different ways of feature aggre gation and ignore the properties of the meta-paths themselves. This paper studies meta-paths in three commonly used data sets and finds that there are huge differences in the number of neighbors connected by different meta paths. At the same time, the noise information contained in large neigh bor paths will have an adverse impact on model performance. Therefore, this paper proposes a Heterogeneous Graph Neural Network Classification and Aggregation Algorithm Based on Large and Small Neighbor Path Iden tification(LSPI). LSPI firstly divides the meta-paths into large and small neighbor paths through the path discriminator , and in order to reduce the noise interference problem in large neighbor paths, LSPI selects neighbor nodes with higher similarity from both topology and feature perspectives, and passes small neighbor paths and filtered large neighbor paths through different graph convolution components. Aggregation is performed to obtain feature information under different subgraphs, and then LSPI uses subgraph level attention to fuse the feature information under different subgraphs to generate the final node embedding. Finally this paper verifies the superiority of the method through extensive experiments and also gives suggestions on the number of nodes to be retained in large neighbor paths through exper iments. The complete reproducible code adn data has been published at: https://github.com/liuhua811/LSPIA.
Authors: Yihang She, Clement Atzberger, Andrew Blake, Adriano Gualandi, Srinivasan Keshav
Abstract: Scientists often model physical processes to understand the natural world and uncover the causation behind observations. Due to unavoidable simplification, discrepancies often arise between model predictions and actual observations, in the form of systematic biases, whose impact varies with model completeness. Classical model inversion methods such as Bayesian inference or regressive neural networks tend either to overlook biases or make assumptions about their nature during data preprocessing, potentially leading to implausible results. Inspired by recent work in inverse graphics, we replace the decoder stage of a standard autoencoder with a physical model followed by a bias-correction layer. This generalisable approach simultaneously inverts the model and corrects its biases in an end-to-end manner without making strong assumptions about the nature of the biases. We demonstrate the effectiveness of our approach using two physical models from disparate domains: a complex radiative transfer model from remote sensing; and a volcanic deformation model from geodesy. Our method matches or surpasses results from classical approaches without requiring biases to be explicitly filtered out, suggesting an effective pathway for understanding the causation of various physical processes.
Authors: Ziqing Fan, Ruipeng Zhang, Jiangchao Yao, Bo Han, Ya Zhang, Yanfeng Wang
Abstract: Partially class-disjoint data (PCDD), a common yet under-explored data formation where each client contributes a part of classes (instead of all classes) of samples, severely challenges the performance of federated algorithms. Without full classes, the local objective will contradict the global objective, yielding the angle collapse problem for locally missing classes and the space waste problem for locally existing classes. As far as we know, none of the existing methods can intrinsically mitigate PCDD challenges to achieve holistic improvement in the bilateral views (both global view and local view) of federated learning. To address this dilemma, we are inspired by the strong generalization of simplex Equiangular Tight Frame~(ETF) on the imbalanced data, and propose a novel approach called FedGELA where the classifier is globally fixed as a simplex ETF while locally adapted to the personal distributions. Globally, FedGELA provides fair and equal discrimination for all classes and avoids inaccurate updates of the classifier, while locally it utilizes the space of locally missing classes for locally existing classes. We conduct extensive experiments on a range of datasets to demonstrate that our FedGELA achieves promising performance~(averaged improvement of 3.9% to FedAvg and 1.5% to best baselines) and provide both local and global convergence guarantees. Source code is available at:https://github.com/MediaBrain-SJTU/FedGELA.git.
Authors: Yanru Sun, Zongxia Xie, Dongyue Chen, Emadeldeen Eldele, Qinghua Hu
Abstract: Deep learning has significantly advanced time series forecasting through its powerful capacity to capture sequence relationships. However, training these models with the Mean Square Error (MSE) loss often results in over-smooth predictions, making it challenging to handle the complexity and learn high-entropy features from time series data with high variability and unpredictability. In this work, we introduce a novel approach by tokenizing time series values to train forecasting models via cross-entropy loss, while considering the continuous nature of time series data. Specifically, we propose Hierarchical Classification Auxiliary Network, HCAN, a general model-agnostic component that can be integrated with any forecasting model. HCAN is based on a Hierarchy-Aware Attention module that integrates multi-granularity high-entropy features at different hierarchy levels. At each level, we assign a class label for timesteps to train an Uncertainty-Aware Classifier. This classifier mitigates the over-confidence in softmax loss via evidence theory. We also implement a Hierarchical Consistency Loss to maintain prediction consistency across hierarchy levels. Extensive experiments integrating HCAN with state-of-the-art forecasting models demonstrate substantial improvements over baselines on several real-world datasets. Code is available at:https://github.com/syrGitHub/HCAN.
Authors: Renchunzi Xie, Ambroise Odonnat, Vasilii Feofanov, Weijian Deng, Jianfeng Zhang, Bo An
Abstract: Leveraging the models' outputs, specifically the logits, is a common approach to estimating the test accuracy of a pre-trained neural network on out-of-distribution (OOD) samples without requiring access to the corresponding ground truth labels. Despite their ease of implementation and computational efficiency, current logit-based methods are vulnerable to overconfidence issues, leading to prediction bias, especially under the natural shift. In this work, we first study the relationship between logits and generalization performance from the view of low-density separation assumption. Our findings motivate our proposed method MaNo which (1) applies a data-dependent normalization on the logits to reduce prediction bias, and (2) takes the $L_p$ norm of the matrix of normalized logits as the estimation score. Our theoretical analysis highlights the connection between the provided score and the model's uncertainty. We conduct an extensive empirical study on common unsupervised accuracy estimation benchmarks and demonstrate that MaNo achieves state-of-the-art performance across various architectures in the presence of synthetic, natural, or subpopulation shifts.
Authors: Ziqing Fan, Jiangchao Yao, Ruipeng Zhang, Lingjuan Lyu, Ya Zhang, Yanfeng Wang
Abstract: Statistical heterogeneity severely limits the performance of federated learning (FL), motivating several explorations e.g., FedProx, MOON and FedDyn, to alleviate this problem. Despite effectiveness, their considered scenario generally requires samples from almost all classes during the local training of each client, although some covariate shifts may exist among clients. In fact, the natural case of partially class-disjoint data (PCDD), where each client contributes a few classes (instead of all classes) of samples, is practical yet underexplored. Specifically, the unique collapse and invasion characteristics of PCDD can induce the biased optimization direction in local training, which prevents the efficiency of federated learning. To address this dilemma, we propose a manifold reshaping approach called FedMR to calibrate the feature space of local training. Our FedMR adds two interplaying losses to the vanilla federated learning: one is intra-class loss to decorrelate feature dimensions for anti-collapse; and the other one is inter-class loss to guarantee the proper margin among categories in the feature expansion. We conduct extensive experiments on a range of datasets to demonstrate that our FedMR achieves much higher accuracy and better communication efficiency. Source code is available at: https://github.com/MediaBrain-SJTU/FedMR.git.
Authors: Zijiang Yan, Ramsundar Tanikella, Hina Tabassum
Abstract: In vehicular networks (VNets), ensuring both road safety and dependable network connectivity is of utmost importance. Achieving this necessitates the creation of resilient and efficient decision-making policies that prioritize multiple objectives. In this paper, we develop a Variational Quantum Circuit (VQC)-based multi-objective reinforcement learning (MORL) framework to characterize efficient network selection and autonomous driving policies in a vehicular network (VNet). Numerical results showcase notable enhancements in both convergence rates and rewards when compared to conventional deep-Q networks (DQNs), validating the efficacy of the VQC-MORL solution.
Authors: Minji Lee, Luiz Felipe Vecchietti, Hyunkyu Jung, Hyun Joo Ro, Meeyoung Cha, Ho Min Kim
Abstract: Proteins are complex molecules responsible for different functions in nature. Enhancing the functionality of proteins and cellular fitness can significantly impact various industries. However, protein optimization using computational methods remains challenging, especially when starting from low-fitness sequences. We propose LatProtRL, an optimization method to efficiently traverse a latent space learned by an encoder-decoder leveraging a large protein language model. To escape local optima, our optimization is modeled as a Markov decision process using reinforcement learning acting directly in latent space. We evaluate our approach on two important fitness optimization tasks, demonstrating its ability to achieve comparable or superior fitness over baseline methods. Our findings and in vitro evaluation show that the generated sequences can reach high-fitness regions, suggesting a substantial potential of LatProtRL in lab-in-the-loop scenarios.
Authors: Fan Zhang, Carlos Esteve-Yag\"ue, S\"oren Dittmer, Carola-Bibiane Sch\"onlieb, Michael Roberts
Abstract: Federated Learning (FL) enables collaborative training of machine learning models on decentralized data while preserving data privacy. However, data across clients often differs significantly due to class imbalance, feature distribution skew, sample size imbalance, and other phenomena. Leveraging information from these not identically distributed (non-IID) datasets poses substantial challenges. FL methods based on a single global model cannot effectively capture the variations in client data and underperform in non-IID settings. Consequently, Personalized FL (PFL) approaches that adapt to each client's data distribution but leverage other clients' data are essential but currently underexplored. We propose a novel Bayesian PFL framework using bi-level optimization to tackle the data heterogeneity challenges. Our proposed framework utilizes the global model as a prior distribution within a Maximum A Posteriori (MAP) estimation of personalized client models. This approach facilitates PFL by integrating shared knowledge from the prior, thereby enhancing local model performance, generalization ability, and communication efficiency. We extensively evaluated our bi-level optimization approach on real-world and synthetic datasets, demonstrating significant improvements in model accuracy compared to existing methods while reducing communication overhead. This study contributes to PFL by establishing a solid theoretical foundation for the proposed method and offering a robust, ready-to-use framework that effectively addresses the challenges posed by non-IID data in FL.
Authors: Jens P\"uttschneider, Timm Faulwasser
Abstract: The training of ResNets and neural ODEs can be formulated and analyzed from the perspective of optimal control. This paper proposes a dissipative formulation of the training of ResNets and neural ODEs for classification problems by including a variant of the cross-entropy as a regularization in the stage cost. Based on the dissipative formulation of the training, we prove that the trained ResNet exhibit the turnpike phenomenon. We then illustrate that the training exhibits the turnpike phenomenon by training on the two spirals and MNIST datasets. This can be used to find very shallow networks suitable for a given classification task.
Authors: Bernd Frauenknecht, Artur Eisele, Devdutt Subhasish, Friedrich Solowjow, Sebastian Trimpe
Abstract: Dyna-style model-based reinforcement learning (MBRL) combines model-free agents with predictive transition models through model-based rollouts. This combination raises a critical question: 'When to trust your model?'; i.e., which rollout length results in the model providing useful data? Janner et al. (2019) address this question by gradually increasing rollout lengths throughout the training. While theoretically tempting, uniform model accuracy is a fallacy that collapses at the latest when extrapolating. Instead, we propose asking the question 'Where to trust your model?'. Using inherent model uncertainty to consider local accuracy, we obtain the Model-Based Actor-Critic with Uncertainty-Aware Rollout Adaption (MACURA) algorithm. We propose an easy-to-tune rollout mechanism and demonstrate substantial improvements in data efficiency and performance compared to state-of-the-art deep MBRL methods on the MuJoCo benchmark.
Authors: Danil Provodin, Maurits Kaptein, Mykola Pechenizkiy
Abstract: We present a new algorithm based on posterior sampling for learning in Constrained Markov Decision Processes (CMDP) in the infinite-horizon undiscounted setting. The algorithm achieves near-optimal regret bounds while being advantageous empirically compared to the existing algorithms. Our main theoretical result is a Bayesian regret bound for each cost component of $\tilde{O} (DS\sqrt{AT})$ for any communicating CMDP with $S$ states, $A$ actions, and diameter $D$. This regret bound matches the lower bound in order of time horizon $T$ and is the best-known regret bound for communicating CMDPs achieved by a computationally tractable algorithm. Empirical results show that our posterior sampling algorithm outperforms the existing algorithms for constrained reinforcement learning.
Authors: Emmanouil Krasanakis, Symeon Papadopoulos
Abstract: Creating fair AI systems is a complex problem that involves the assessment of context-dependent bias concerns. Existing research and programming libraries express specific concerns as measures of bias that they aim to constrain or mitigate. In practice, one should explore a wide variety of (sometimes incompatible) measures before deciding which ones warrant corrective action, but their narrow scope means that most new situations can only be examined after devising new measures. In this work, we present a mathematical framework that distils literature measures of bias into building blocks, hereby facilitating new combinations to cover a wide range of fairness concerns, such as classification or recommendation differences across multiple multi-value sensitive attributes (e.g., many genders and races, and their intersections). We show how this framework generalizes existing concepts and present frequently used blocks. We provide an open-source implementation of our framework as a Python library, called FairBench, that facilitates systematic and extensible exploration of potential bias concerns.
Authors: Mustafa Mert \c{C}elikok, Frans A. Oliehoek, Jan-Willem van de Meent
Abstract: We consider inverse reinforcement learning problems with concave utilities. Concave Utility Reinforcement Learning (CURL) is a generalisation of the standard RL objective, which employs a concave function of the state occupancy measure, rather than a linear function. CURL has garnered recent attention for its ability to represent instances of many important applications including the standard RL such as imitation learning, pure exploration, constrained MDPs, offline RL, human-regularized RL, and others. Inverse reinforcement learning is a powerful paradigm that focuses on recovering an unknown reward function that can rationalize the observed behaviour of an agent. There has been recent theoretical advances in inverse RL where the problem is formulated as identifying the set of feasible reward functions. However, inverse RL for CURL problems has not been considered previously. In this paper we show that most of the standard IRL results do not apply to CURL in general, since CURL invalidates the classical Bellman equations. This calls for a new theoretical framework for the inverse CURL problem. Using a recent equivalence result between CURL and Mean-field Games, we propose a new definition for the feasible rewards for I-CURL by proving that this problem is equivalent to an inverse game theory problem in a subclass of mean-field games. We present initial query and sample complexity results for the I-CURL problem under assumptions such as Lipschitz-continuity. Finally, we outline future directions and applications in human--AI collaboration enabled by our results.
Authors: Andrew Zhao, Quentin Xu, Matthieu Lin, Shenzhi Wang, Yong-jin Liu, Zilong Zheng, Gao Huang
Abstract: Recent advances in large language models (LLMs) have made them indispensable, raising significant concerns over managing their safety. Automated red teaming offers a promising alternative to the labor-intensive and error-prone manual probing for vulnerabilities, providing more consistent and scalable safety evaluations. However, existing approaches often compromise diversity by focusing on maximizing attack success rate. Additionally, methods that decrease the cosine similarity from historical embeddings with semantic diversity rewards lead to novelty stagnation as history grows. To address these issues, we introduce DiveR-CT, which relaxes conventional constraints on the objective and semantic reward, granting greater freedom for the policy to enhance diversity. Our experiments demonstrate DiveR-CT's marked superiority over baselines by 1) generating data that perform better in various diversity metrics across different attack success rate levels, 2) better-enhancing resiliency in blue team models through safety tuning based on collected data, 3) allowing dynamic control of objective weights for reliable and controllable attack success rates, and 4) reducing susceptibility to reward overoptimization. Project details and code can be found at https://andrewzh112.github.io/#diverct.
Authors: Yuxi Han, Jihe Wang, Danghui Wang
Abstract: Graph Neural Networks (GNNs) are computationally demanding and inefficient when applied to graph classification tasks in resource-constrained edge scenarios due to their inherent process, involving multiple rounds of forward and backward propagation. As a lightweight alternative, Hyper-Dimensional Computing (HDC), which leverages high-dimensional vectors for data encoding and processing, offers a more efficient solution by addressing computational bottleneck. However, current HDC methods primarily focus on static graphs and neglect to effectively capture node attributes and structural information, which leads to poor accuracy. In this work, we propose CiliaGraph, an enhanced expressive yet ultra-lightweight HDC model for graph classification. This model introduces a novel node encoding strategy that preserves relative distance isomorphism for accurate node connection representation. In addition, node distances are utilized as edge weights for information aggregation, and the encoded node attributes and structural information are concatenated to obtain a comprehensive graph representation. Furthermore, we explore the relationship between orthogonality and dimensionality to reduce the dimensions, thereby further enhancing computational efficiency. Compared to the SOTA GNNs, extensive experiments show that CiliaGraph reduces memory usage and accelerates training speed by an average of 292 times(up to 2341 times) and 103 times(up to 313 times) respectively while maintaining comparable accuracy.
Authors: Jeffery Dick, Saptarshi Nath, Christos Peridis, Eseoghene Benjamin, Soheil Kolouri, Andrea Soltoggio
Abstract: Context detection involves labeling segments of an online stream of data as belonging to different tasks. Task labels are used in lifelong learning algorithms to perform consolidation or other procedures that prevent catastrophic forgetting. Inferring task labels from online experiences remains a challenging problem. Most approaches assume finite and low-dimension observation spaces or a preliminary training phase during which task labels are learned. Moreover, changes in the transition or reward functions can be detected only in combination with a policy, and therefore are more difficult to detect than changes in the input distribution. This paper presents an approach to learning both policies and labels in an online deep reinforcement learning setting. The key idea is to use distance metrics, obtained via optimal transport methods, i.e., Wasserstein distance, on suitable latent action-reward spaces to measure distances between sets of data points from past and current streams. Such distances can then be used for statistical tests based on an adapted Kolmogorov-Smirnov calculation to assign labels to sequences of experiences. A rollback procedure is introduced to learn multiple policies by ensuring that only the appropriate data is used to train the corresponding policy. The combination of task detection and policy deployment allows for the optimization of lifelong reinforcement learning agents without an oracle that provides task labels. The approach is tested using two benchmarks and the results show promising performance when compared with related context detection algorithms. The results suggest that optimal transport statistical methods provide an explainable and justifiable procedure for online context detection and reward optimization in lifelong reinforcement learning.
Authors: Dorina Weichert, Alexander Kister, Patrick Link, Sebastian Houben, Gunar Ernis
Abstract: The practical use of Bayesian Optimization (BO) in engineering applications imposes special requirements: high sampling efficiency on the one hand and finding a robust solution on the other hand. We address the case of adversarial robustness, where all parameters are controllable during the optimization process, but a subset of them is uncontrollable or even adversely perturbed at the time of application. To this end, we develop an efficient information-based acquisition function that we call Robust Entropy Search (RES). We empirically demonstrate its benefits in experiments on synthetic and real-life data. The results showthat RES reliably finds robust optima, outperforming state-of-the-art algorithms.
Authors: Lanting Fang, Yulian Yang, Kai Wang, Shanshan Feng, Kaiyu Feng, Jie Gui, Shuliang Wang, Yew-Soon Ong
Abstract: While dynamic graph neural networks have shown promise in various applications, explaining their predictions on continuous-time dynamic graphs (CTDGs) is difficult. This paper investigates a new research task: self-interpretable GNNs for CTDGs. We aim to predict future links within the dynamic graph while simultaneously providing causal explanations for these predictions. There are two key challenges: (1) capturing the underlying structural and temporal information that remains consistent across both independent and identically distributed (IID) and out-of-distribution (OOD) data, and (2) efficiently generating high-quality link prediction results and explanations. To tackle these challenges, we propose a novel causal inference model, namely the Independent and Confounded Causal Model (ICCM). ICCM is then integrated into a deep learning architecture that considers both effectiveness and efficiency. Extensive experiments demonstrate that our proposed model significantly outperforms existing methods across link prediction accuracy, explanation quality, and robustness to shortcut features. Our code and datasets are anonymously released at https://github.com/2024SIG/SIG.
Authors: Dongwhi Kim, Nuno Moniz
Abstract: As machine learning continues to gain prominence, transparency and explainability are increasingly critical. Without an understanding of these models, they can replicate and worsen human bias, adversely affecting marginalized communities. Algorithmic recourse emerges as a tool for clarifying decisions made by predictive models, providing actionable insights to alter outcomes. They answer, 'What do I have to change?' to achieve the desired result. Despite their importance, current algorithmic recourse methods treat all domain values equally, which is unrealistic in real-world settings. In this paper, we propose a novel framework, Relevance-Aware Algorithmic Recourse (RAAR), that leverages the concept of relevance in applying algorithmic recourse to regression tasks. We conducted multiple experiments on 15 datasets to outline how relevance influences recourses. Results show that relevance contributes algorithmic recourses comparable to well-known baselines, with greater efficiency and lower relative costs.
Authors: Yu Luo, Tianying Ji, Fuchun Sun, Jianwei Zhang, Huazhe Xu, Xianyuan Zhan
Abstract: Training reinforcement learning policies using environment interaction data collected from varying policies or dynamics presents a fundamental challenge. Existing works often overlook the distribution discrepancies induced by policy or dynamics shifts, or rely on specialized algorithms with task priors, thus often resulting in suboptimal policy performances and high learning variances. In this paper, we identify a unified strategy for online RL policy learning under diverse settings of policy and dynamics shifts: transition occupancy matching. In light of this, we introduce a surrogate policy learning objective by considering the transition occupancy discrepancies and then cast it into a tractable min-max optimization problem through dual reformulation. Our method, dubbed Occupancy-Matching Policy Optimization (OMPO), features a specialized actor-critic structure equipped with a distribution discriminator and a small-size local buffer. We conduct extensive experiments based on the OpenAI Gym, Meta-World, and Panda Robots environments, encompassing policy shifts under stationary and nonstationary dynamics, as well as domain adaption. The results demonstrate that OMPO outperforms the specialized baselines from different categories in all settings. We also find that OMPO exhibits particularly strong performance when combined with domain randomization, highlighting its potential in RL-based robotics applications
Authors: Shuyu Cheng, Yibo Miao, Yinpeng Dong, Xiao Yang, Xiao-Shan Gao, Jun Zhu
Abstract: This paper studies the challenging black-box adversarial attack that aims to generate adversarial examples against a black-box model by only using output feedback of the model to input queries. Some previous methods improve the query efficiency by incorporating the gradient of a surrogate white-box model into query-based attacks due to the adversarial transferability. However, the localized gradient is not informative enough, making these methods still query-intensive. In this paper, we propose a Prior-guided Bayesian Optimization (P-BO) algorithm that leverages the surrogate model as a global function prior in black-box adversarial attacks. As the surrogate model contains rich prior information of the black-box one, P-BO models the attack objective with a Gaussian process whose mean function is initialized as the surrogate model's loss. Our theoretical analysis on the regret bound indicates that the performance of P-BO may be affected by a bad prior. Therefore, we further propose an adaptive integration strategy to automatically adjust a coefficient on the function prior by minimizing the regret bound. Extensive experiments on image classifiers and large vision-language models demonstrate the superiority of the proposed algorithm in reducing queries and improving attack success rates compared with the state-of-the-art black-box attacks. Code is available at https://github.com/yibo-miao/PBO-Attack.
Authors: Maximilian Herde, Bogdan Raoni\'c, Tobias Rohner, Roger K\"appeli, Roberto Molinaro, Emmanuel de B\'ezenac, Siddhartha Mishra
Abstract: We introduce Poseidon, a foundation model for learning the solution operators of PDEs. It is based on a multiscale operator transformer, with time-conditioned layer norms that enable continuous-in-time evaluations. A novel training strategy leveraging the semi-group property of time-dependent PDEs to allow for significant scaling-up of the training data is also proposed. Poseidon is pretrained on a diverse, large scale dataset for the governing equations of fluid dynamics. It is then evaluated on a suite of 15 challenging downstream tasks that include a wide variety of PDE types and operators. We show that Poseidon exhibits excellent performance across the board by outperforming baselines significantly, both in terms of sample efficiency and accuracy. Poseidon also generalizes very well to new physics that is not seen during pretraining. Moreover, Poseidon scales with respect to model and data size, both for pretraining and for downstream tasks. Taken together, our results showcase the surprising ability of Poseidon to learn effective representations from a very small set of PDEs during pretraining in order to generalize well to unseen and unrelated PDEs downstream, demonstrating its potential as an effective, general purpose PDE foundation model. Finally, the Poseidon model as well as underlying pretraining and downstream datasets are open sourced, with code being available at https://github.com/camlab-ethz/poseidon and pretrained models and datasets at https://huggingface.co/camlab-ethz.
URLs: https://github.com/camlab-ethz/poseidon, https://huggingface.co/camlab-ethz.
Authors: Pierre Harvey Richemond, Yunhao Tang, Daniel Guo, Daniele Calandriello, Mohammad Gheshlaghi Azar, Rafael Rafailov, Bernardo Avila Pires, Eugene Tarassov, Lucas Spangher, Will Ellsworth, Aliaksei Severyn, Jonathan Mallinson, Lior Shani, Gil Shamir, Rishabh Joshi, Tianqi Liu, Remi Munos, Bilal Piot
Abstract: The dominant framework for alignment of large language models (LLM), whether through reinforcement learning from human feedback or direct preference optimisation, is to learn from preference data. This involves building datasets where each element is a quadruplet composed of a prompt, two independent responses (completions of the prompt) and a human preference between the two independent responses, yielding a preferred and a dis-preferred response. Such data is typically scarce and expensive to collect. On the other hand, \emph{single-trajectory} datasets where each element is a triplet composed of a prompt, a response and a human feedback is naturally more abundant. The canonical element of such datasets is for instance an LLM's response to a user's prompt followed by a user's feedback such as a thumbs-up/down. Consequently, in this work, we propose DRO, or \emph{Direct Reward Optimisation}, as a framework and associated algorithms that do not require pairwise preferences. DRO uses a simple mean-squared objective that can be implemented in various ways. We validate our findings empirically, using T5 encoder-decoder language models, and show DRO's performance over selected baselines such as Kahneman-Tversky Optimization (KTO). Thus, we confirm that DRO is a simple and empirically compelling method for single-trajectory policy optimisation.
Authors: Xixi Wu, Yifei Shen, Caihua Shan, Kaitao Song, Siwei Wang, Bohang Zhang, Jiarui Feng, Hong Cheng, Wei Chen, Yun Xiong, Dongsheng Li
Abstract: Task planning is emerging as an important research topic alongside the development of large language models (LLMs). It aims to break down complex user requests into solvable sub-tasks, thereby fulfilling the original requests. In this context, the sub-tasks can be naturally viewed as a graph, where the nodes represent the sub-tasks, and the edges denote the dependencies among them. Consequently, task planning is a decision-making problem that involves selecting a connected path or subgraph within the corresponding graph and invoking it. In this paper, we explore graph learning-based methods for task planning, a direction that is orthogonal to the prevalent focus on prompt design. Our interest in graph learning stems from a theoretical discovery: the biases of attention and auto-regressive loss impede LLMs' ability to effectively navigate decision-making on graphs, which is adeptly addressed by graph neural networks (GNNs). This theoretical insight led us to integrate GNNs with LLMs to enhance overall performance. Extensive experiments demonstrate that GNN-based methods surpass existing solutions even without training, and minimal training can further enhance their performance. Additionally, our approach complements prompt engineering and fine-tuning techniques, with performance further enhanced by improved prompts or a fine-tuned model.
Authors: Simon Geisler, Arthur Kosmala, Daniel Herbst, Stephan G\"unnemann
Abstract: Spatial Message Passing Graph Neural Networks (MPGNNs) are widely used for learning on graph-structured data. However, key limitations of l-step MPGNNs are that their "receptive field" is typically limited to the l-hop neighborhood of a node and that information exchange between distant nodes is limited by over-squashing. Motivated by these limitations, we propose Spatio-Spectral Graph Neural Networks (S$^2$GNNs) -- a new modeling paradigm for Graph Neural Networks (GNNs) that synergistically combines spatially and spectrally parametrized graph filters. Parameterizing filters partially in the frequency domain enables global yet efficient information propagation. We show that S$^2$GNNs vanquish over-squashing and yield strictly tighter approximation-theoretic error bounds than MPGNNs. Further, rethinking graph convolutions at a fundamental level unlocks new design spaces. For example, S$^2$GNNs allow for free positional encodings that make them strictly more expressive than the 1-Weisfeiler-Lehman (WL) test. Moreover, to obtain general-purpose S$^2$GNNs, we propose spectrally parametrized filters for directed graphs. S$^2$GNNs outperform spatial MPGNNs, graph transformers, and graph rewirings, e.g., on the peptide long-range benchmark tasks, and are competitive with state-of-the-art sequence modeling. On a 40 GB GPU, S$^2$GNNs scale to millions of nodes.
Authors: Arthur Juliani, Jordan T. Ash
Abstract: Continual learning with deep neural networks presents challenges distinct from both the fixed-dataset and convex continual learning regimes. One such challenge is plasticity loss, wherein a neural network trained in an online fashion displays a degraded ability to fit new tasks. This problem has been extensively studied in both supervised learning and off-policy reinforcement learning (RL), where a number of remedies have been proposed. Still, plasticity loss has received less attention in the on-policy deep RL setting. Here we perform an extensive set of experiments examining plasticity loss and a variety of mitigation methods in on-policy deep RL. We demonstrate that plasticity loss is pervasive under domain shift in this regime, and that a number of methods developed to resolve it in other settings fail, sometimes even resulting in performance that is worse than performing no intervention at all. In contrast, we find that a class of ``regenerative'' methods are able to consistently mitigate plasticity loss in a variety of contexts, including in gridworld tasks and more challenging environments like Montezuma's Revenge and ProcGen.
Authors: Robi Bhattacharjee, Nick Rittler, Kamalika Chaudhuri
Abstract: Many machine learning models appear to deploy effortlessly under distribution shift, and perform well on a target distribution that is considerably different from the training distribution. Yet, learning theory of distribution shift bounds performance on the target distribution as a function of the discrepancy between the source and target, rarely guaranteeing high target accuracy. Motivated by this gap, this work takes a closer look at the theory of distribution shift for a classifier from a source to a target distribution. Instead of relying on the discrepancy, we adopt an Invariant-Risk-Minimization (IRM)-like assumption connecting the distributions, and characterize conditions under which data from a source distribution is sufficient for accurate classification of the target. When these conditions are not met, we show when only unlabeled data from the target is sufficient, and when labeled target data is needed. In all cases, we provide rigorous theoretical guarantees in the large sample regime.
Authors: Sarthak Mittal, Eric Elmoznino, Leo Gagnon, Sangnie Bhardwaj, Dhanya Sridhar, Guillaume Lajoie
Abstract: Large autoregressive models like Transformers can solve tasks through in-context learning (ICL) without learning new weights, suggesting avenues for efficiently solving new tasks. For many tasks, e.g., linear regression, the data factorizes: examples are independent given a task latent that generates the data, e.g., linear coefficients. While an optimal predictor leverages this factorization by inferring task latents, it is unclear if Transformers implicitly do so or if they instead exploit heuristics and statistical shortcuts enabled by attention layers. Both scenarios have inspired active ongoing work. In this paper, we systematically investigate the effect of explicitly inferring task latents. We minimally modify the Transformer architecture with a bottleneck designed to prevent shortcuts in favor of more structured solutions, and then compare performance against standard Transformers across various ICL tasks. Contrary to intuition and some recent works, we find little discernible difference between the two; biasing towards task-relevant latent variables does not lead to better out-of-distribution performance, in general. Curiously, we find that while the bottleneck effectively learns to extract latent task variables from context, downstream processing struggles to utilize them for robust prediction. Our study highlights the intrinsic limitations of Transformers in achieving structured ICL solutions that generalize, and shows that while inferring the right latents aids interpretability, it is not sufficient to alleviate this problem.
Authors: Benjamin Shih, Ahmad Peyvan, Zhongqiang Zhang, George Em Karniadakis
Abstract: Neural operator learning models have emerged as very effective surrogates in data-driven methods for partial differential equations (PDEs) across different applications from computational science and engineering. Such operator learning models not only predict particular instances of a physical or biological system in real-time but also forecast classes of solutions corresponding to a distribution of initial and boundary conditions or forcing terms. % DeepONet is the first neural operator model and has been tested extensively for a broad class of solutions, including Riemann problems. Transformers have not been used in that capacity, and specifically, they have not been tested for solutions of PDEs with low regularity. % In this work, we first establish the theoretical groundwork that transformers possess the universal approximation property as operator learning models. We then apply transformers to forecast solutions of diverse dynamical systems with solutions of finite regularity for a plurality of initial conditions and forcing terms. In particular, we consider three examples: the Izhikevich neuron model, the tempered fractional-order Leaky Integrate-and-Fire (LIF) model, and the one-dimensional Euler equation Riemann problem. For the latter problem, we also compare with variants of DeepONet, and we find that transformers outperform DeepONet in accuracy but they are computationally more expensive.
Authors: Andrew Jacobsen, Ashok Cutkosky
Abstract: We develop algorithms for online linear regression which achieve optimal static and dynamic regret guarantees \emph{even in the complete absence of prior knowledge}. We present a novel analysis showing that a discounted variant of the Vovk-Azoury-Warmuth forecaster achieves dynamic regret of the form $R_{T}(\vec{u})\le O\left(d\log(T)\vee \sqrt{dP_{T}^{\gamma}(\vec{u})T}\right)$, where $P_{T}^{\gamma}(\vec{u})$ is a measure of variability of the comparator sequence, and show that the discount factor achieving this result can be learned on-the-fly. We show that this result is optimal by providing a matching lower bound. We also extend our results to \emph{strongly-adaptive} guarantees which hold over every sub-interval $[a,b]\subseteq[1,T]$ simultaneously.
Authors: Hanye Zhao, Xiaoshen Han, Zhengbang Zhu, Minghuan Liu, Yong Yu, Weinan Zhang
Abstract: With the great success of diffusion models (DMs) in generating realistic synthetic vision data, many researchers have investigated their potential in decision-making and control. Most of these works utilized DMs to sample directly from the trajectory space, where DMs can be viewed as a combination of dynamics models and policies. In this work, we explore how to decouple DMs' ability as dynamics models in fully offline settings, allowing the learning policy to roll out trajectories. As DMs learn the data distribution from the dataset, their intrinsic policy is actually the behavior policy induced from the dataset, which results in a mismatch between the behavior policy and the learning policy. We propose Dynamics Diffusion, short as DyDiff, which can inject information from the learning policy to DMs iteratively. DyDiff ensures long-horizon rollout accuracy while maintaining policy consistency and can be easily deployed on model-free algorithms. We provide theoretical analysis to show the advantage of DMs on long-horizon rollout over models and demonstrate the effectiveness of DyDiff in the context of offline reinforcement learning, where the rollout dataset is provided but no online environment for interaction. Our code is at https://github.com/FineArtz/DyDiff.
Authors: Renato M. Silva, Greg\'orio F. Azevedo, Matheus V. V. Berto, Jean R. Rocha, Eduardo C. Fidelis, Matheus V. Nogueira, Pedro H. Lisboa, Tiago A. Almeida
Abstract: Traffic incidents involving vulnerable road users (VRUs) constitute a significant proportion of global road accidents. Advances in traffic communication ecosystems, coupled with sophisticated signal processing and machine learning techniques, have facilitated the utilization of data from diverse sensors. Despite these advancements and the availability of extensive datasets, substantial progress is required to mitigate traffic casualties. This paper provides a comprehensive survey of state-of-the-art technologies and methodologies to enhance the safety of VRUs. The study delves into the communication networks between vehicles and VRUs, emphasizing the integration of advanced sensors and the availability of relevant datasets. It explores preprocessing techniques and data fusion methods to enhance sensor data quality. Furthermore, our study assesses critical simulation environments essential for developing and testing VRU safety systems. Our research also highlights recent advances in VRU detection and classification algorithms, addressing challenges such as variable environmental conditions. Additionally, we cover cutting-edge research in predicting VRU intentions and behaviors, which is crucial for proactive collision avoidance strategies. Through this survey, we aim to provide a comprehensive understanding of the current landscape of VRU safety technologies, identifying areas of progress and areas needing further research and development.
Authors: Paulo Neves, Joerg K. Wegner, Philippe Schwaller
Abstract: Ensuring high-quality data is paramount for maximizing the performance of machine learning models and business intelligence systems. However, challenges in data quality, including noise in data capture, missing records, limited data production, and confounding variables, significantly constrain the potential performance of these systems. In this study, we propose an architecture-agnostic algorithm, Gradient Guided Hypotheses (GGH), designed to address these challenges. GGH analyses gradients from hypotheses as a proxy of distinct and possibly contradictory patterns in the data. This framework entails an additional step in machine learning training, where gradients can be included or excluded from backpropagation. In this manner, missing and noisy data are addressed through a unified solution that perceives both challenges as facets of the same overarching issue: the propagation of erroneous information. Experimental validation of GGH is conducted using real-world open-source datasets, where records with missing rates of up to 98.5% are simulated. Comparative analysis with state-of-the-art imputation methods demonstrates a substantial improvement in model performance achieved by GGH. Specifically in very high scarcity regimes, GGH was found to be the only viable solution. Additionally, GGH's noise detection capabilities are showcased by introducing simulated noise into the datasets and observing enhanced model performance after filtering out the noisy data. This study presents GGH as a promising solution for improving data quality and model performance in various applications.
Authors: Keltin Grimes, Collin Abidi, Cole Frank, Shannon Gallagher
Abstract: Machine learning models are vulnerable to adversarial attacks, including attacks that leak information about the model's training data. There has recently been an increase in interest about how to best address privacy concerns, especially in the presence of data-removal requests. Machine unlearning algorithms aim to efficiently update trained models to comply with data deletion requests while maintaining performance and without having to resort to retraining the model from scratch, a costly endeavor. Several algorithms in the machine unlearning literature demonstrate some level of privacy gains, but they are often evaluated only on rudimentary membership inference attacks, which do not represent realistic threats. In this paper we describe and propose alternative evaluation methods for three key shortcomings in the current evaluation of unlearning algorithms. We show the utility of our alternative evaluations via a series of experiments of state-of-the-art unlearning algorithms on different computer vision datasets, presenting a more detailed picture of the state of the field.
Authors: Charles Westphal, Stephen Hailes, Mirco Musolesi
Abstract: In this paper, we introduce Partial Information Decomposition of Features (PIDF), a new paradigm for simultaneous data interpretability and feature selection. Contrary to traditional methods that assign a single importance value, our approach is based on three metrics per feature: the mutual information shared with the target variable, the feature's contribution to synergistic information, and the amount of this information that is redundant. In particular, we develop a novel procedure based on these three metrics, which reveals not only how features are correlated with the target but also the additional and overlapping information provided by considering them in combination with other features. We extensively evaluate PIDF using both synthetic and real-world data, demonstrating its potential applications and effectiveness, by considering case studies from genetics and neuroscience.
Authors: Bijan Mazaheri, Chandler Squires, Caroline Uhler
Abstract: Modern data analysis frequently relies on the use of large datasets, often constructed as amalgamations of diverse populations or data-sources. Heterogeneity across these smaller datasets constitutes two major challenges for causal inference: (1) the source of each sample can introduce latent confounding between treatment and effect, and (2) diverse populations may respond differently to the same treatment, giving rise to heterogeneous treatment effects (HTEs). The issues of latent confounding and HTEs have been studied separately but not in conjunction. In particular, previous works only report the conditional average treatment effect (CATE) among similar individuals (with respect to the measured covariates). CATEs cannot resolve mixtures of potential treatment effects driven by latent heterogeneity, which we call mixtures of treatment effects (MTEs). Inspired by method of moment approaches to mixture models, we propose "synthetic potential outcomes" (SPOs). Our new approach deconfounds heterogeneity while also guaranteeing the identifiability of MTEs. This technique bypasses full recovery of a mixture, which significantly simplifies its requirements for identifiability. We demonstrate the efficacy of SPOs on synthetic data.
Authors: Mohammadreza Sadeghi, Zihan Wang, Narges Armanfard
Abstract: Unsupervised Continual Learning (UCL) is a burgeoning field in machine learning, focusing on enabling neural networks to sequentially learn tasks without explicit label information. Catastrophic Forgetting (CF), where models forget previously learned tasks upon learning new ones, poses a significant challenge in continual learning, especially in UCL, where labeled information of data is not accessible. CF mitigation strategies, such as knowledge distillation and replay buffers, often face memory inefficiency and privacy issues. Although current research in UCL has endeavored to refine data representations and address CF in streaming data contexts, there is a noticeable lack of algorithms specifically designed for unsupervised clustering. To fill this gap, in this paper, we introduce the concept of Unsupervised Continual Clustering (UCC). We propose Forward-Backward Knowledge Distillation for unsupervised Continual Clustering (FBCC) to counteract CF within the context of UCC. FBCC employs a single continual learner (the ``teacher'') with a cluster projector, along with multiple student models, to address the CF issue. The proposed method consists of two phases: Forward Knowledge Distillation, where the teacher learns new clusters while retaining knowledge from previous tasks with guidance from specialized student models, and Backward Knowledge Distillation, where a student model mimics the teacher's behavior to retain task-specific knowledge, aiding the teacher in subsequent tasks. FBCC marks a pioneering approach to UCC, demonstrating enhanced performance and memory efficiency in clustering across various tasks, outperforming the application of clustering algorithms to the latent space of state-of-the-art UCL algorithms.
Authors: Zhuang Qi, Junlin Zhang, Xiaming Chen, Xin Qi
Abstract: The neighbor-based method has become a powerful tool to handle the outlier detection problem, which aims to infer the abnormal degree of the sample based on the compactness of the sample and its neighbors. However, the existing methods commonly focus on designing different processes to locate outliers in the dataset, while the contributions of different types neighbors to outlier detection has not been well discussed. To this end, this paper studies the neighbor in the existing outlier detection algorithms and a taxonomy is introduced, which uses the three-level components of information, neighbor and methodology to define hybrid methods. This taxonomy can serve as a paradigm where a novel neighbor-based outlier detection method can be proposed by combining different components in this taxonomy. A large number of comparative experiments were conducted on synthetic and real-world datasets in terms of performance comparison and case study, and the results show that reverse K-nearest neighbor based methods achieve promising performance and dynamic selection method is suitable for working in high-dimensional space. Notably, it is verified that rationally selecting components from this taxonomy may create an algorithms superior to existing methods.
Authors: Zhiqiang Cai, Yu Cao, Yuanfei Huang, Xiang Zhou
Abstract: Sampling invariant distributions from an Ito diffusion process presents a significant challenge in stochastic simulation. Traditional numerical solvers for stochastic differential equations require both a fine step size and a lengthy simulation period, resulting in both biased and correlated samples. Current deep learning-based method solves the stationary Fokker--Planck equation to determine the invariant probability density function in form of deep neural networks, but they generally do not directly address the problem of sampling from the computed density function. In this work, we introduce a framework that employs a weak generative sampler (WGS) to directly generate independent and identically distributed (iid) samples induced by a transformation map derived from the stationary Fokker--Planck equation. Our proposed loss function is based on the weak form of the Fokker--Planck equation, integrating normalizing flows to characterize the invariant distribution and facilitate sample generation from the base distribution. Our randomized test function circumvents the need for mini-max optimization in the traditional weak formulation. Distinct from conventional generative models, our method neither necessitates the computationally intensive calculation of the Jacobian determinant nor the invertibility of the transformation map. A crucial component of our framework is the adaptively chosen family of test functions in the form of Gaussian kernel functions with centres selected from the generated data samples. Experimental results on several benchmark examples demonstrate the effectiveness of our method, which offers both low computational costs and excellent capability in exploring multiple metastable states.
Authors: Yuda Song, Lili Wu, Dylan J. Foster, Akshay Krishnamurthy
Abstract: Sample-efficiency and reliability remain major bottlenecks toward wide adoption of reinforcement learning algorithms in continuous settings with high-dimensional perceptual inputs. Toward addressing these challenges, we introduce a new theoretical framework, RichCLD (Rich-Observation RL with Continuous Latent Dynamics), in which the agent performs control based on high-dimensional observations, but the environment is governed by low-dimensional latent states and Lipschitz continuous dynamics. Our main contribution is a new algorithm for this setting that is provably statistically and computationally efficient. The core of our algorithm is a new representation learning objective; we show that prior representation learning schemes tailored to discrete dynamics do not naturally extend to the continuous setting. Our new objective is amenable to practical implementation, and empirically, we find that it compares favorably to prior schemes in a standard evaluation protocol. We further provide several insights into the statistical complexity of the RichCLD framework, in particular proving that certain notions of Lipschitzness that admit sample-efficient learning in the absence of rich observations are insufficient in the rich-observation setting.
Authors: Saber Malekmohammadi, Afaf Taik, Golnoosh Farnadi
Abstract: Federated Learning (FL) is a decentralized machine learning (ML) approach that keeps data localized and often incorporates Differential Privacy (DP) to enhance privacy guarantees. Similar to previous work on DP in ML, we observed that differentially private federated learning (DPFL) introduces performance disparities, particularly affecting minority groups. Recent work has attempted to address performance fairness in vanilla FL through clustering, but this method remains sensitive and prone to errors, which are further exacerbated by the DP noise in DPFL. To fill this gap, in this paper, we propose a novel clustered DPFL algorithm designed to effectively identify clients' clusters in highly heterogeneous settings while maintaining high accuracy with DP guarantees. To this end, we propose to cluster clients based on both their model updates and training loss values. Our proposed approach also addresses the server's uncertainties in clustering clients' model updates by employing larger batch sizes along with Gaussian Mixture Model (GMM) to alleviate the impact of noise and potential clustering errors, especially in privacy-sensitive scenarios. We provide theoretical analysis of the effectiveness of our proposed approach. We also extensively evaluate our approach across diverse data distributions and privacy budgets and show its effectiveness in mitigating the disparate impact of DP in FL settings with a small computational cost.
Authors: Khuong Vo
Abstract: A deep latent variable model is a powerful method for capturing complex distributions. These models assume that underlying structures, but unobserved, are present within the data. In this dissertation, we explore high-dimensional problems related to physiological monitoring using latent variable models. First, we present a novel deep state-space model to generate electrical waveforms of the heart using optically obtained signals as inputs. This can bring about clinical diagnoses of heart disease via simple assessment through wearable devices. Second, we present a brain signal modeling scheme that combines the strengths of probabilistic graphical models and deep adversarial learning. The structured representations can provide interpretability and encode inductive biases to reduce the data complexity of neural oscillations. The efficacy of the learned representations is further studied in epilepsy seizure detection formulated as an unsupervised learning problem. Third, we propose a framework for the joint modeling of physiological measures and behavior. Existing methods to combine multiple sources of brain data provided are limited. Direct analysis of the relationship between different types of physiological measures usually does not involve behavioral data. Our method can identify the unique and shared contributions of brain regions to behavior and can be used to discover new functions of brain regions. The success of these innovative computational methods would allow the translation of biomarker findings across species and provide insight into neurocognitive analysis in numerous biological studies and clinical diagnoses, as well as emerging consumer applications.
Authors: Bobby He, Lorenzo Noci, Daniele Paliotta, Imanol Schlag, Thomas Hofmann
Abstract: Outlier Features (OF) are neurons whose activation magnitudes significantly exceed the average over a neural network's (NN) width. They are well known to emerge during standard transformer training and have the undesirable effect of hindering quantisation in afflicted models. Despite their practical importance, little is known behind why OFs emerge during training, nor how one can minimise them. Our work focuses on the above questions, first identifying several quantitative metrics, such as the kurtosis over neuron activation norms, to measure OFs. With these metrics, we study how architectural and optimisation choices influence OFs, and provide practical insights to minimise OFs during training. As highlights, we emphasise the importance of controlling signal propagation throughout training, and propose the Outlier Protected transformer block, which removes standard Pre-Norm layers to mitigate OFs, without loss of convergence speed or training stability. Overall, our findings shed new light on our understanding of, our ability to prevent, and the complexity of this important facet in NN training dynamics.
Authors: Manh Khoi Duong, Stefan Conrad
Abstract: Motivated by the recital (67) of the current corrigendum of the AI Act in the European Union, we propose and present measures and mitigation strategies for discrimination in tabular datasets. We specifically focus on datasets that contain multiple protected attributes, such as nationality, age, and sex. This makes measuring and mitigating bias more challenging, as many existing methods are designed for a single protected attribute. This paper comes with a twofold contribution: Firstly, new discrimination measures are introduced. These measures are categorized in our framework along with existing ones, guiding researchers and practitioners in choosing the right measure to assess the fairness of the underlying dataset. Secondly, a novel application of an existing bias mitigation method, FairDo, is presented. We show that this strategy can mitigate any type of discrimination, including intersectional discrimination, by transforming the dataset. By conducting experiments on real-world datasets (Adult, Bank, Compas), we demonstrate that de-biasing datasets with multiple protected attributes is achievable. Further, the transformed fair datasets do not compromise any of the tested machine learning models' performances significantly when trained on these datasets compared to the original datasets. Discrimination was reduced by up to 83% in our experimentation. For most experiments, the disparity between protected groups was reduced by at least 7% and 27% on average. Generally, the findings show that the mitigation strategy used is effective, and this study contributes to the ongoing discussion on the implementation of the European Union's AI Act.
Authors: Adam Fisch, Jacob Eisenstein, Vicky Zayats, Alekh Agarwal, Ahmad Beirami, Chirag Nagpal, Pete Shaw, Jonathan Berant
Abstract: Language model (LM) post-training (or alignment) involves maximizing a reward function that is derived from preference annotations. Direct Preference Optimization (DPO) is a popular offline alignment method that trains a policy directly on preference data without the need to train a reward model or apply reinforcement learning. However, typical preference datasets have only a single, or at most a few, annotation per preference pair, which causes DPO to overconfidently assign rewards that trend towards infinite magnitude. This frequently leads to degenerate policies, sometimes causing even the probabilities of the preferred generations to go to zero. In this work, we analyze this phenomenon and propose distillation to get a better proxy for the true preference distribution over generation pairs: we train the LM to produce probabilities that match the distribution induced by a reward model trained on the preference data. Moreover, to account for uncertainty in the reward model we are distilling from, we optimize against a family of reward models that, as a whole, is likely to include at least one reasonable proxy for the preference distribution. Our results show that distilling from such a family of reward models leads to improved robustness to distribution shift in preference annotations, while preserving the simple supervised nature of DPO.
Authors: Masahiro Kato
Abstract: This study investigates a local asymptotic minimax optimal strategy for fixed-budget best arm identification (BAI). We propose the Adaptive Generalized Neyman Allocation (AGNA) strategy and show that its worst-case upper bound of the probability of misidentifying the best arm aligns with the worst-case lower bound under the small-gap regime, where the gap between the expected outcomes of the best and suboptimal arms is small. Our strategy corresponds to a generalization of the Neyman allocation for two-armed bandits (Neyman, 1934; Kaufmann et al., 2016) and a refinement of existing strategies such as the ones proposed by Glynn & Juneja (2004) and Shin et al. (2018). Compared to Komiyama et al. (2022), which proposes a minimax rate-optimal strategy, our proposed strategy has a tighter upper bound that exactly matches the lower bound, including the constant terms, by restricting the class of distributions to the class of small-gap distributions. Our result contributes to the longstanding open issue about the existence of asymptotically optimal strategies in fixed-budget BAI, by presenting the local asymptotic minimax optimal strategy.
Authors: Shicong Cen, Jincheng Mei, Katayoon Goshvadi, Hanjun Dai, Tong Yang, Sherry Yang, Dale Schuurmans, Yuejie Chi, Bo Dai
Abstract: Reinforcement learning from human feedback (RLHF) has demonstrated great promise in aligning large language models (LLMs) with human preference. Depending on the availability of preference data, both online and offline RLHF are active areas of investigation. A key bottleneck is understanding how to incorporate uncertainty estimation in the reward function learned from the preference data for RLHF, regardless of how the preference data is collected. While the principles of optimism or pessimism under uncertainty are well-established in standard reinforcement learning (RL), a practically-implementable and theoretically-grounded form amenable to large language models is not yet available, as standard techniques for constructing confidence intervals become intractable under arbitrary policy parameterizations. In this paper, we introduce a unified approach to online and offline RLHF -- value-incentivized preference optimization (VPO) -- which regularizes the maximum-likelihood estimate of the reward function with the corresponding value function, modulated by a $\textit{sign}$ to indicate whether the optimism or pessimism is chosen. VPO also directly optimizes the policy with implicit reward modeling, and therefore shares a simpler RLHF pipeline similar to direct preference optimization. Theoretical guarantees of VPO are provided for both online and offline settings, matching the rates of their standard RL counterparts. Moreover, experiments on text summarization and dialog verify the practicality and effectiveness of VPO.
Authors: Shenao Zhang, Donghan Yu, Hiteshi Sharma, Ziyi Yang, Shuohang Wang, Hany Hassan, Zhaoran Wang
Abstract: Preference optimization, particularly through Reinforcement Learning from Human Feedback (RLHF), has achieved significant success in aligning Large Language Models (LLMs) to adhere to human intentions. Unlike offline alignment with a fixed dataset, online feedback collection from humans or AI on model generations typically leads to more capable reward models and better-aligned LLMs through an iterative process. However, achieving a globally accurate reward model requires systematic exploration to generate diverse responses that span the vast space of natural language. Random sampling from standard reward-maximizing LLMs alone is insufficient to fulfill this requirement. To address this issue, we propose a bilevel objective optimistically biased towards potentially high-reward responses to actively explore out-of-distribution regions. By solving the inner-level problem with the reparameterized reward function, the resulting algorithm, named Self-Exploring Language Models (SELM), eliminates the need for a separate RM and iteratively updates the LLM with a straightforward objective. Compared to Direct Preference Optimization (DPO), the SELM objective reduces indiscriminate favor of unseen extrapolations and enhances exploration efficiency. Our experimental results demonstrate that when finetuned on Zephyr-7B-SFT and Llama-3-8B-Instruct models, SELM significantly boosts the performance on instruction-following benchmarks such as MT-Bench and AlpacaEval 2.0, as well as various standard academic benchmarks in different settings. Our code and models are available at https://github.com/shenao-zhang/SELM.
Authors: Geremy Loacham\'in Suntaxi, Paris Papavasileiou, Eleni D. Koronaki, Dimitrios G. Giovanis, Georgios Gakis, Ioannis G. Aviziotis, Martin Kathrein, Gabriele Pozzetti, Christoph Czettl, St\'ephane P. A. Bordas, Andreas G. Boudouvis
Abstract: This work introduces a comprehensive approach utilizing data-driven methods to elucidate the deposition process regimes in Chemical Vapor Deposition (CVD) reactors and the interplay of physical mechanism that dominate in each one of them. Through this work, we address three key objectives. Firstly, our methodology relies on process outcomes, derived by a detailed CFD model, to identify clusters of "outcomes" corresponding to distinct process regimes, wherein the relative influence of input variables undergoes notable shifts. This phenomenon is experimentally validated through Arrhenius plot analysis, affirming the efficacy of our approach. Secondly, we demonstrate the development of an efficient surrogate model, based on Polynomial Chaos Expansion (PCE), that maintains accuracy, facilitating streamlined computational analyses. Finally, as a result of PCE, sensitivity analysis is made possible by means of Sobol' indices, that quantify the impact of process inputs across identified regimes. The insights gained from our analysis contribute to the formulation of hypotheses regarding phenomena occurring beyond the transition regime. Notably, the significance of temperature even in the diffusion-limited regime, as evidenced by the Arrhenius plot, suggests activation of gas phase reactions at elevated temperatures. Importantly, our proposed methods yield insights that align with experimental observations and theoretical principles, aiding decision-making in process design and optimization. By circumventing the need for costly and time-consuming experiments, our approach offers a pragmatic pathway towards enhanced process efficiency. Moreover, this study underscores the potential of data-driven computational methods for innovating reactor design paradigms.
Authors: Yavuz Selim Inan
Abstract: WHO has declared that more than 2.2 billion people worldwide are suffering from visual disorders, such as media haze, glaucoma, and drusen. At least 1 billion of these cases could have been either prevented or successfully treated, yet they remain unaddressed due to poverty, a lack of specialists, inaccurate ocular fundus diagnoses by ophthalmologists, or the presence of a rare disease. To address this, the research has developed the Hybrid Trio-Network Model Algorithm for accurately diagnosing 12 distinct common and rare eye diseases. This algorithm utilized the RFMiD dataset of 3,200 fundus images and the Binary Relevance Method to detect diseases separately, ensuring expandability and avoiding incorrect correlations. Each detector, incorporating finely tuned hyperparameters to optimize performance, consisted of three feature components: A classical transfer learning CNN model, a two-stage CNN model, and a Siamese Network. The diagnosis was made using features extracted through this Trio-Model with Ensembled Machine Learning algorithms. The proposed model achieved an average accuracy of 97% and an AUC score of 0.96. Compared to past benchmark studies, an increase of over 10% in the F1-score was observed for most diseases. Furthermore, using the Siamese Network, the model successfully made predictions in diseases like optic disc pallor, which past studies failed to predict due to low confidence. This diagnostic tool presents a stable, adaptive, cost-effective, efficient, accessible, and fast solution for globalizing early detection of both common and rare diseases.
Authors: Zhangyu Wang, Krzysztof Janowicz, Gengchen Mai, Ivan Majic
Abstract: Intuitively, there is a relation between measures of spatial dependence and information theoretical measures of entropy. For instance, we can provide an intuition of why spatial data is special by stating that, on average, spatial data samples contain less than expected information. Similarly, spatial data, e.g., remotely sensed imagery, that is easy to compress is also likely to show significant spatial autocorrelation. Formulating our (highly specific) core concepts of spatial information theory in the widely used language of information theory opens new perspectives on their differences and similarities and also fosters cross-disciplinary collaboration, e.g., with the broader AI/ML communities. Interestingly, however, this intuitive relation is challenging to formalize and generalize, leading prior work to rely mostly on experimental results, e.g., for describing landscape patterns. In this work, we will explore the information theoretical roots of spatial autocorrelation, more specifically Moran's I, through the lens of self-information (also known as surprisal) and provide both formal proofs and experiments.
Authors: Andrew Balch
Abstract: Algorithms play an increasingly-significant role in our social lives. Unfortunately, they often perpetuate social injustices while doing so. The popular means of addressing these algorithmic injustices has been through algorithmic reformism: fine-tuning the algorithm itself to be more fair, accountable, and transparent. While commendable, the emerging discipline of critical algorithm studies shows that reformist approaches have failed to curtail algorithmic injustice because they ignore the power structure surrounding algorithms. Heeding calls from critical algorithm studies to analyze this power structure, I employ a framework developed by Erik Olin Wright to examine the configuration of power surrounding Algorithmic Activity: the ways in which algorithms are researched, developed, trained, and deployed within society. I argue that the reason Algorithmic Activity is unequal, undemocratic, and unsustainable is that the power structure shaping it is one of economic empowerment rather than social empowerment. For Algorithmic Activity to be socially just, we need to transform this power configuration to empower the people at the other end of an algorithm. To this end, I explore Wright's symbiotic, interstitial, and raptural transformations in the context of Algorithmic Activity, as well as how they may be applied in a hypothetical research project that uses algorithms to address a social issue. I conclude with my vision for socially just Algorithmic Activity, asking that future work strives to integrate the proposed transformations and develop new mechanisms for social empowerment.
Authors: Shehu AbdusSalam, Steve Abel, Miguel Crispim Romao
Abstract: We propose symbolic regression as a powerful tool for studying Beyond the Standard Model physics. As a benchmark model, we consider the so-called Constrained Minimal Supersymmetric Standard Model, which has a four-dimensional parameter space defined at the GUT scale. We provide a set of analytical expressions that reproduce three low-energy observables of interest in terms of the parameters of the theory: the Higgs mass, the contribution to the anomalous magnetic moment of the muon, and the cold dark matter relic density. To demonstrate the power of the approach, we employ the symbolic expressions in a global fits analysis to derive the posterior probability densities of the parameters, which are obtained extremely rapidly in comparison with conventional methods.
Authors: Marc Wanner, Laura Lewis, Chiranjib Bhattacharyya, Devdatt Dubhashi, Alexandru Gheorghiu
Abstract: A fundamental problem in quantum many-body physics is that of finding ground states of local Hamiltonians. A number of recent works gave provably efficient machine learning (ML) algorithms for learning ground states. Specifically, [Huang et al. Science 2022], introduced an approach for learning properties of the ground state of an $n$-qubit gapped local Hamiltonian $H$ from only $n^{\mathcal{O}(1)}$ data points sampled from Hamiltonians in the same phase of matter. This was subsequently improved by [Lewis et al. Nature Communications 2024], to $\mathcal{O}(\log n)$ samples when the geometry of the $n$-qubit system is known. In this work, we introduce two approaches that achieve a constant sample complexity, independent of system size $n$, for learning ground state properties. Our first algorithm consists of a simple modification of the ML model used by Lewis et al. and applies to a property of interest known beforehand. Our second algorithm, which applies even if a description of the property is not known, is a deep neural network model. While empirical results showing the performance of neural networks have been demonstrated, to our knowledge, this is the first rigorous sample complexity bound on a neural network model for predicting ground state properties. We also perform numerical experiments that confirm the improved scaling of our approach compared to earlier results.
Authors: Hai-Vy Nguyen, Fabrice Gamboa, Sixin Zhang, Reda Chhaibi, Serge Gratton, Thierry Giaccone
Abstract: In this paper, we introduce a novel discriminative loss function with large margin in the context of Deep Learning. This loss boosts the discriminative power of neural nets, represented by intra-class compactness and inter-class separability. On the one hand, the class compactness is ensured by close distance of samples of the same class to each other. On the other hand, the inter-class separability is boosted by a margin loss that ensures the minimum distance of each class to its closest boundary. All the terms in our loss have an explicit meaning, giving a direct view of the feature space obtained. We analyze mathematically the relation between compactness and margin term, giving a guideline about the impact of the hyper-parameters on the learned features. Moreover, we also analyze properties of the gradient of the loss with respect to the parameters of the neural net. Based on this, we design a strategy called partial momentum updating that enjoys simultaneously stability and consistency in training. Furthermore, we also investigate generalization errors to have better theoretical insights. Our loss function systematically boosts the test accuracy of models compared to the standard softmax loss in our experiments.
Authors: Koichi Saito, Dongjun Kim, Takashi Shibuya, Chieh-Hsin Lai, Zhi Zhong, Yuhta Takida, Yuki Mitsufuji
Abstract: Sound content is an indispensable element for multimedia works such as video games, music, and films. Recent high-quality diffusion-based sound generation models can serve as valuable tools for the creators. However, despite producing high-quality sounds, these models often suffer from slow inference speeds. This drawback burdens creators, who typically refine their sounds through trial and error to align them with their artistic intentions. To address this issue, we introduce Sound Consistency Trajectory Models (SoundCTM). Our model enables flexible transitioning between high-quality 1-step sound generation and superior sound quality through multi-step generation. This allows creators to initially control sounds with 1-step samples before refining them through multi-step generation. While CTM fundamentally achieves flexible 1-step and multi-step generation, its impressive performance heavily depends on an additional pretrained feature extractor and an adversarial loss, which are expensive to train and not always available in other domains. Thus, we reframe CTM's training framework and introduce a novel feature distance by utilizing the teacher's network for a distillation loss. Additionally, while distilling classifier-free guided trajectories, we train conditional and unconditional student models simultaneously and interpolate between these models during inference. We also propose training-free controllable frameworks for SoundCTM, leveraging its flexible sampling capability. SoundCTM achieves both promising 1-step and multi-step real-time sound generation without using any extra off-the-shelf networks. Furthermore, we demonstrate SoundCTM's capability of controllable sound generation in a training-free manner.
Authors: Shivesh Jadon, Mehrad Faridan, Edward Mah, Rajan Vaish, Wesley Willett, Ryo Suzuki
Abstract: This paper introduces the concept of augmented conversation, which aims to support co-located in-person conversations via embedded speech-driven on-the-fly referencing in augmented reality (AR). Today computing technologies like smartphones allow quick access to a variety of references during the conversation. However, these tools often create distractions, reducing eye contact and forcing users to focus their attention on phone screens and manually enter keywords to access relevant information. In contrast, AR-based on-the-fly referencing provides relevant visual references in real-time, based on keywords extracted automatically from the spoken conversation. By embedding these visual references in AR around the conversation partner, augmented conversation reduces distraction and friction, allowing users to maintain eye contact and supporting more natural social interactions. To demonstrate this concept, we developed \system, a Hololens-based interface that leverages real-time speech recognition, natural language processing and gaze-based interactions for on-the-fly embedded visual referencing. In this paper, we explore the design space of visual referencing for conversations, and describe our our implementation -- building on seven design guidelines identified through a user-centered design process. An initial user study confirms that our system decreases distraction and friction in conversations compared to smartphone searches, while providing highly useful and relevant information.
Authors: Seanie Lee, Minsu Kim, Lynn Cherif, David Dobre, Juho Lee, Sung Ju Hwang, Kenji Kawaguchi, Gauthier Gidel, Yoshua Bengio, Nikolay Malkin, Moksh Jain
Abstract: Red-teaming, or identifying prompts that elicit harmful responses, is a critical step in ensuring the safe and responsible deployment of large language models (LLMs). Developing effective protection against many modes of attack prompts requires discovering diverse attacks. Automated red-teaming typically uses reinforcement learning to fine-tune an attacker language model to generate prompts that elicit undesirable responses from a target LLM, as measured, for example, by an auxiliary toxicity classifier. We show that even with explicit regularization to favor novelty and diversity, existing approaches suffer from mode collapse or fail to generate effective attacks. As a flexible and probabilistically principled alternative, we propose to use GFlowNet fine-tuning, followed by a secondary smoothing phase, to train the attacker model to generate diverse and effective attack prompts. We find that the attacks generated by our method are effective against a wide range of target LLMs, both with and without safety tuning, and transfer well between target LLMs. Finally, we demonstrate that models safety-tuned using a dataset of red-teaming prompts generated by our method are robust to attacks from other RL-based red-teaming approaches.
Authors: Francisco de Arriba-P\'erez, Silvia Garc\'ia-M\'endez, Francisco J. Gonz\'alez-Casta\~no, Enrique Costa-Montenegro
Abstract: Previous researchers have proposed intelligent systems for therapeutic monitoring of cognitive impairments. However, most existing practical approaches for this purpose are based on manual tests. This raises issues such as excessive caretaking effort and the white-coat effect. To avoid these issues, we present an intelligent conversational system for entertaining elderly people with news of their interest that monitors cognitive impairment transparently. Automatic chatbot dialogue stages allow assessing content description skills and detecting cognitive impairment with Machine Learning algorithms. We create these dialogue flows automatically from updated news items using Natural Language Generation techniques. The system also infers the gold standard of the answers to the questions, so it can assess cognitive capabilities automatically by comparing these answers with the user responses. It employs a similarity metric with values in [0, 1], in increasing level of similarity. To evaluate the performance and usability of our approach, we have conducted field tests with a test group of 30 elderly people in the earliest stages of dementia, under the supervision of gerontologists. In the experiments, we have analysed the effect of stress and concentration in these users. Those without cognitive impairment performed up to five times better. In particular, the similarity metric varied between 0.03, for stressed and unfocused participants, and 0.36, for relaxed and focused users. Finally, we developed a Machine Learning algorithm based on textual analysis features for automatic cognitive impairment detection, which attained accuracy, F-measure and recall levels above 80%. We have thus validated the automatic approach to detect cognitive impairment in elderly people based on entertainment content.
Authors: Marco S\"alzer, Eric Alsmann, Martin Lange
Abstract: We investigate challenges and possibilities of formal reasoning for encoder-only transformers (EOT), meaning sound and complete methods for verifying or interpreting behaviour. In detail, we condense related formal reasoning tasks in the form of a naturally occurring satisfiability problem (SAT). We find that SAT is undecidable if we consider EOT, commonly considered in the expressiveness community. Furthermore, we identify practical scenarios where SAT is decidable and establish corresponding complexity bounds. Besides trivial cases, we find that quantized EOT, namely those restricted by some fixed-width arithmetic, lead to the decidability of SAT due to their limited attention capabilities. However, the problem remains difficult, as we establish those scenarios where SAT is NEXPTIME-hard and those where we can show that it is solvable in NEXPTIME for quantized EOT. To complement our theoretical results, we put our findings and their implications in the overall perspective of formal reasoning.
Authors: Shubhang Bhatnagar, Narendra Ahuja
Abstract: Deep metric learning (DML) involves training a network to learn a semantically meaningful representation space. Many current approaches mine n-tuples of examples and model interactions within each tuplets. We present a novel, compositional DML model, inspired by electrostatic fields in physics that, instead of in tuples, represents the influence of each example (embedding) by a continuous potential field, and superposes the fields to obtain their combined global potential field. We use attractive/repulsive potential fields to represent interactions among embeddings from images of the same/different classes. Contrary to typical learning methods, where mutual influence of samples is proportional to their distance, we enforce reduction in such influence with distance, leading to a decaying field. We show that such decay helps improve performance on real world datasets with large intra-class variations and label noise. Like other proxy-based methods, we also use proxies to succinctly represent sub-populations of examples. We evaluate our method on three standard DML benchmarks- Cars-196, CUB-200-2011, and SOP datasets where it outperforms state-of-the-art baselines.
Authors: Sami Davies, Sergei Vassilvitskii, Yuyan Wang
Abstract: Push-Relabel is one of the most celebrated network flow algorithms. Maintaining a pre-flow that saturates a cut, it enjoys better theoretical and empirical running time than other flow algorithms, such as Ford-Fulkerson. In practice, Push-Relabel is even faster than what theoretical guarantees can promise, in part because of the use of good heuristics for seeding and updating the iterative algorithm. However, it remains unclear how to run Push-Relabel on an arbitrary initialization that is not necessarily a pre-flow or cut-saturating. We provide the first theoretical guarantees for warm-starting Push-Relabel with a predicted flow, where our learning-augmented version benefits from fast running time when the predicted flow is close to an optimal flow, while maintaining robust worst-case guarantees. Interestingly, our algorithm uses the gap relabeling heuristic, which has long been employed in practice, even though prior to our work there was no rigorous theoretical justification for why it can lead to run-time improvements. We then provide experiments that show our warm-started Push-Relabel also works well in practice.
Authors: Abrar Fahim, Alex Murphy, Alona Fyshe
Abstract: Multi-modal contrastive models such as CLIP achieve state-of-the-art performance in zero-shot classification by embedding input images and texts on a joint representational space. Recently, a modality gap has been reported in two-encoder contrastive models like CLIP, meaning that the image and text embeddings reside in disjoint areas of the latent space. Previous studies suggest that this gap exists due to 1) the cone effect, 2) mismatched pairs in the dataset, and 3) insufficient training. We show that, even when accounting for all these factors, and even when using the same modality, the contrastive loss actually creates a gap during training. As a result, We propose that the modality gap is inherent to the two-encoder contrastive loss and rename it the contrastive gap. We present evidence that attributes this contrastive gap to low uniformity in CLIP space, resulting in embeddings that occupy only a small portion of the latent space. To close the gap, we adapt the uniformity and alignment properties of unimodal contrastive loss to the multi-modal setting and show that simply adding these terms to the CLIP loss distributes the embeddings more uniformly in the representational space, closing the gap. In our experiments, we show that the modified representational space achieves better performance than default CLIP loss in downstream tasks such as zero-shot image classification and multi-modal arithmetic.
Authors: Quanqi Hu, Qi Qi, Zhaosong Lu, Tianbao Yang
Abstract: In this paper, we study a class of non-smooth non-convex problems in the form of $\min_{x}[\max_{y\in Y}\phi(x, y) - \max_{z\in Z}\psi(x, z)]$, where both $\Phi(x) = \max_{y\in Y}\phi(x, y)$ and $\Psi(x)=\max_{z\in Z}\psi(x, z)$ are weakly convex functions, and $\phi(x, y), \psi(x, z)$ are strongly concave functions in terms of $y$ and $z$, respectively. It covers two families of problems that have been studied but are missing single-loop stochastic algorithms, i.e., difference of weakly convex functions and weakly convex strongly-concave min-max problems. We propose a stochastic Moreau envelope approximate gradient method dubbed SMAG, the first single-loop algorithm for solving these problems, and provide a state-of-the-art non-asymptotic convergence rate. The key idea of the design is to compute an approximate gradient of the Moreau envelopes of $\Phi, \Psi$ using only one step of stochastic gradient update of the primal and dual variables. Empirically, we conduct experiments on positive-unlabeled (PU) learning and partial area under ROC curve (pAUC) optimization with an adversarial fairness regularizer to validate the effectiveness of our proposed algorithms.
Authors: Alexander Windmann, Philipp Wittenberg, Marvin Schieseck, Oliver Niggemann
Abstract: In Industry 4.0, Cyber-Physical Systems (CPS) generate vast data sets that can be leveraged by Artificial Intelligence (AI) for applications including predictive maintenance and production planning. However, despite the demonstrated potential of AI, its widespread adoption in sectors like manufacturing remains limited. Our comprehensive review of recent literature, including standards and reports, pinpoints key challenges: system integration, data-related issues, managing workforce-related concerns and ensuring trustworthy AI. A quantitative analysis highlights particular challenges and topics that are important for practitioners but still need to be sufficiently investigated by academics. The paper briefly discusses existing solutions to these challenges and proposes avenues for future research. We hope that this survey serves as a resource for practitioners evaluating the cost-benefit implications of AI in CPS and for researchers aiming to address these urgent challenges.
Authors: Michael Munn, Benoit Dherin, Javier Gonzalvo
Abstract: There has been considerable effort to better understand the generalization capabilities of deep neural networks both as a means to unlock a theoretical understanding of their success as well as providing directions for further improvements. In this paper, we investigate margin-based multiclass generalization bounds for neural networks which rely on a recent complexity measure, the geometric complexity, developed for neural networks. We derive a new upper bound on the generalization error which scales with the margin-normalized geometric complexity of the network and which holds for a broad family of data distributions and model classes. Our generalization bound is empirically investigated for a ResNet-18 model trained with SGD on the CIFAR-10 and CIFAR-100 datasets with both original and random labels.
Authors: Charles Guille-Escuret, Eugene Ndiaye
Abstract: Conformal prediction methodologies have significantly advanced the quantification of uncertainties in predictive models. Yet, the construction of confidence regions for model parameters presents a notable challenge, often necessitating stringent assumptions regarding data distribution or merely providing asymptotic guarantees. We introduce a novel approach termed CCR, which employs a combination of conformal prediction intervals for the model outputs to establish confidence regions for model parameters. We present coverage guarantees under minimal assumptions on noise and that is valid in finite sample regime. Our approach is applicable to both split conformal predictions and black-box methodologies including full or cross-conformal approaches. In the specific case of linear models, the derived confidence region manifests as the feasible set of a Mixed-Integer Linear Program (MILP), facilitating the deduction of confidence intervals for individual parameters and enabling robust optimization. We empirically compare CCR to recent advancements in challenging settings such as with heteroskedastic and non-Gaussian noise.
Authors: Ali H\"urriyeto\u{g}lu, Osman Mutlu, F{\i}rat Duru\c{s}an, Erdem Y\"or\"uk
Abstract: GLOCON is a database of contentious events automatically extracted from national news sources from various countries in multiple languages. National news sources are utilized, and complete news archives are processed to create an event list for each source. Automation is achieved using a gold standard corpus sampled randomly from complete news archives (Y\"or\"uk et al. 2022) and all annotated by at least two domain experts based on the event definition provided in Duru\c{s}an et al. (2022).
Authors: Aditya Gunturu, Yi Wen, Jarin Thundathil, Nandi Zhang, Rubaiat Habib Kazi, Ryo Suzuki
Abstract: We introduce Augmented Physics, a machine learning-powered tool designed for creating interactive physics simulations from static textbook diagrams. Leveraging computer vision techniques, such as Segment Anything and OpenCV, our web-based system enables users to semi-automatically extract diagrams from physics textbooks and then generate interactive simulations based on the extracted content. These interactive diagrams are seamlessly integrated into scanned textbook pages, facilitating interactive and personalized learning experiences across various physics concepts, including gravity, optics, circuits, and kinematics. Drawing on an elicitation study with seven physics instructors, we explore four key augmentation techniques: 1) augmented experiments, 2) animated diagrams, 3) bi-directional manipulatives, and 4) parameter visualization. We evaluate our system through technical evaluation, a usability study (N=12), and expert interviews (N=12). The study findings suggest that our system can facilitate more engaging and personalized learning experiences in physics education.
Authors: Ajinkya Borle, Ameya Bhave
Abstract: Biclustering is a problem in machine learning and data mining that seeks to group together rows and columns of a dataset according to certain criteria. In this work, we highlight the natural relation that quantum computing models like boson and Gaussian boson sampling (GBS) have to this problem. We first explore the use of boson sampling to identify biclusters based on matrix permanents. We then propose a heuristic that finds clusters in a dataset using Gaussian boson sampling by (i) converting the dataset into a bipartite graph and then (ii) running GBS to find the densest sub-graph(s) within the larger bipartite graph. Our simulations for the above proposed heuristics show promising results for future exploration in this area.
Authors: Brian A. Yuan, Joseph G. Makin
Abstract: Recent work on intracranial brain-machine interfaces has demonstrated that spoken speech can be decoded with high accuracy, essentially by treating the problem as an instance of supervised learning and training deep neural networks to map from neural activity to text. However, such networks pay for their expressiveness with very large numbers of labeled data, a requirement that is particularly burdensome for invasive neural recordings acquired from human patients. On the other hand, these patients typically produce speech outside of the experimental blocks used for training decoders. Making use of such data, and data from other patients, to improve decoding would ease the burden of data collection -- especially onerous for dys- and anarthric patients. Here we demonstrate that this is possible, by reengineering wav2vec -- a simple, self-supervised, fully convolutional model that learns latent representations of audio using a noise-contrastive loss -- for electrocorticographic (ECoG) data. We train this model on unlabelled ECoG recordings, and subsequently use it to transform ECoG from labeled speech sessions into wav2vec's representation space, before finally training a supervised encoder-decoder to map these representations to text. We experiment with various numbers of labeled blocks; for almost all choices, the new representations yield superior decoding performance to the original ECoG data, and in no cases do they yield worse. Performance can also be improved in some cases by pretraining wav2vec on another patient's data. In the best cases, wav2vec's representations decrease word error rates over the original data by upwards of 50%.
Authors: Mina Arzaghi, Florian Carichon, Golnoosh Farnadi
Abstract: Large Language Models (LLMs) are increasingly integrated into critical decision-making processes, such as loan approvals and visa applications, where inherent biases can lead to discriminatory outcomes. In this paper, we examine the nuanced relationship between demographic attributes and socioeconomic biases in LLMs, a crucial yet understudied area of fairness in LLMs. We introduce a novel dataset of one million English sentences to systematically quantify socioeconomic biases across various demographic groups. Our findings reveal pervasive socioeconomic biases in both established models such as GPT-2 and state-of-the-art models like Llama 2 and Falcon. We demonstrate that these biases are significantly amplified when considering intersectionality, with LLMs exhibiting a remarkable capacity to extract multiple demographic attributes from names and then correlate them with specific socioeconomic biases. This research highlights the urgent necessity for proactive and robust bias mitigation techniques to safeguard against discriminatory outcomes when deploying these powerful models in critical real-world applications.
Authors: Haya Diwan, Jinrui Gou, Cameron Musco, Christopher Musco, Torsten Suel
Abstract: There has been significant recent interest in graph-based nearest neighbor search methods, many of which are centered on the construction of navigable graphs over high-dimensional point sets. A graph is navigable if we can successfully move from any starting node to any target node using a greedy routing strategy where we always move to the neighbor that is closest to the destination according to a given distance function. The complete graph is navigable for any point set, but the important question for applications is if sparser graphs can be constructed. While this question is fairly well understood in low-dimensions, we establish some of the first upper and lower bounds for high-dimensional point sets. First, we give a simple and efficient way to construct a navigable graph with average degree $O(\sqrt{n \log n })$ for any set of $n$ points, in any dimension, for any distance function. We compliment this result with a nearly matching lower bound: even under the Euclidean metric in $O(\log n)$ dimensions, a random point set has no navigable graph with average degree $O(n^{\alpha})$ for any $\alpha < 1/2$. Our lower bound relies on sharp anti-concentration bounds for binomial random variables, which we use to show that the near-neighborhoods of a set of random points do not overlap significantly, forcing any navigable graph to have many edges.
Authors: Shubham Vatsal, Ayush Singh
Abstract: Large language models (LLMs) have shown remarkable performance on many tasks in different domains. However, their performance in closed-book biomedical machine reading comprehension (MRC) has not been evaluated in depth. In this work, we evaluate GPT on four closed-book biomedical MRC benchmarks. We experiment with different conventional prompting techniques as well as introduce our own novel prompting method. To solve some of the retrieval problems inherent to LLMs, we propose a prompting strategy named Implicit Retrieval Augmented Generation (RAG) that alleviates the need for using vector databases to retrieve important chunks in traditional RAG setups. Moreover, we report qualitative assessments on the natural language generation outputs from our approach. The results show that our new prompting technique is able to get the best performance in two out of four datasets and ranks second in rest of them. Experiments show that modern-day LLMs like GPT even in a zero-shot setting can outperform supervised models, leading to new state-of-the-art (SoTA) results on two of the benchmarks.
Authors: Alexander Soen, Hisham Husain, Philip Schulz, Vu Nguyen
Abstract: Classification with rejection emerges as a learning paradigm which allows models to abstain from making predictions. The predominant approach is to alter the supervised learning pipeline by augmenting typical loss functions, letting model rejection incur a lower loss than an incorrect prediction. Instead, we propose a different distributional perspective, where we seek to find an idealized data distribution which maximizes a pretrained model's performance. This can be formalized via the optimization of a loss's risk with a $ \phi$-divergence regularization term. Through this idealized distribution, a rejection decision can be made by utilizing the density ratio between this distribution and the data distribution. We focus on the setting where our $ \phi $-divergences are specified by the family of $ \alpha $-divergence. Our framework is tested empirically over clean and noisy datasets.
Authors: Arpita Soni, Sujatha Alla, Suresh Dodda, Hemanth Volikatla
Abstract: The market for domestic robots made to perform household chores is growing as these robots relieve people of everyday responsibilities. Domestic robots are generally welcomed for their role in easing human labor, in contrast to industrial robots, which are frequently criticized for displacing human workers. But before these robots can carry out domestic chores, they need to become proficient in several minor activities, such as recognizing their surroundings, making decisions, and picking up on human behaviors. Reinforcement learning, or RL, has emerged as a key robotics technology that enables robots to interact with their environment and learn how to optimize their actions to maximize rewards. However, the goal of Deep Reinforcement Learning is to address more complicated, continuous action-state spaces in real-world settings by combining RL with Neural Networks. The efficacy of DeepRL can be further augmented through interactive feedback, in which a trainer offers real-time guidance to expedite the robot's learning process. Nevertheless, the current methods have drawbacks, namely the transient application of guidance that results in repeated learning under identical conditions. Therefore, we present a novel method to preserve and reuse information and advice via Deep Interactive Reinforcement Learning, which utilizes a persistent rule-based system. This method not only expedites the training process but also lessens the number of repetitions that instructors will have to carry out. This study has the potential to advance the development of household robots and improve their effectiveness and efficiency as learners.
Authors: Linjia Kang, Songhua Zhou, Shuyan Fang, Shichao Liu, Wen Zhang
Abstract: Accurate prediction of molecular properties is critical in the field of drug discovery. However, existing methods do not fully consider the fact that molecules in the real world usually possess multiple property labels, and complex high-order relationships may exist among these labels. Therefore, molecular representation learning models should generate differential molecular representations that consider multi-granularity correlation information among tasks. To this end, our research introduces a Hierarchical Prompted Molecular Representation Learning Framework (HiPM), which enhances the differential expression of tasks in molecular representations through task-aware prompts, and utilizes shared information among labels to mitigate negative transfer between different tasks. HiPM primarily consists of two core components: the Molecular Representation Encoder (MRE) and the Task-Aware Prompter (TAP). The MRE employs a hierarchical message-passing network architecture to capture molecular features at both the atomic and motif levels, while the TAP uses agglomerative hierarchical clustering to build a prompt tree that reflects the affinity and distinctiveness of tasks, enabling the model to effectively handle the complexity of multi-label property predictions. Extensive experiments demonstrate that HiPM achieves state-of-the-art performance across various multi-label datasets, offering a new perspective on multi-label molecular representation learning.
Authors: S. Mostafa Mousavi, Marc Stogaitis, Tajinder Gadh, Richard M Allen, Alexei Barski, Robert Bosch, Patrick Robertson, Nivetha Thiruverahan, Youngmin Cho
Abstract: This paper presents a novel approach for estimating the ground shaking intensity using social media data and CCTV footage. Employing the Gemini Pro (Reid et al. 2024) model, a multi-modal language model, we demonstrate the ability to extract relevant information from unstructured data utilizing generative AI and natural language processing. The model output, in the form of Modified Mercalli Intensity (MMI) values, align well with independent observational data. Furthermore, our results suggest that beyond its advanced visual and auditory understanding abilities, Gemini appears to utilize additional sources of knowledge, including a simplified understanding of the general relationship between earthquake magnitude, distance, and MMI intensity, which it presumably acquired during its training, in its reasoning and decision-making processes. These findings raise intriguing questions about the extent of Gemini's general understanding of the physical world and its phenomena. The ability of Gemini to generate results consistent with established scientific knowledge highlights the potential of LLMs like Gemini in augmenting our understanding of complex physical phenomena such as earthquakes. More specifically, the results of this study highlight the potential of LLMs like Gemini to revolutionize citizen seismology by enabling rapid, effective, and flexible analysis of crowdsourced data from eyewitness accounts for assessing earthquake impact and providing crisis situational awareness. This approach holds great promise for improving early warning systems, disaster response, and overall resilience in earthquake-prone regions. This study provides a significant step toward harnessing the power of social media and AI for earthquake disaster mitigation.
Authors: Satwik Kundu, Swaroop Ghosh
Abstract: The high expenses imposed by current quantum cloud providers, coupled with the escalating need for quantum resources, may incentivize the emergence of cheaper cloud-based quantum services from potentially untrusted providers. Deploying or hosting quantum models, such as Quantum Neural Networks (QNNs), on these untrusted platforms introduces a myriad of security concerns, with the most critical one being model theft. This vulnerability stems from the cloud provider's full access to these circuits during training and/or inference. In this work, we introduce STIQ, a novel ensemble-based strategy designed to safeguard QNNs against such cloud-based adversaries. Our method innovatively trains two distinct QNNs concurrently, hosting them on same or different platforms, in a manner that each network yields obfuscated outputs rendering the individual QNNs ineffective for adversaries operating within cloud environments. However, when these outputs are combined locally (using an aggregate function), they reveal the correct result. Through extensive experiments across various QNNs and datasets, our technique has proven to effectively masks the accuracy and losses of the individually hosted models by upto 76\%, albeit at the expense of $\leq 2\times$ increase in the total computational overhead. This trade-off, however, is a small price to pay for the enhanced security and integrity of QNNs in a cloud-based environment prone to untrusted adversaries. We also demonstrated STIQ's practical application by evaluating it on real 127-qubit IBM\_Sherbrooke hardware, showing that STIQ achieves up to 60\% obfuscation, with combined performance comparable to an unobfuscated model.
Authors: Matthew Fahrbach, Srikumar Ramalingam, Morteza Zadimoghaddam, Sara Ahmadian, Gui Citovsky, Giulia DeSalvo
Abstract: We propose a novel subset selection task called min-distance diverse data summarization ($\textsf{MDDS}$), which has a wide variety of applications in machine learning, e.g., data sampling and feature selection. Given a set of points in a metric space, the goal is to maximize an objective that combines the total utility of the points and a diversity term that captures the minimum distance between any pair of selected points, subject to the constraint $|S| \le k$. For example, the points may correspond to training examples in a data sampling problem, e.g., learned embeddings of images extracted from a deep neural network. This work presents the $\texttt{GIST}$ algorithm, which achieves a $\frac{2}{3}$-approximation guarantee for $\textsf{MDDS}$ by approximating a series of maximum independent set problems with a bicriteria greedy algorithm. We also prove a complementary $(\frac{2}{3}+\varepsilon)$-hardness of approximation, for any $\varepsilon > 0$. Finally, we provide an empirical study that demonstrates $\texttt{GIST}$ outperforms existing methods for $\textsf{MDDS}$ on synthetic data, and also for a real-world image classification experiment the studies single-shot subset selection for ImageNet.
Authors: Divya Nori, Wengong Jin
Abstract: The growing significance of RNA engineering in diverse biological applications has spurred interest in developing AI methods for structure-based RNA design. While diffusion models have excelled in protein design, adapting them for RNA presents new challenges due to RNA's conformational flexibility and the computational cost of fine-tuning large structure prediction models. To this end, we propose RNAFlow, a flow matching model for protein-conditioned RNA sequence-structure design. Its denoising network integrates an RNA inverse folding model and a pre-trained RosettaFold2NA network for generation of RNA sequences and structures. The integration of inverse folding in the structure denoising process allows us to simplify training by fixing the structure prediction network. We further enhance the inverse folding model by conditioning it on inferred conformational ensembles to model dynamic RNA conformations. Evaluation on protein-conditioned RNA structure and sequence generation tasks demonstrates RNAFlow's advantage over existing RNA design methods.
Authors: Qin Yang, Meisam Mohammad, Han Wang, Ali Payani, Ashish Kundu, Kai Shu, Yan Yan, Yuan Hong
Abstract: Differentially Private Stochastic Gradient Descent (DP-SGD) and its variants have been proposed to ensure rigorous privacy for fine-tuning large-scale pre-trained language models. However, they rely heavily on the Gaussian mechanism, which may overly perturb the gradients and degrade the accuracy, especially in stronger privacy regimes (e.g., the privacy budget $\epsilon < 3$). To address such limitations, we propose a novel Language Model-based Optimal Differential Privacy (LMO-DP) mechanism, which takes the first step to enable the tight composition of accurately fine-tuning (large) language models with a sub-optimal DP mechanism, even in strong privacy regimes (e.g., $0.1\leq \epsilon<3$). Furthermore, we propose a novel offline optimal noise search method to efficiently derive the sub-optimal DP that significantly reduces the noise magnitude. For instance, fine-tuning RoBERTa-large (with 300M parameters) on the SST-2 dataset can achieve an accuracy of 92.20% (given $\epsilon=0.3$, $\delta=10^{-10}$) by drastically outperforming the Gaussian mechanism (e.g., $\sim 50\%$ for small $\epsilon$ and $\delta$). We also draw similar findings on the text generation tasks on GPT-2. Finally, to our best knowledge, LMO-DP is also the first solution to accurately fine-tune Llama-2 with strong differential privacy guarantees. The code will be released soon and available upon request.
Authors: Tianshu Chu, Dachuan Xu, Wei Yao, Jin Zhang
Abstract: While stochastic bilevel optimization methods have been extensively studied for addressing large-scale nested optimization problems in machine learning, it remains an open question whether the optimal complexity bounds for solving bilevel optimization are the same as those in single-level optimization. Our main result resolves this question: SPABA, an adaptation of the PAGE method for nonconvex optimization in (Li et al., 2021) to the bilevel setting, can achieve optimal sample complexity in both the finite-sum and expectation settings. We show the optimality of SPABA by proving that there is no gap in complexity analysis between stochastic bilevel and single-level optimization when implementing PAGE. Notably, as indicated by the results of (Dagr\'eou et al., 2022), there might exist a gap in complexity analysis when implementing other stochastic gradient estimators, like SGD and SAGA. In addition to SPABA, we propose several other single-loop stochastic bilevel algorithms, that either match or improve the state-of-the-art sample complexity results, leveraging our convergence rate and complexity analysis. Numerical experiments demonstrate the superior practical performance of the proposed methods.
Authors: Isha Chaudhary, Qian Hu, Manoj Kumar, Morteza Ziyadi, Rahul Gupta, Gagandeep Singh
Abstract: Large Language Models (LLMs) can produce responses that exhibit social biases and support stereotypes. However, conventional benchmarking is insufficient to thoroughly evaluate LLM bias, as it can not scale to large sets of prompts and provides no guarantees. Therefore, we propose a novel certification framework QuaCer-B (Quantitative Certification of Bias) that provides formal guarantees on obtaining unbiased responses from target LLMs under large sets of prompts. A certificate consists of high-confidence bounds on the probability of obtaining biased responses from the LLM for any set of prompts containing sensitive attributes, sampled from a distribution. We illustrate the bias certification in LLMs for prompts with various prefixes drawn from given distributions. We consider distributions of random token sequences, mixtures of manual jailbreaks, and jailbreaks in the LLM's embedding space to certify its bias. We certify popular LLMs with QuaCer-B and present novel insights into their biases.
Authors: Zhong Zheng, Haochen Zhang, Lingzhou Xue
Abstract: In this paper, we consider model-free federated reinforcement learning for tabular episodic Markov decision processes. Under the coordination of a central server, multiple agents collaboratively explore the environment and learn an optimal policy without sharing their raw data. Despite recent advances in federated Q-learning algorithms achieving near-linear regret speedup with low communication cost, existing algorithms only attain suboptimal regrets compared to the information bound. We propose a novel model-free federated Q-learning algorithm, termed FedQ-Advantage. Our algorithm leverages reference-advantage decomposition for variance reduction and operates under two distinct mechanisms: synchronization between the agents and the server, and policy update, both triggered by events. We prove that our algorithm not only requires a lower logarithmic communication cost but also achieves an almost optimal regret, reaching the information bound up to a logarithmic factor and near-linear regret speedup compared to its single-agent counterpart when the time horizon is sufficiently large.
Authors: Yasi Zhang, Peiyu Yu, Yaxuan Zhu, Yingshan Chang, Feng Gao, Ying Nian Wu, Oscar Leong
Abstract: Generative models based on flow matching have attracted significant attention for their simplicity and superior performance in high-resolution image synthesis. By leveraging the instantaneous change-of-variables formula, one can directly compute image likelihoods from a learned flow, making them enticing candidates as priors for downstream tasks such as inverse problems. In particular, a natural approach would be to incorporate such image probabilities in a maximum-a-posteriori (MAP) estimation problem. A major obstacle, however, lies in the slow computation of the log-likelihood, as it requires backpropagating through an ODE solver, which can be prohibitively slow for high-dimensional problems. In this work, we propose an iterative algorithm to approximate the MAP estimator efficiently to solve a variety of linear inverse problems. Our algorithm is mathematically justified by the observation that the MAP objective can be approximated by a sum of $N$ ``local MAP'' objectives, where $N$ is the number of function evaluations. By leveraging Tweedie's formula, we show that we can perform gradient steps to sequentially optimize these objectives. We validate our approach for various linear inverse problems, such as super-resolution, deblurring, inpainting, and compressed sensing, and demonstrate that we can outperform other methods based on flow matching.
Authors: Simranjit Singh, Georgios Pavlakos, Dimitrios Stamoulis
Abstract: As interest in "reformulating" the 3D Visual Question Answering (VQA) problem in the context of foundation models grows, it is imperative to assess how these new paradigms influence existing closed-vocabulary datasets. In this case study, we evaluate the zero-shot performance of foundational models (GPT-4 Vision and GPT-4) on well-established 3D VQA benchmarks, namely 3D-VQA and ScanQA. We provide an investigation to contextualize the performance of GPT-based agents relative to traditional modeling approaches. We find that GPT-based agents without any fine-tuning perform on par with the closed vocabulary approaches. Our findings corroborate recent results that "blind" models establish a surprisingly strong baseline in closed-vocabulary settings. We demonstrate that agents benefit significantly from scene-specific vocabulary via in-context textual grounding. By presenting a preliminary comparison with previous baselines, we hope to inform the community's ongoing efforts to refine multi-modal 3D benchmarks.
Authors: Siyuan Guo, Chi Zhang, Karthika Mohan, Ferenc Husz\'ar, Bernhard Sch\"olkopf
Abstract: We study causal effect estimation in a setting where the data are not i.i.d. (independent and identically distributed). We focus on exchangeable data satisfying an assumption of independent causal mechanisms. Traditional causal effect estimation frameworks, e.g., relying on structural causal models and do-calculus, are typically limited to i.i.d. data and do not extend to more general exchangeable generative processes, which naturally arise in multi-environment data. To address this gap, we develop a generalized framework for exchangeable data and introduce a truncated factorization formula that facilitates both the identification and estimation of causal effects in our setting. To illustrate potential applications, we introduce a causal P\'olya urn model and demonstrate how intervention propagates effects in exchangeable data settings. Finally, we develop an algorithm that performs simultaneous causal discovery and effect estimation given multi-environment data.
Authors: Dhiraj Neupane, Mohamed Reda Bouadjenek, Richard Dazeley, Sunil Aryal
Abstract: In this era of advanced manufacturing, it's now more crucial than ever to diagnose machine faults as early as possible to guarantee their safe and efficient operation. With the massive surge in industrial big data and advancement in sensing and computational technologies, data-driven Machinery Fault Diagnosis (MFD) solutions based on machine/deep learning approaches have been used ubiquitously in manufacturing. Timely and accurately identifying faulty machine signals is vital in industrial applications for which many relevant solutions have been proposed and are reviewed in many articles. Despite the availability of numerous solutions and reviews on MFD, existing works often lack several aspects. Most of the available literature has limited applicability in a wide range of manufacturing settings due to their concentration on a particular type of equipment or method of analysis. Additionally, discussions regarding the challenges associated with implementing data-driven approaches, such as dealing with noisy data, selecting appropriate features, and adapting models to accommodate new or unforeseen faults, are often superficial or completely overlooked. Thus, this survey provides a comprehensive review of the articles using different types of machine learning approaches for the detection and diagnosis of various types of machinery faults, highlights their strengths and limitations, provides a review of the methods used for condition-based analyses, comprehensively discusses the available machinery fault datasets, introduces future researchers to the possible challenges they have to encounter while using these approaches for MFD and recommends the probable solutions to mitigate those problems. The future research prospects are also pointed out for a better understanding of the field. We believe this article will help researchers and contribute to the further development of the field.
Authors: Silvia Garc\'ia M\'endez, F\'atima Leal, Benedita Malheiro, Juan Carlos Burguillo Rial, Bruno Veloso, Adriana E. Chis, Horacio Gonz\'alez V\'elez
Abstract: Data crowdsourcing is a data acquisition process where groups of voluntary contributors feed platforms with highly relevant data ranging from news, comments, and media to knowledge and classifications. It typically processes user-generated data streams to provide and refine popular services such as wikis, collaborative maps, e-commerce sites, and social networks. Nevertheless, this modus operandi raises severe concerns regarding ill-intentioned data manipulation in adversarial environments. This paper presents a simulation, modelling, and classification approach to automatically identify human and non-human (bots) as well as benign and malign contributors by using data fabrication to balance classes within experimental data sets, data stream modelling to build and update contributor profiles and, finally, autonomic data stream classification. By employing WikiVoyage - a free worldwide wiki travel guide open to contribution from the general public - as a testbed, our approach proves to significantly boost the confidence and quality of the classifier by using a class-balanced data stream, comprising both real and synthetic data. Our empirical results show that the proposed method distinguishes between benign and malign bots as well as human contributors with a classification accuracy of up to 92 %.
Authors: Ruipeng Zhang, Ziqing Fan, Jiangchao Yao, Ya Zhang, Yanfeng Wang
Abstract: This paper presents a Domain-Inspired Sharpness-Aware Minimization (DISAM) algorithm for optimization under domain shifts. It is motivated by the inconsistent convergence degree of SAM across different domains, which induces optimization bias towards certain domains and thus impairs the overall convergence. To address this issue, we consider the domain-level convergence consistency in the sharpness estimation to prevent the overwhelming (deficient) perturbations for less (well) optimized domains. Specifically, DISAM introduces the constraint of minimizing variance in the domain loss, which allows the elastic gradient calibration in perturbation generation: when one domain is optimized above the averaging level \textit{w.r.t.} loss, the gradient perturbation towards that domain will be weakened automatically, and vice versa. Under this mechanism, we theoretically show that DISAM can achieve faster overall convergence and improved generalization in principle when inconsistent convergence emerges. Extensive experiments on various domain generalization benchmarks show the superiority of DISAM over a range of state-of-the-art methods. Furthermore, we show the superior efficiency of DISAM in parameter-efficient fine-tuning combined with the pretraining models. The source code is released at https://github.com/MediaBrain-SJTU/DISAM.
Authors: Daniele Dell'Erba, Yong Li, Sven Schewe
Abstract: We propose DFAMiner, a passive learning tool for learning minimal separating deterministic finite automata (DFA) from a set of labelled samples. Separating automata are an interesting class of automata that occurs generally in regular model checking and has raised interest in foundational questions of parity game solving. We first propose a simple and linear-time algorithm that incrementally constructs a three-valued DFA (3DFA) from a set of labelled samples given in the usual lexicographical order. This 3DFA has accepting and rejecting states as well as don't-care states, so that it can exactly recognise the labelled examples. We then apply our tool to mining a minimal separating DFA for the labelled samples by minimising the constructed automata via a reduction to solving SAT problems. Empirical evaluation shows that our tool outperforms current state-of-the-art tools significantly on standard benchmarks for learning minimal separating DFAs from samples. Progress in the efficient construction of separating DFAs can also lead to finding the lower bound of parity game solving, where we show that DFAMiner can create optimal separating automata for simple languages with up to 7 colours. Future improvements might offer inroads to better data structures.
Authors: Julia Jentsch, Ali Burak \"Unal, \c{S}eyma Selcan Ma\u{g}ara, Mete Akg\"un
Abstract: Handling missing data is crucial in machine learning, but many datasets contain gaps due to errors or non-response. Unlike traditional methods such as listwise deletion, which are simple but inadequate, the literature offers more sophisticated and effective methods, thereby improving sample size and accuracy. However, these methods require accessing the whole dataset, which contradicts the privacy regulations when the data is distributed among multiple sources. Especially in the medical and healthcare domain, such access reveals sensitive information about patients. This study addresses privacy-preserving imputation methods for sensitive data using secure multi-party computation, enabling secure computations without revealing any party's sensitive information. In this study, we realized the mean, median, regression, and kNN imputation methods in a privacy-preserving way. We specifically target the medical and healthcare domains considering the significance of protection of the patient data, showcasing our methods on a diabetes dataset. Experiments on the diabetes dataset validated the correctness of our privacy-preserving imputation methods, yielding the largest error around $3 \times 10^{-3}$, closely matching plaintext methods. We also analyzed the scalability of our methods to varying numbers of samples, showing their applicability to real-world healthcare problems. Our analysis demonstrated that all our methods scale linearly with the number of samples. Except for kNN, the runtime of all our methods indicates that they can be utilized for large datasets.
Authors: Ruichang Zhang, Youcheng Sun, Mustafa A. Mustafa
Abstract: Smart meters play a crucial role in enhancing energy management and efficiency, but they raise significant privacy concerns by potentially revealing detailed user behaviors through energy consumption patterns. Recent scholarly efforts have focused on developing battery-aided load-shaping techniques to protect user privacy while balancing costs. This paper proposes a novel deep reinforcement learning-based load-shaping algorithm (PLS-DQN) designed to protect user privacy by proactively creating artificial load signatures that mislead potential attackers. We evaluate our proposed algorithm against a non-intrusive load monitoring (NILM) adversary. The results demonstrate that our approach not only effectively conceals real energy usage patterns but also outperforms state-of-the-art methods in enhancing user privacy while maintaining cost efficiency.
Authors: Chenze Shao, Fandong Meng, Yijin Liu, Jie Zhou
Abstract: Language generation based on maximum likelihood estimation (MLE) has become the fundamental approach for text generation. Maximum likelihood estimation is typically performed by minimizing the log-likelihood loss, also known as the logarithmic score in statistical decision theory. The logarithmic score is strictly proper in the sense that it encourages honest forecasts, where the expected score is maximized only when the model reports true probabilities. Although many strictly proper scoring rules exist, the logarithmic score is the only local scoring rule among them that depends exclusively on the probability of the observed sample, making it capable of handling the exponentially large sample space of natural text. In this work, we propose a straightforward strategy for adapting scoring rules to language generation, allowing for language modeling with any non-local scoring rules. Leveraging this strategy, we train language generation models using two classic strictly proper scoring rules, the Brier score and the Spherical score, as alternatives to the logarithmic score. Experimental results indicate that simply substituting the loss function, without adjusting other hyperparameters, can yield substantial improvements in model's generation capabilities. Moreover, these improvements can scale up to large language models (LLMs) such as LLaMA-7B and LLaMA-13B. Source code: \url{https://github.com/shaochenze/ScoringRulesLM}.
Authors: Giacomo Acciarini, Laurent Beauregard, Dario Izzo
Abstract: Low-thrust trajectories play a crucial role in optimizing scientific output and cost efficiency in asteroid belt missions. Unlike high-thrust transfers, low-thrust trajectories require solving complex optimal control problems. This complexity grows exponentially with the number of asteroids visited due to orbital mechanics intricacies. In the literature, methods for approximating low-thrust transfers without full optimization have been proposed, including analytical and machine learning techniques. In this work, we propose new analytical approximations and compare their accuracy and performance to machine learning methods. While analytical approximations leverage orbit theory to estimate trajectory costs, machine learning employs a more black-box approach, utilizing neural networks to predict optimal transfers based on various attributes. We build a dataset of about 3 million transfers, found by solving the time and fuel optimal control problems, for different time of flights, which we also release open-source. Comparison between the two methods on this database reveals the superiority of machine learning, especially for longer transfers. Despite challenges such as multi revolution transfers, both approaches maintain accuracy within a few percent in the final mass errors, on a database of trajectories involving numerous asteroids. This work contributes to the efficient exploration of mission opportunities in the asteroid belt, providing insights into the strengths and limitations of different approximation strategies.
Authors: Hiroshi Takahashi, Tomoharu Iwata, Atsutoshi Kumagai, Yuuki Yamanaka
Abstract: Semi-supervised anomaly detection, which aims to improve the performance of the anomaly detector by using a small amount of anomaly data in addition to unlabeled data, has attracted attention. Existing semi-supervised approaches assume that unlabeled data are mostly normal. They train the anomaly detector to minimize the anomaly scores for the unlabeled data, and to maximize those for the anomaly data. However, in practice, the unlabeled data are often contaminated with anomalies. This weakens the effect of maximizing the anomaly scores for anomalies, and prevents us from improving the detection performance. To solve this problem, we propose the positive-unlabeled autoencoder, which is based on positive-unlabeled learning and the anomaly detector such as the autoencoder. With our approach, we can approximate the anomaly scores for normal data using the unlabeled and anomaly data. Therefore, without the labeled normal data, we can train the anomaly detector to minimize the anomaly scores for normal data, and to maximize those for the anomaly data. In addition, our approach is applicable to various anomaly detectors such as the DeepSVDD. Experiments on various datasets show that our approach achieves better detection performance than existing approaches.
Authors: Shohei Enomoto
Abstract: Deep neural networks (DNNs) struggle to generalize to out-of-distribution domains that are different from those in training despite their impressive performance. In practical applications, it is important for DNNs to have both high standard accuracy and robustness against out-of-distribution domains. One technique that achieves both of these improvements is disentangled learning with mixture distribution via auxiliary batch normalization layers (ABNs). This technique treats clean and transformed samples as different domains, allowing a DNN to learn better features from mixed domains. However, if we distinguish the domains of the samples based on entropy, we find that some transformed samples are drawn from the same domain as clean samples, and these samples are not completely different domains. To generate samples drawn from a completely different domain than clean samples, we hypothesize that transforming clean high-entropy samples to further increase the entropy generates out-of-distribution samples that are much further away from the in-distribution domain. On the basis of the hypothesis, we propose high entropy propagation~(EntProp), which feeds high-entropy samples to the network that uses ABNs. We introduce two techniques, data augmentation and free adversarial training, that increase entropy and bring the sample further away from the in-distribution domain. These techniques do not require additional training costs. Our experimental results show that EntProp achieves higher standard accuracy and robustness with a lower training cost than the baseline methods. In particular, EntProp is highly effective at training on small datasets.
Authors: Gaoxiang Zhao, Lu Wang, Xiaoqiang Wang
Abstract: The effectiveness of anomaly signal detection can be significantly undermined by the inherent uncertainty of relying on one specified model. Under the framework of model average methods, this paper proposes a novel criterion to select the weights on aggregation of multiple models, wherein the focal loss function accounts for the classification of extremely imbalanced data. This strategy is further integrated into Random Forest algorithm by replacing the conventional voting method. We have evaluated the proposed method on benchmark datasets across various domains, including network intrusion. The findings indicate that our proposed method not only surpasses the model averaging with typical loss functions but also outstrips common anomaly detection algorithms in terms of accuracy and robustness.
Authors: Antonio Briola, Silvia Bartolucci, Tomaso Aste
Abstract: We introduce a novel large-scale deep learning model for Limit Order Book mid-price changes forecasting, and we name it `HLOB'. This architecture (i) exploits the information encoded by an Information Filtering Network, namely the Triangulated Maximally Filtered Graph, to unveil deeper and non-trivial dependency structures among volume levels; and (ii) guarantees deterministic design choices to handle the complexity of the underlying system by drawing inspiration from the groundbreaking class of Homological Convolutional Neural Networks. We test our model against 9 state-of-the-art deep learning alternatives on 3 real-world Limit Order Book datasets, each including 15 stocks traded on the NASDAQ exchange, and we systematically characterize the scenarios where HLOB outperforms state-of-the-art architectures. Our approach sheds new light on the spatial distribution of information in Limit Order Books and on its degradation over increasing prediction horizons, narrowing the gap between microstructural modeling and deep learning-based forecasting in high-frequency financial markets.
Authors: Nan Li, Bo Kang, Tijl De Bie
Abstract: Personalized recommendation systems often drive users towards more extreme content, exacerbating opinion polarization. While (content-aware) moderation has been proposed to mitigate these effects, such approaches risk curtailing the freedom of speech and of information. To address this concern, we propose and explore the feasibility of \emph{content-agnostic} moderation as an alternative approach for reducing polarization. Content-agnostic moderation does not rely on the actual content being moderated, arguably making it less prone to forms of censorship. We establish theoretically that content-agnostic moderation cannot be guaranteed to work in a fully generic setting. However, we show that it can often be effectively achieved in practice with plausible assumptions. We introduce two novel content-agnostic moderation methods that modify the recommendations from the content recommender to disperse user-item co-clusters without relying on content features. To evaluate the potential of content-agnostic moderation in controlled experiments, we built a simulation environment to analyze the closed-loop behavior of a system with a given set of users, recommendation system, and moderation approach. Through comprehensive experiments in this environment, we show that our proposed moderation methods significantly enhance stance neutrality and maintain high recommendation quality across various data scenarios. Our results indicate that achieving stance neutrality without direct content information is not only feasible but can also help in developing more balanced and informative recommendation systems without substantially degrading user engagement.
Authors: Linus Jeary, Tom Kuipers, Mehran Hosseini, Nicola Paoletti
Abstract: Conformal Prediction (CP) is a popular uncertainty quantification method that provides distribution-free, statistically valid prediction sets, assuming that training and test data are exchangeable. In such a case, CP's prediction sets are guaranteed to cover the (unknown) true test output with a user-specified probability. Nevertheless, this guarantee is violated when the data is subjected to adversarial attacks, which often result in a significant loss of coverage. Recently, several approaches have been put forward to recover CP guarantees in this setting. These approaches leverage variations of randomised smoothing to produce conservative sets which account for the effect of the adversarial perturbations. They are, however, limited in that they only support $\ell^2$-bounded perturbations and classification tasks. This paper introduces \emph{VRCP (Verifiably Robust Conformal Prediction)}, a new framework that leverages recent neural network verification methods to recover coverage guarantees under adversarial attacks. Our VRCP method is the first to support perturbations bounded by arbitrary norms including $\ell^1$, $\ell^2$, and $\ell^\infty$, as well as regression tasks. We evaluate and compare our approach on image classification tasks (CIFAR10, CIFAR100, and TinyImageNet) and regression tasks for deep reinforcement learning environments. In every case, VRCP achieves above nominal coverage and yields significantly more efficient and informative prediction regions than the SotA.
Authors: Haosheng Xu, Dongheng Qian, Jing Wang
Abstract: The use of machine learning methods for predicting the properties of crystalline materials encounters significant challenges, primarily related to input encoding, output versatility, and interpretability. Here, we introduce CrystalBERT, an adaptable transformer-based framework with novel structure that integrates space group, elemental, and unit cell information. The method's adaptability lies not only in its ability to seamlessly combine diverse features but also in its capability to accurately predict a wide range of physically important properties, including topological properties, superconducting transition temperatures, dielectric constants, and more. CrystalBERT also provides insightful physical interpretations regarding the features that most significantly influence the target properties. Our findings indicate that space group and elemental information are more important for predicting topological and superconducting properties, in contrast to some properties that primarily depend on the unit cell information. This underscores the intricate nature of topological and superconducting properties. By incorporating all these features, we achieve a high accuracy of 91% in topological classification, surpassing prior studies and identifying previously misclassified topological materials, further demonstrating the effectiveness of our model.
Authors: Ho Chun Wu, Esther Hoi Shan Lau, Paul Yuen, Kevin Hung, John Kwok Tai Chui, Andrew Kwok Fai Lui
Abstract: Pedestrian trajectory modelling in an urban complex is challenging because pedestrians can have many possible destinations, such as shops, escalators, and attractions. Moreover, weather and time-of-day may affect pedestrian behavior. In this paper, a new weather-time-trajectory fusion network (WTTFNet) is proposed to improve the performance of baseline deep neural network architecture. By incorporating weather and time-of-day information as an embedding structure, a novel WTTFNet based on gate multimodal unit is used to fuse the multimodal information and deep representation of trajectories. A joint loss function based on focal loss is used to co-optimize both the deep trajectory features and final classifier, which helps to improve the accuracy in predicting the intended destination of pedestrians and hence the trajectories under possible scenarios of class imbalances. Experimental results using the Osaka Asia and Pacific Trade Center (ATC) dataset shows improved performance of the proposed approach over state-of-the-art algorithms by 23.67% increase in classification accuracy, 9.16% and 7.07% reduction of average and final displacement error. The proposed approach may serve as an attractive approach for improving existing baseline trajectory prediction models when they are applied to scenarios with influences of weather-time conditions. It can be employed in numerous applications such as pedestrian facility engineering, public space development and technology-driven retail.
Authors: Namasivayam Kalithasan, Arnav Tuli, Vishal Bindal, Himanshu Gaurav Singh, Parag Singla, Rohan Paul
Abstract: Automatically detecting and recovering from failures is an important but challenging problem for autonomous robots. Most of the recent work on learning to plan from demonstrations lacks the ability to detect and recover from errors in the absence of an explicit state representation and/or a (sub-) goal check function. We propose an approach (blending learning with symbolic search) for automated error discovery and recovery, without needing annotated data of failures. Central to our approach is a neuro-symbolic state representation, in the form of dense scene graph, structured based on the objects present within the environment. This enables efficient learning of the transition function and a discriminator that not only identifies failures but also localizes them facilitating fast re-planning via computation of heuristic distance function. We also present an anytime version of our algorithm, where instead of recovering to the last correct state, we search for a sub-goal in the original plan minimizing the total distance to the goal given a re-planning budget. Experiments on a physics simulator with a variety of simulated failures show the effectiveness of our approach compared to existing baselines, both in terms of efficiency as well as accuracy of our recovery mechanism.
Authors: Peter Devine
Abstract: Training Large Language Models (LLMs) with Reinforcement Learning from AI Feedback (RLAIF) aligns model outputs more closely with human preferences. This involves an evaluator model ranking multiple candidate responses to user prompts. However, the rankings from popular evaluator models such as GPT-4 can be inconsistent. We propose the Repeat Ranking method - where we evaluate the same responses multiple times and train only on those responses which are consistently ranked. Using 2,714 prompts in 62 languages, we generated responses from 7 top multilingual LLMs and had GPT-4 rank them five times each. Evaluating on MT-Bench chat benchmarks in six languages, our method outperformed the standard practice of training on all available prompts. Our work highlights the quality versus quantity trade-off in RLAIF dataset generation and offers a stackable strategy for enhancing dataset and thus model quality.
Authors: Zhangyang Gao, Jue Wang, Cheng Tan, Lirong Wu, Yufei Huang, Siyuan Li, Zhirui Ye, Stan Z. Li
Abstract: Molecule inverse folding has been a long-standing challenge in chemistry and biology, with the potential to revolutionize drug discovery and material science. Despite specified models have been proposed for different small- or macro-molecules, few have attempted to unify the learning process, resulting in redundant efforts. Complementary to recent advancements in molecular structure prediction, such as RoseTTAFold All-Atom and AlphaFold3, we propose the unified model UniIF for the inverse folding of all molecules. We do such unification in two levels: 1) Data-Level: We propose a unified block graph data form for all molecules, including the local frame building and geometric feature initialization. 2) Model-Level: We introduce a geometric block attention network, comprising a geometric interaction, interactive attention and virtual long-term dependency modules, to capture the 3D interactions of all molecules. Through comprehensive evaluations across various tasks such as protein design, RNA design, and material design, we demonstrate that our proposed method surpasses state-of-the-art methods on all tasks. UniIF offers a versatile and effective solution for general molecule inverse folding.
Authors: Ziheng Cheng, Longlin Yu, Tianyu Xie, Shiyue Zhang, Cheng Zhang
Abstract: Semi-implicit variational inference (SIVI) extends traditional variational families with semi-implicit distributions defined in a hierarchical manner. Due to the intractable densities of semi-implicit distributions, classical SIVI often resorts to surrogates of evidence lower bound (ELBO) that would introduce biases for training. A recent advancement in SIVI, named SIVI-SM, utilizes an alternative score matching objective made tractable via a minimax formulation, albeit requiring an additional lower-level optimization. In this paper, we propose kernel SIVI (KSIVI), a variant of SIVI-SM that eliminates the need for lower-level optimization through kernel tricks. Specifically, we show that when optimizing over a reproducing kernel Hilbert space (RKHS), the lower-level problem has an explicit solution. This way, the upper-level objective becomes the kernel Stein discrepancy (KSD), which is readily computable for stochastic gradient descent due to the hierarchical structure of semi-implicit variational distributions. An upper bound for the variance of the Monte Carlo gradient estimators of the KSD objective is derived, which allows us to establish novel convergence guarantees of KSIVI. We demonstrate the effectiveness and efficiency of KSIVI on both synthetic distributions and a variety of real data Bayesian inference tasks.
Authors: Xiaotong Cheng, Ioannis Tsetis, Setareh Maghsudi
Abstract: Modern power systems integrate renewable distributed energy resources (DERs) as an environment-friendly enhancement to meet the ever-increasing demands. However, the inherent unreliability of renewable energy renders developing DER management algorithms imperative. We study the energy-sharing problem in a system consisting of several DERs. Each agent harvests and distributes renewable energy in its neighborhood to optimize the network's performance while minimizing energy waste. We model this problem as a bandit convex optimization problem with constraints that correspond to each node's limitations for energy production. We propose distributed decision-making policies to solve the formulated problem, where we utilize the notion of dynamic regret as the performance metric. We also include an adjustment strategy in our developed algorithm to reduce the constraint violations. Besides, we design a policy that deals with the non-stationary environment. Theoretical analysis shows the effectiveness of our proposed algorithm. Numerical experiments using a real-world dataset show superior performance of our proposal compared to state-of-the-art methods.
Authors: Matthaios Chatzopoulos, Phaedon-Stelios Koutsourelakis
Abstract: We propose Physics-Aware Neural Implicit Solvers (PANIS), a novel, data-driven framework for learning surrogates for parametrized Partial Differential Equations (PDEs). It consists of a probabilistic, learning objective in which weighted residuals are used to probe the PDE and provide a source of {\em virtual} data i.e. the actual PDE never needs to be solved. This is combined with a physics-aware implicit solver that consists of a much coarser, discretized version of the original PDE, which provides the requisite information bottleneck for high-dimensional problems and enables generalization in out-of-distribution settings (e.g. different boundary conditions). We demonstrate its capability in the context of random heterogeneous materials where the input parameters represent the material microstructure. We extend the framework to multiscale problems and show that a surrogate can be learned for the effective (homogenized) solution without ever solving the reference problem. We further demonstrate how the proposed framework can accommodate and generalize several existing learning objectives and architectures while yielding probabilistic surrogates that can quantify predictive uncertainty.
Authors: Bal\'azs Szalontai, Gerg\H{o} Szalay, Tam\'as M\'arton, Anna Sike, Bal\'azs Pint\'er, Tibor Gregorics
Abstract: Recently, there has been increasing activity in using deep learning for software engineering, including tasks like code generation and summarization. In particular, the most recent coding Large Language Models seem to perform well on these problems. In this technical report, we aim to review how these models perform in code explanation/summarization, while also investigating their code generation capabilities (based on natural language descriptions).
Authors: Naoki Nishikawa, Taiji Suzuki
Abstract: Deep neural networks based on state space models (SSMs) are attracting much attention in sequence modeling since their computational cost is significantly smaller than that of Transformers. While the capabilities of SSMs have been primarily investigated through experimental comparisons, theoretical understanding of SSMs is still limited. In particular, there is a lack of statistical and quantitative evaluation of whether SSM can replace Transformers. In this paper, we theoretically explore in which tasks SSMs can be alternatives of Transformers from the perspective of estimating sequence-to-sequence functions. We consider the setting where the target function has direction-dependent smoothness and prove that SSMs can estimate such functions with the same convergence rate as Transformers. Additionally, we prove that SSMs can estimate the target function, even if the smoothness changes depending on the input sequence, as well as Transformers. Our results show the possibility that SSMs can replace Transformers when estimating the functions in certain classes that appear in practice.
Authors: Zongbao Zhang, Jiao Hao, Wenmeng Zhao, Yan Liu, Yaohui Huang, Xinhang Luo
Abstract: The rapid expansion of electric vehicles (EVs) has rendered the load forecasting of electric vehicle charging stations (EVCS) increasingly critical. The primary challenge in achieving precise load forecasting for EVCS lies in accounting for the nonlinear of charging behaviors, the spatial interactions among different stations, and the intricate temporal variations in usage patterns. To address these challenges, we propose a Multiscale Spatio-Temporal Enhanced Model (MSTEM) for effective load forecasting at EVCS. MSTEM incorporates a multiscale graph neural network to discern hierarchical nonlinear temporal dependencies across various time scales. Besides, it also integrates a recurrent learning component and a residual fusion mechanism, enhancing its capability to accurately capture spatial and temporal variations in charging patterns. The effectiveness of the proposed MSTEM has been validated through comparative analysis with six baseline models using three evaluation metrics. The case studies utilize real-world datasets for both fast and slow charging loads at EVCS in Perth, UK. The experimental results demonstrate the superiority of MSTEM in short-term continuous load forecasting for EVCS.
Authors: Georg Rutishauser, Joan Mihali, Moritz Scherer, Luca Benini
Abstract: Ternary neural networks (TNNs) offer a superior accuracy-energy trade-off compared to binary neural networks. However, until now, they have required specialized accelerators to realize their efficiency potential, which has hindered widespread adoption. To address this, we present xTern, a lightweight extension of the RISC-V instruction set architecture (ISA) targeted at accelerating TNN inference on general-purpose cores. To complement the ISA extension, we developed a set of optimized kernels leveraging xTern, achieving 67% higher throughput than their 2-bit equivalents. Power consumption is only marginally increased by 5.2%, resulting in an energy efficiency improvement by 57.1%. We demonstrate that the proposed xTern extension, integrated into an octa-core compute cluster, incurs a minimal silicon area overhead of 0.9% with no impact on timing. In end-to-end benchmarks, we demonstrate that xTern enables the deployment of TNNs achieving up to 1.6 percentage points higher CIFAR-10 classification accuracy than 2-bit networks at equal inference latency. Our results show that xTern enables RISC-V-based ultra-low-power edge AI platforms to benefit from the efficiency potential of TNNs.
Authors: Markus J. Buehler
Abstract: We present Cephalo, a series of multimodal vision large language models (V-LLMs) designed for materials science applications, integrating visual and linguistic data for enhanced understanding and interaction within human-AI and multi-agent AI frameworks. A key innovation of Cephalo is its advanced dataset generation method, which employs a sophisticated algorithm to accurately detect and separate images and their corresponding textual descriptions from PDF documents, such as scientific papers. The method includes a careful refinement of image-text pairs through integrated vision and language processing, ensuring high-quality, contextually relevant, and well reasoned training data. Cephalo is trained on integrated image and text data extracted from thousands of scientific papers and science-focused Wikipedia pages demonstrates can interpret complex visual scenes, generate precise language descriptions, and answer queries about images effectively. The combination of a vision encoder with an autoregressive transformer supports complex natural language understanding in an integrated model, which can be coupled with other generative methods to create an image-to-text-to-image or image-to-text-to-3D pipeline. To explore the development of larger models from smaller ones, we merge sets of layers that originate from different pre-trained source models. This hybrid approach allows us to leverage the domain-specific expertise and general conversational capabilities to harness the strengths of multiple models. We examine the models in diverse use cases that incorporate biological materials, fracture and engineering analysis, protein biophysics, and bio-inspired design based on insect behavior. Generative applications include bio-inspired designs, including pollen-inspired architected materials, as well as the synthesis of bio-inspired material microstructures from a photograph of a solar eclipse.
Authors: Xinyue Shen, Yixin Wu, Michael Backes, Yang Zhang
Abstract: Recently, the concept of artificial assistants has evolved from science fiction into real-world applications. GPT-4o, the newest multimodal large language model (MLLM) across audio, vision, and text, has further blurred the line between fiction and reality by enabling more natural human-computer interactions. However, the advent of GPT-4o's voice mode may also introduce a new attack surface. In this paper, we present the first systematic measurement of jailbreak attacks against the voice mode of GPT-4o. We show that GPT-4o demonstrates good resistance to forbidden questions and text jailbreak prompts when directly transferring them to voice mode. This resistance is primarily due to GPT-4o's internal safeguards and the difficulty of adapting text jailbreak prompts to voice mode. Inspired by GPT-4o's human-like behaviors, we propose VoiceJailbreak, a novel voice jailbreak attack that humanizes GPT-4o and attempts to persuade it through fictional storytelling (setting, character, and plot). VoiceJailbreak is capable of generating simple, audible, yet effective jailbreak prompts, which significantly increases the average attack success rate (ASR) from 0.033 to 0.778 in six forbidden scenarios. We also conduct extensive experiments to explore the impacts of interaction steps, key elements of fictional writing, and different languages on VoiceJailbreak's effectiveness and further enhance the attack performance with advanced fictional writing techniques. We hope our study can assist the research community in building more secure and well-regulated MLLMs.
Authors: Jacopo Teneggi, Jeremias Sulam
Abstract: Recent works have extended notions of feature importance to \emph{semantic concepts} that are inherently interpretable to the users interacting with a black-box predictive model. Yet, precise statistical guarantees, such as false positive rate control, are needed to communicate findings transparently and to avoid unintended consequences in real-world scenarios. In this paper, we formalize the global (i.e., over a population) and local (i.e., for a sample) statistical importance of semantic concepts for the predictions of opaque models, by means of conditional independence, which allows for rigorous testing. We use recent ideas of sequential kernelized testing (SKIT) to induce a rank of importance across concepts, and showcase the effectiveness and flexibility of our framework on synthetic datasets as well as on image classification tasks using vision-language models such as CLIP.
Authors: Purba Mukherjee, Anjan Ananda Sen
Abstract: In this work, we implement Gaussian process regression to reconstruct the expansion history of the universe in a model-agnostic manner, using the Pantheon-Plus SN-Ia compilation in combination with two different BAO measurements (SDSS-IV and DESI DR1). In both the reconstructions, the $\Lambda$CDM model is always included in the 95\% confidence intervals. We find evidence that the DESI LRG data at $z_{\text{eff}} = 0.51$ is not an outlier within our model-independent framework. We study the $\mathcal{O}m$-diagnostics and the evolution of the total equation of state (EoS) of our universe, which hint towards the possibility of a quintessence-like dark energy scenario with a very slowly varying EoS, and a phantom-crossing in higher $z$. The entire exercise is later complemented by considering two more SN-Ia compilations - DES-5YR and Union3 - in combination with DESI BAO. Reconstruction with the DESI BAO + DES-5YR SN data sets predicts that the $\Lambda$CDM model lies outside the 3$\sigma$ confidence levels, whereas with DESI BAO + Union3 data, the $\Lambda$CDM model is always included within 1$\sigma$. We also report constraints on $H_0 r_d$ from our model-agnostic analysis, independent of the pre-recombination physics. Our results point towards an $\approx$ 2$\sigma$ discrepancy between the DESI + Pantheon-Plus and DESI + DES-5YR data sets, which calls for further investigation.
Authors: Laura Fieback (Volkswagen AG, TU Berlin), Jakob Spiegelberg (Volkswagen AG), Hanno Gottschalk (TU Berlin)
Abstract: Large Vision Language Models (LVLMs) have shown remarkable capabilities in multimodal tasks like visual question answering or image captioning. However, inconsistencies between the visual information and the generated text, a phenomenon referred to as hallucinations, remain an unsolved problem with regard to the trustworthiness of LVLMs. To address this problem, recent works proposed to incorporate computationally costly Large (Vision) Language Models in order to detect hallucinations on a sentence- or subsentence-level. In this work, we introduce MetaToken, a lightweight binary classifier to detect hallucinations on the token-level at negligible cost. Based on a statistical analysis, we reveal key factors of hallucinations in LVLMs which have been overseen in previous works. MetaToken can be applied to any open-source LVLM without any knowledge about ground truth data providing a reliable detection of hallucinations. We evaluate our method on four state-of-the-art LVLMs demonstrating the effectiveness of our approach.
Authors: Xuan Son Nguyen, Shuo Yang, Aymeric Histace
Abstract: Deep neural networks (DNNs) on Riemannian manifolds have garnered increasing interest in various applied areas. For instance, DNNs on spherical and hyperbolic manifolds have been designed to solve a wide range of computer vision and nature language processing tasks. One of the key factors that contribute to the success of these networks is that spherical and hyperbolic manifolds have the rich algebraic structures of gyrogroups and gyrovector spaces. This enables principled and effective generalizations of the most successful DNNs to these manifolds. Recently, some works have shown that many concepts in the theory of gyrogroups and gyrovector spaces can also be generalized to matrix manifolds such as Symmetric Positive Definite (SPD) and Grassmann manifolds. As a result, some building blocks for SPD and Grassmann neural networks, e.g., isometric models and multinomial logistic regression (MLR) can be derived in a way that is fully analogous to their spherical and hyperbolic counterparts. Building upon these works, we design fully-connected (FC) and convolutional layers for SPD neural networks. We also develop MLR on Symmetric Positive Semi-definite (SPSD) manifolds, and propose a method for performing backpropagation with the Grassmann logarithmic map in the projector perspective. We demonstrate the effectiveness of the proposed approach in the human action recognition and node classification tasks.
Authors: ChonLam Lao, Jiaqi Gao, Ganesh Ananthanarayanan, Aditya Akella, Minlan Yu
Abstract: The trend of modeless ML inference is increasingly growing in popularity as it hides the complexity of model inference from users and caters to diverse user and application accuracy requirements. Previous work mostly focuses on modeless inference in data centers. To provide low-latency inference, in this paper, we promote modeless inference at the edge. The edge environment introduces additional challenges related to low power consumption, limited device memory, and volatile network environments. To address these challenges, we propose HawkVision, which provides low-latency modeless serving of vision DNNs. HawkVision leverages a two-layer edge-DC architecture that employs confidence scaling to reduce the number of model options while meeting diverse accuracy requirements. It also supports lossy inference under volatile network environments. Our experimental results show that HawkVision outperforms current serving systems by up to 1.6X in P99 latency for providing modeless service. Our FPGA prototype demonstrates similar performance at certain accuracy levels with up to a 3.34X reduction in power consumption.
Authors: Yue Xia, Christoph Hofmeister, Maximilian Egger, Rawad Bitar
Abstract: Federated Learning (FL) faces several challenges, such as the privacy of the clients data and security against Byzantine clients. Existing works treating privacy and security jointly make sacrifices on the privacy guarantee. In this work, we introduce LoByITFL, the first communication-efficient Information-Theoretic (IT) private and secure FL scheme that makes no sacrifices on the privacy guarantees while ensuring security against Byzantine adversaries. The key ingredients are a small and representative dataset available to the federator, a careful transformation of the FLTrust algorithm and the use of a trusted third party only in a one-time preprocessing phase before the start of the learning algorithm. We provide theoretical guarantees on privacy and Byzantine-resilience, and provide convergence guarantee and experimental results validating our theoretical findings.
Authors: Seyedmehdi Orouji, Martin C. Liu, Tal Korem, Megan A. K. Peters
Abstract: Machine learning techniques are steadily becoming more important in modern biology, and are used to build predictive models, discover patterns, and investigate biological problems. However, models trained on one dataset are often not generalizable to other datasets from different cohorts or laboratories, due to differences in the statistical properties of these datasets. These could stem from technical differences, such as the measurement technique used, or from relevant biological differences between the populations studied. Domain adaptation, a type of transfer learning, can alleviate this problem by aligning the statistical distributions of features and samples among different datasets so that similar models can be applied across them. However, a majority of state-of-the-art domain adaptation methods are designed to work with large-scale data, mostly text and images, while biological datasets often suffer from small sample sizes, and possess complexities such as heterogeneity of the feature space. This Review aims to synthetically discuss domain adaptation methods in the context of small-scale and highly heterogeneous biological data. We describe the benefits and challenges of domain adaptation in biological research and critically discuss some of its objectives, strengths, and weaknesses through key representative methodologies. We argue for the incorporation of domain adaptation techniques to the computational biologist's toolkit, with further development of customized approaches.
Authors: Ed Davis, Ian Gallagher, Daniel John Lawson, Patrick Rubin-Delanchy
Abstract: Graph neural networks (GNNs) are powerful black-box models which have shown impressive empirical performance. However, without any form of uncertainty quantification, it can be difficult to trust such models in high-risk scenarios. Conformal prediction aims to address this problem, however, an assumption of exchangeability is required for its validity which has limited its applicability to static graphs and transductive regimes. We propose to use unfolding, which allows any existing static GNN to output a dynamic graph embedding with exchangeability properties. Using this, we extend the validity of conformal prediction to dynamic GNNs in both transductive and semi-inductive regimes. We provide a theoretical guarantee of valid conformal prediction in these cases and demonstrate the empirical validity, as well as the performance gains, of unfolded GNNs against standard GNN architectures on both simulated and real datasets.
Authors: Amir Hossein Karbasi, Hao Yang, Saiedeh Razavi
Abstract: In transportation networks, intersections pose significant risks of collisions due to conflicting movements of vehicles approaching from different directions. To address this issue, various tools can exert influence on traffic safety both directly and indirectly. This study focuses on investigating the impact of adaptive signal control and connected and automated vehicles (CAVs) on intersection safety using a deep reinforcement learning approach. The objective is to assess the individual and combined effects of CAVs and adaptive traffic signal control on traffic safety, considering rear-end and crossing conflicts. The study employs a Deep Q Network (DQN) to regulate traffic signals and driving behaviors of both CAVs and Human Drive Vehicles (HDVs), and uses Time To Collision (TTC) metric to evaluate safety. The findings demonstrate a significant reduction in rear-end and crossing conflicts through the combined implementation of CAVs and DQNs-based traffic signal control. Additionally, the long-term positive effects of CAVs on safety are similar to the short-term effects of combined CAVs and DQNs-based traffic signal control. Overall, the study emphasizes the potential benefits of integrating CAVs and adaptive traffic signal control approaches in order to enhance traffic safety. The findings of this study could provide valuable insights for city officials and transportation authorities in developing effective strategies to improve safety at signalized intersections.
Authors: Ruchika Chavhan, Da Li, Timothy Hospedales
Abstract: While large-scale text-to-image diffusion models have demonstrated impressive image-generation capabilities, there are significant concerns about their potential misuse for generating unsafe content, violating copyright, and perpetuating societal biases. Recently, the text-to-image generation community has begun addressing these concerns by editing or unlearning undesired concepts from pre-trained models. However, these methods often involve data-intensive and inefficient fine-tuning or utilize various forms of token remapping, rendering them susceptible to adversarial jailbreaks. In this paper, we present a simple and effective training-free approach, ConceptPrune, wherein we first identify critical regions within pre-trained models responsible for generating undesirable concepts, thereby facilitating straightforward concept unlearning via weight pruning. Experiments across a range of concepts including artistic styles, nudity, object erasure, and gender debiasing demonstrate that target concepts can be efficiently erased by pruning a tiny fraction, approximately 0.12% of total weights, enabling multi-concept erasure and robustness against various white-box and black-box adversarial attacks.
Authors: Harikrishna Narasimhan, Wittawat Jitkrittum, Ankit Singh Rawat, Seungyeon Kim, Neha Gupta, Aditya Krishna Menon, Sanjiv Kumar
Abstract: Cascades and speculative decoding are two common approaches to improving language models' inference efficiency. Both approaches involve interleaving models of different sizes, but via fundamentally distinct mechanisms: cascades employ a deferral rule that invokes the larger model only for "hard" inputs, while speculative decoding uses speculative execution to primarily invoke the larger model in parallel verification mode. These mechanisms offer different benefits: empirically, cascades are often capable of yielding better quality than even the larger model, while theoretically, speculative decoding offers a guarantee of quality-neutrality. In this paper, we leverage the best of both these approaches by designing new speculative cascading techniques that implement their deferral rule through speculative execution. We characterize the optimal deferral rule for our speculative cascades, and employ a plug-in approximation to the optimal rule. Through experiments with T5 models on benchmark language tasks, we show that the proposed approach yields better cost-quality trade-offs than cascading and speculative decoding baselines.
Authors: Zhanhui Zhou, Zhixuan Liu, Jie Liu, Zhichen Dong, Chao Yang, Yu Qiao
Abstract: Large language models are usually fine-tuned to align with human preferences. However, fine-tuning a large language model can be challenging. In this work, we introduce $\textit{weak-to-strong search}$, framing the alignment of a large language model as a test-time greedy search to maximize the log-likelihood difference between small tuned and untuned models while sampling from the frozen large model. This method serves both as (i) a compute-efficient model up-scaling strategy that avoids directly tuning the large model and as (ii) an instance of weak-to-strong generalization that enhances a strong model with weak test-time guidance. Empirically, we demonstrate the flexibility of weak-to-strong search across different tasks. In controlled-sentiment generation and summarization, we use tuned and untuned $\texttt{gpt2}$s to effectively improve the alignment of large models without additional training. Crucially, in a more difficult instruction-following benchmark, AlpacaEval 2.0, we show that reusing off-the-shelf small model pairs (e.g., $\texttt{zephyr-7b-beta}$ and its untuned version) can significantly improve the length-controlled win rates of both white-box and black-box large models against $\texttt{gpt-4-turbo}$ (e.g., $34.4 \rightarrow 37.9$ for $\texttt{Llama-3-70B-Instruct}$ and $16.0 \rightarrow 20.1$ for $\texttt{gpt-3.5-turbo-instruct}$), despite the small models' low win rates $\approx 10.0$.
Authors: Xiang Fu, Andrew Rosen, Kyle Bystrom, Rui Wang, Albert Musaelian, Boris Kozinsky, Tess Smidt, Tommi Jaakkola
Abstract: In density functional theory, charge density is the core attribute of atomic systems from which all chemical properties can be derived. Machine learning methods are promising in significantly accelerating charge density prediction, yet existing approaches either lack accuracy or scalability. We propose a recipe that can achieve both. In particular, we identify three key ingredients: (1) representing the charge density with atomic and virtual orbitals (spherical fields centered at atom/virtual coordinates); (2) using expressive and learnable orbital basis sets (basis function for the spherical fields); and (3) using high-capacity equivariant neural network architecture. Our method achieves state-of-the-art accuracy while being more than an order of magnitude faster than existing methods. Furthermore, our method enables flexible efficiency-accuracy trade-offs by adjusting the model/basis sizes.
Authors: Thomas W. Mitchel, Michael Taylor, Vincent Sitzmann
Abstract: Real-world geometry and 3D vision tasks are replete with challenging symmetries that defy tractable analytical expression. In this paper, we introduce Neural Isometries, an autoencoder framework which learns to map the observation space to a general-purpose latent space wherein encodings are related by isometries whenever their corresponding observations are geometrically related in world space. Specifically, we regularize the latent space such that maps between encodings preserve a learned inner product and commute with a learned functional operator, in the same manner as rigid-body transformations commute with the Laplacian. This approach forms an effective backbone for self-supervised representation learning, and we demonstrate that a simple off-the-shelf equivariant network operating in the pre-trained latent space can achieve results on par with meticulously-engineered, handcrafted networks designed to handle complex, nonlinear symmetries. Furthermore, isometric maps capture information about the respective transformations in world space, and we show that this allows us to regress camera poses directly from the coefficients of the maps between encodings of adjacent views of a scene.
Authors: Wenbo Hu, Zi-Yi Dou, Liunian Harold Li, Amita Kamath, Nanyun Peng, Kai-Wei Chang
Abstract: Large Vision-Language Models (LVLMs) typically encode an image into a fixed number of visual tokens (e.g., 576) and process these tokens with a language model. Despite their strong performance, LVLMs face challenges in adapting to varying computational constraints. This raises the question: can we achieve flexibility in the number of visual tokens to suit different tasks and computational resources? We answer this with an emphatic yes. Inspired by Matryoshka Representation Learning, we introduce the Matryoshka Query Transformer (MQT), capable of encoding an image into m visual tokens during inference, where m can be any number up to a predefined maximum. This is achieved by employing a query transformer with M latent query tokens to compress the visual embeddings. During each training step, we randomly select m <= M latent query tokens and train the model using only these first m tokens, discarding the rest. Combining MQT with LLaVA, we train a single model once, and flexibly and drastically reduce the number of inference-time visual tokens while maintaining similar or better performance compared to training independent models for each number of tokens. Our model, MQT-LLAVA, matches LLaVA-1.5 performance across 11 benchmarks using a maximum of 256 tokens instead of LLaVA's fixed 576. Reducing to 16 tokens (8x less TFLOPs) only sacrifices the performance by 2.4 points on MMBench. On certain tasks such as ScienceQA and MMMU, we can even go down to only 2 visual tokens with performance drops of just 3% and 6% each. Our exploration of the trade-off between the accuracy and computational cost brought about by the number of visual tokens facilitates future research to achieve the best of both worlds.
Authors: Mingmeng Geng, Sihong He, Roberto Trotta
Abstract: Do large language models (LLMs) have their own worldviews and personality tendencies? Simulations in which an LLM was asked to answer subjective questions were conducted more than 1 million times. Comparison of the responses from different LLMs with real data from the European Social Survey (ESS) suggests that the effect of prompts on bias and variability is fundamental, highlighting major cultural, age, and gender biases. Methods for measuring the difference between LLMs and survey data are discussed, such as calculating weighted means and a new proposed measure inspired by Jaccard similarity. We conclude that it is important to analyze the robustness and variability of prompts before using LLMs to model individual decisions or collective behavior, as their imitation abilities are approximate at best.
Authors: Ge Zhang, Scott Qu, Jiaheng Liu, Chenchen Zhang, Chenghua Lin, Chou Leuang Yu, Danny Pan, Esther Cheng, Jie Liu, Qunshu Lin, Raven Yuan, Tuney Zheng, Wei Pang, Xinrun Du, Yiming Liang, Yinghao Ma, Yizhi Li, Ziyang Ma, Bill Lin, Emmanouil Benetos, Huan Yang, Junting Zhou, Kaijing Ma, Minghao Liu, Morry Niu, Noah Wang, Quehry Que, Ruibo Liu, Sine Liu, Shawn Guo, Soren Gao, Wangchunshu Zhou, Xinyue Zhang, Yizhi Zhou, Yubo Wang, Yuelin Bai, Yuhan Zhang, Yuxiang Zhang, Zenith Wang, Zhenzhu Yang, Zijian Zhao, Jiajun Zhang, Wanli Ouyang, Wenhao Huang, Wenhu Chen
Abstract: Large Language Models (LLMs) have made great strides in recent years to achieve unprecedented performance across different tasks. However, due to commercial interest, the most competitive models like GPT, Gemini, and Claude have been gated behind proprietary interfaces without disclosing the training details. Recently, many institutions have open-sourced several strong LLMs like LLaMA-3, comparable to existing closed-source LLMs. However, only the model's weights are provided with most details (e.g., intermediate checkpoints, pre-training corpus, and training code, etc.) being undisclosed. To improve the transparency of LLMs, the research community has formed to open-source truly open LLMs (e.g., Pythia, Amber, OLMo), where more details (e.g., pre-training corpus and training code) are being provided. These models have greatly advanced the scientific study of these large models including their strengths, weaknesses, biases and risks. However, we observe that the existing truly open LLMs on reasoning, knowledge, and coding tasks are still inferior to existing state-of-the-art LLMs with similar model sizes. To this end, we open-source MAP-Neo, a highly capable and transparent bilingual language model with 7B parameters trained from scratch on 4.5T high-quality tokens. Our MAP-Neo is the first fully open-sourced bilingual LLM with comparable performance compared to existing state-of-the-art LLMs. Moreover, we open-source all details to reproduce our MAP-Neo, where the cleaned pre-training corpus, data cleaning pipeline, checkpoints, and well-optimized training/evaluation framework are provided. Finally, we hope our MAP-Neo will enhance and strengthen the open research community and inspire more innovations and creativities to facilitate the further improvements of LLMs.
Authors: Hanrong Ye, De-An Huang, Yao Lu, Zhiding Yu, Wei Ping, Andrew Tao, Jan Kautz, Song Han, Dan Xu, Pavlo Molchanov, Hongxu Yin
Abstract: We introduce X-VILA, an omni-modality model designed to extend the capabilities of large language models (LLMs) by incorporating image, video, and audio modalities. By aligning modality-specific encoders with LLM inputs and diffusion decoders with LLM outputs, X-VILA achieves cross-modality understanding, reasoning, and generation. To facilitate this cross-modality alignment, we curate an effective interleaved any-to-any modality instruction-following dataset. Furthermore, we identify a significant problem with the current cross-modality alignment method, which results in visual information loss. To address the issue, we propose a visual alignment mechanism with a visual embedding highway module. We then introduce a resource-efficient recipe for training X-VILA, that exhibits proficiency in any-to-any modality conversation, surpassing previous approaches by large margins. X-VILA also showcases emergent properties across modalities even in the absence of similar training data. The project will be made open-source.
Authors: Lijun Zhang, Tianbao Yang, Rong Jin, Zhi-Hua Zhou
Abstract: In this paper, we develop a relative error bound for nuclear norm regularized matrix completion, with the focus on the completion of full-rank matrices. Under the assumption that the top eigenspaces of the target matrix are incoherent, we derive a relative upper bound for recovering the best low-rank approximation of the unknown matrix. Although multiple works have been devoted to analyzing the recovery error of full-rank matrix completion, their error bounds are usually additive, making it impossible to obtain the perfect recovery case and more generally difficult to leverage the skewed distribution of eigenvalues. Our analysis is built upon the optimality condition of the regularized formulation and existing guarantees for low-rank matrix completion. To the best of our knowledge, this is the first relative bound that has been proved for the regularized formulation of matrix completion.
Authors: Amine M. Remita, Abdoulaye Banir\'e Diallo
Abstract: Viral sequence classification is an important task in pathogen detection, epidemiological surveys and evolutionary studies. Statistical learning methods are widely used to classify and identify viral sequences in samples from environments. These methods face several challenges associated with the nature and properties of viral genomes such as recombination, mutation rate and diversity. Also, new generations of sequencing technologies rise other difficulties by generating massive amounts of fragmented sequences. While linear classifiers are often used to classify viruses, there is a lack of exploration of the accuracy space of existing models in the context of alignment free approaches. In this study, we present an exhaustive assessment procedure exploring the power of linear classifiers in genotyping and subtyping partial and complete genomes. It is applied to the Hepatitis C viruses (HCV). Several variables are considered in this investigation such as classifier types (generative and discriminative) and their hyper-parameters (smoothing value and regularization penalty function), the classification task (genotyping and subtyping), the length of the tested sequences (partial and complete) and the length of k-mer words. Overall, several classifiers perform well given a set of precise combination of the experimental variables mentioned above. Finally, we provide the procedure and benchmark data to allow for more robust assessment of classification from virus genomes.
Authors: Meghna Singh, Saksham Goel, Abhiraj Mohan, Jaideep Srivastava
Abstract: The quality of sleep has a deep impact on people's physical and mental health. People with insufficient sleep are more likely to report physical and mental distress, activity limitation, anxiety, and pain. Moreover, in the past few years, there has been an explosion of applications and devices for activity monitoring and health tracking. Signals collected from these wearable devices can be used to study and improve sleep quality. In this paper, we utilize the relationship between physical activity and sleep quality to find ways of assisting people improve their sleep using machine learning techniques. People usually have several behavior modes that their bio-functions can be divided into. Performing time series clustering on activity data, we find cluster centers that would correlate to the most evident behavior modes for a specific subject. Activity recipes are then generated for good sleep quality for each behavior mode within each cluster. These activity recipes are supplied to an activity recommendation engine for suggesting a mix of relaxed to intense activities to subjects during their daily routines. The recommendations are further personalized based on the subjects' lifestyle constraints, i.e. their age, gender, body mass index (BMI), resting heart rate, etc, with the objective of the recommendation being the improvement of that night's quality of sleep. This would in turn serve a longer-term health objective, like lowering heart rate, improving the overall quality of sleep, etc.
Authors: Yuantong Li, Chi-hua Wang, Guang Cheng, Will Wei Sun
Abstract: Two-sided online matching platforms are employed in various markets. However, agents' preferences in the current market are usually implicit and unknown, thus needing to be learned from data. With the growing availability of dynamic side information involved in the decision process, modern online matching methodology demands the capability to track shifting preferences for agents based on contextual information. This motivates us to propose a novel framework for this dynamic online matching problem with contextual information, which allows for dynamic preferences in matching decisions. Existing works focus on online matching with static preferences, but this is insufficient: the two-sided preference changes as soon as one side's contextual information updates, resulting in non-static matching. In this paper, we propose a dynamic matching bandit algorithm to adapt to this problem. The key component of the proposed dynamic matching algorithm is an online estimation of the preference ranking with a statistical guarantee. Theoretically, we show that the proposed dynamic matching algorithm delivers an agent-optimal stable matching result with high probability. In particular, we prove a logarithmic regret upper bound $\mathcal{O}(\log(T))$ and construct a corresponding instance-dependent matching regret lower bound. In the experiments, we demonstrate that dynamic matching algorithm is robust to various preference schemes, dimensions of contexts, reward noise levels, and context variation levels, and its application to a job-seeking market further demonstrates the practical usage of the proposed method.
Authors: Orin Levy, Asaf Cassel, Alon Cohen, Yishay Mansour
Abstract: We present the E-UC$^3$RL algorithm for regret minimization in Stochastic Contextual Markov Decision Processes (CMDPs). The algorithm operates under the minimal assumptions of realizable function class and access to \emph{offline} least squares and log loss regression oracles. Our algorithm is efficient (assuming efficient offline regression oracles) and enjoys a regret guarantee of $ \widetilde{O}(H^3 \sqrt{T |S| |A|d_{\mathrm{E}}(\mathcal{P}) \log (|\mathcal{F}| |\mathcal{P}|/ \delta) )}) , $ with $T$ being the number of episodes, $S$ the state space, $A$ the action space, $H$ the horizon, $\mathcal{P}$ and $\mathcal{F}$ are finite function classes used to approximate the context-dependent dynamics and rewards, respectively, and $d_{\mathrm{E}}(\mathcal{P})$ is the Eluder dimension of $\mathcal{P}$ w.r.t the Hellinger distance. To the best of our knowledge, our algorithm is the first efficient and rate-optimal regret minimization algorithm for CMDPs that operates under the general offline function approximation setting. In addition, we extend the Eluder dimension to general bounded metrics which may be of separate interest.
Authors: Travis Dick, Alex Kulesza, Ziteng Sun, Ananda Theertha Suresh
Abstract: We propose a new definition of instance optimality for differentially private estimation algorithms. Our definition requires an optimal algorithm to compete, simultaneously for every dataset $D$, with the best private benchmark algorithm that (a) knows $D$ in advance and (b) is evaluated by its worst-case performance on large subsets of $D$. That is, the benchmark algorithm need not perform well when potentially extreme points are added to $D$; it only has to handle the removal of a small number of real data points that already exist. This makes our benchmark significantly stronger than those proposed in prior work. We nevertheless show, for real-valued datasets, how to construct private algorithms that achieve our notion of instance optimality when estimating a broad class of dataset properties, including means, quantiles, and $\ell_p$-norm minimizers. For means in particular, we provide a detailed analysis and show that our algorithm simultaneously matches or exceeds the asymptotic performance of existing algorithms under a range of distributional assumptions.
Authors: Tao Li, Xiwei Zhang
Abstract: We establish a framework of distributed random inverse problems over network graphs with online measurements, and propose a decentralized online learning algorithm. This unifies the distributed parameter estimation in Hilbert spaces and the least mean square problem in reproducing kernel Hilbert spaces (RKHS-LMS). We transform the convergence of the algorithm into the asymptotic stability of a class of inhomogeneous random difference equations in Hilbert spaces with L2-bounded martingale difference terms and develop the L2 -asymptotic stability theory in Hilbert spaces. It is shown that if the network graph is connected and the sequence of forward operators satisfies the infinite-dimensional spatio-temporal persistence of excitation condition, then the estimates of all nodes are mean square and almost surely strongly consistent. Moreover, we propose a decentralized online learning algorithm in RKHS based on non-stationary and non-independent online data streams, and prove that the algorithm is mean square and almost surely strongly consistent if the operators induced by the random input data satisfy the infinite-dimensional spatio-temporal persistence of excitation condition.
Authors: Rebecca Salles, Janio Lima, Rafaelli Coutinho, Esther Pacitti, Florent Masseglia, Reza Akbarinia, Chao Chen, Jonathan Garibaldi, Fabio Porto, Eduardo Ogasawara
Abstract: Time series event detection methods are evaluated mainly by standard classification metrics that focus solely on detection accuracy. However, inaccuracy in detecting an event can often result from its preceding or delayed effects reflected in neighboring detections. These detections are valuable to trigger necessary actions or help mitigate unwelcome consequences. In this context, current metrics are insufficient and inadequate for the context of event detection. There is a demand for metrics that incorporate both the concept of time and temporal tolerance for neighboring detections. This paper introduces SoftED metrics, a new set of metrics designed for soft evaluating event detection methods. They enable the evaluation of both detection accuracy and the degree to which their detections represent events. They improved event detection evaluation by associating events and their representative detections, incorporating temporal tolerance in over 36\% of experiments compared to the usual classification metrics. SoftED metrics were validated by domain specialists that indicated their contribution to detection evaluation and method selection.
Authors: Emil Pitz, Kishore Pochiraju
Abstract: Heterogeneity and uncertainty in a composite microstructure lead to either computational bottlenecks if modeled rigorously or to solution inaccuracies in the stress field and failure predictions if approximated. Although methods suitable for analyzing arbitrary and non-linear microstructures exist, their computational cost makes them impractical to use in large-scale structural analysis. Surrogate models or Reduced Order Models (ROMs) commonly enhance efficiencies but are typically calibrated with a single microstructure. Homogenization methods, such as the Mori-Tanaka method, offer rapid homogenization for a wide range of constituent properties. However, simplifying assumptions, like stress and strain averaging in phases, render the consideration of both deterministic and stochastic variations in microstructure infeasible. This paper illustrates a transformer neural network architecture that captures the knowledge of various microstructures and constituents, enabling it to function as a computationally efficient homogenization surrogate model. Given an image or an abstraction of an arbitrary composite microstructure of linearly elastic fibers in an elastoplastic matrix, the transformer network predicts the history-dependent, non-linear, and homogenized stress-strain response. Two methods for encoding microstructure features were tested: calculating two-point statistics using Principal Component Analysis (PCA) for dimensionality reduction and employing an autoencoder with a Convolutional Neural Network (CNN). Both methods accurately predict the homogenized material response. The developed transformer neural network offers an efficient means for microstructure-to-property translation, generalizable and extendable to a variety of microstructures. The paper describes the network architecture, training and testing data generation, and performance under cycling and random loadings.
Authors: Francisco Eiras, Adel Bibi, Rudy Bunel, Krishnamurthy Dj Dvijotham, Philip Torr, M. Pawan Kumar
Abstract: Recent work provides promising evidence that Physics-Informed Neural Networks (PINN) can efficiently solve partial differential equations (PDE). However, previous works have failed to provide guarantees on the worst-case residual error of a PINN across the spatio-temporal domain - a measure akin to the tolerance of numerical solvers - focusing instead on point-wise comparisons between their solution and the ones obtained by a solver on a set of inputs. In real-world applications, one cannot consider tests on a finite set of points to be sufficient grounds for deployment, as the performance could be substantially worse on a different set. To alleviate this issue, we establish guaranteed error-based conditions for PINNs over their continuous applicability domain. To verify the extent to which they hold, we introduce $\partial$-CROWN: a general, efficient and scalable post-training framework to bound PINN residual errors. We demonstrate its effectiveness in obtaining tight certificates by applying it to two classically studied PINNs - Burgers' and Schr\"odinger's equations -, and two more challenging ones with real-world applications - the Allan-Cahn and Diffusion-Sorption equations.
Authors: Jikun Kang, Romain Laroche, Xingdi Yuan, Adam Trischler, Xue Liu, Jie Fu
Abstract: Decision Transformer-based decision-making agents have shown the ability to generalize across multiple tasks. However, their performance relies on massive data and computation. We argue that this inefficiency stems from the forgetting phenomenon, in which a model memorizes its behaviors in parameters throughout training. As a result, training on a new task may deteriorate the model's performance on previous tasks. In contrast to LLMs' implicit memory mechanism, the human brain utilizes distributed memory storage, which helps manage and organize multiple skills efficiently, mitigating the forgetting phenomenon. Inspired by this, we propose a working memory module to store, blend, and retrieve information for different downstream tasks. Evaluation results show that the proposed method improves training efficiency and generalization in Atari games and Meta-World object manipulation tasks. Moreover, we demonstrate that memory fine-tuning further enhances the adaptability of the proposed architecture.
Authors: Feng Chen, Daniel Kunin, Atsushi Yamamura, Surya Ganguli
Abstract: In this work, we reveal a strong implicit bias of stochastic gradient descent (SGD) that drives overly expressive networks to much simpler subnetworks, thereby dramatically reducing the number of independent parameters, and improving generalization. To reveal this bias, we identify invariant sets, or subsets of parameter space that remain unmodified by SGD. We focus on two classes of invariant sets that correspond to simpler (sparse or low-rank) subnetworks and commonly appear in modern architectures. Our analysis uncovers that SGD exhibits a property of stochastic attractivity towards these simpler invariant sets. We establish a sufficient condition for stochastic attractivity based on a competition between the loss landscape's curvature around the invariant set and the noise introduced by stochastic gradients. Remarkably, we find that an increased level of noise strengthens attractivity, leading to the emergence of attractive invariant sets associated with saddle-points or local maxima of the train loss. We observe empirically the existence of attractive invariant sets in trained deep neural networks, implying that SGD dynamics often collapses to simple subnetworks with either vanishing or redundant neurons. We further demonstrate how this simplifying process of stochastic collapse benefits generalization in a linear teacher-student framework. Finally, through this analysis, we mechanistically explain why early training with large learning rates for extended periods benefits subsequent generalization.
Authors: Gerry Windiarto Mohamad Dunda, Shenghui Song
Abstract: Federated learning (FL) has garnered considerable attention due to its privacy-preserving feature. Nonetheless, the lack of freedom in managing user data can lead to group fairness issues, where models are biased towards sensitive factors such as race or gender. To tackle this issue, this paper proposes a novel algorithm, fair federated averaging with augmented Lagrangian method (FFALM), designed explicitly to address group fairness issues in FL. Specifically, we impose a fairness constraint on the training objective and solve the minimax reformulation of the constrained optimization problem. Then, we derive the theoretical upper bound for the convergence rate of FFALM. The effectiveness of FFALM in improving fairness is shown empirically on CelebA and UTKFace datasets in the presence of severe statistical heterogeneity.
Authors: Jonathan Wenger, Felix Dangel, Agustinus Kristiadi
Abstract: The neural tangent kernel (NTK) has garnered significant attention as a theoretical framework for describing the behavior of large-scale neural networks. Kernel methods are theoretically well-understood and as a result enjoy algorithmic benefits, which can be demonstrated to hold in wide synthetic neural network architectures. These advantages include faster optimization, reliable uncertainty quantification and improved continual learning. However, current results quantifying the rate of convergence to the kernel regime suggest that exploiting these benefits requires architectures that are orders of magnitude wider than they are deep. This assumption raises concerns that architectures used in practice do not exhibit behaviors as predicted by the NTK. Here, we supplement previous work on the NTK by empirically investigating whether the limiting regime predicts practically relevant behavior of large-width architectures. Our results demonstrate that this is not the case across multiple domains. This observed disconnect between theory and practice further calls into question to what degree NTK theory should inform architectural and algorithmic choices.
Authors: Jiaxu Liu, Xinping Yi, Xiaowei Huang
Abstract: Hyperbolic graph convolutional networks (HGCN) have demonstrated significant potential in extracting information from hierarchical graphs. However, existing HGCNs are limited to shallow architectures, due to the expensive hyperbolic operations and the over-smoothing issue as depth increases. Although in GCNs, treatments have been applied to alleviate over-smoothing, developing a hyperbolic therapy presents distinct challenges since operations should be carefully designed to fit the hyperbolic nature. Addressing the above challenges, in this work, we propose DeepHGCN, the first deep multi-layer HGCN architecture with dramatically improved computational efficiency and substantially alleviated over-smoothing effect. DeepHGCN presents two key enablers of deep HGCNs: (1) a novel hyperbolic feature transformation layer that enables fast and accurate linear maps; and (2) techniques such as hyperbolic residual connections and regularization for both weights and features facilitated by an efficient hyperbolic midpoint method. Extensive experiments demonstrate that DeepHGCN obtains significant improvements in link prediction and node classification tasks compared to both Euclidean and shallow hyperbolic GCN variants.
Authors: Fangshuo Liao, Junhyung Lyle Kim, Cruz Barnum, Anastasios Kyrillidis
Abstract: Principal Component Analysis (PCA) aims to find subspaces spanned by the so-called principal components that best represent the variance in the dataset. The deflation method is a popular meta-algorithm that sequentially finds individual principal components, starting from the most important ones and working towards the less important ones. However, as deflation proceeds, numerical errors from the imprecise estimation of principal components propagate due to its sequential nature. This paper mathematically characterizes the error propagation of the inexact Hotelling's deflation method. We consider two scenarios: $i)$ when the sub-routine for finding the leading eigenvector is abstract and can represent various algorithms; and $ii)$ when power iteration is used as the sub-routine. In the latter case, the additional directional information from power iteration allows us to obtain a tighter error bound than the sub-routine agnostic case. For both scenarios, we explicitly characterize how the errors progress and affect subsequent principal component estimations.
Authors: Chenghua Gong, Xiang Li, Jianxiang Yu, Cheng Yao, Jiaqi Tan, Chengcheng Yu
Abstract: Graphs have become an important modeling tool for Web applications, and graph neural networks (GNNs) have achieved great success in graph representation learning. However, their performance heavily relies on a large amount of supervision. Recently, ``pre-train, fine-tune'' has become the paradigm to address the issues of label dependency and poor generalization. However, the pre-training strategies vary for graphs with homophily and heterophily, and the objectives for various downstream tasks also differ. This leads to a gap between pretexts and downstream tasks, resulting in ``negative transfer'' and poor performance. Inspired by prompt learning in natural language processing, many studies turn to bridge the gap and fully leverage the pre-trained model. However, existing methods for graph prompting are tailored to homophily, neglecting inherent heterophily on graphs. Meanwhile, most of them rely on randomly initialized prompts, which negatively impact on the stability. Therefore, we propose Self-Prompt, a prompting framework for graphs based on the model and data itself. We first introduce asymmetric graph contrastive learning as pretext to address heterophily and align the objectives of pretext and downstream tasks. Then we reuse the component from pre-training as the self adapter and introduce self-prompts based on graph itself for task adaptation. Finally, we conduct extensive experiments on 11 benchmark datasets to demonstrate its superiority. We provide our codes at \url{https://github.com/gongchenghua/Self-Pro}.
Authors: Julien Pourcel, C\'edric Colas, Gaia Molinaro, Pierre-Yves Oudeyer, Laetitia Teodorescu
Abstract: The ability to invent novel and interesting problems is a remarkable feature of human intelligence that drives innovation, art, and science. We propose a method that aims to automate this process by harnessing the power of state-of-the-art generative models to produce a diversity of challenging yet solvable problems, here in the context of Python programming puzzles. Inspired by the intrinsically motivated literature, Autotelic CodE Search (ACES) jointly optimizes for the diversity and difficulty of generated problems. We represent problems in a space of LLM-generated semantic descriptors describing the programming skills required to solve them (e.g. string manipulation, dynamic programming, etc.) and measure their difficulty empirically as a linearly decreasing function of the success rate of Llama-3-70B, a state-of-the-art LLM problem solver. ACES iteratively prompts a large language model to generate difficult problems achieving a diversity of target semantic descriptors (goal-directed exploration) using previously generated problems as in-context examples. ACES generates problems that are more diverse and more challenging than problems produced by baseline methods and three times more challenging than problems found in existing Python programming benchmarks on average across 11 state-of-the-art code LLMs.
Authors: Prashansa Panda, Shalabh Bhatnagar
Abstract: Actor Critic methods have found immense applications on a wide range of Reinforcement Learning tasks especially when the state-action space is large. In this paper, we consider actor critic and natural actor critic algorithms with function approximation for constrained Markov decision processes (C-MDP) involving inequality constraints and carry out a non-asymptotic analysis for both of these algorithms in a non-i.i.d (Markovian) setting. We consider the long-run average cost criterion where both the objective and the constraint functions are suitable policy-dependent long-run averages of certain prescribed cost functions. We handle the inequality constraints using the Lagrange multiplier method. We prove that these algorithms are guaranteed to find a first-order stationary point (i.e., $\Vert \nabla L(\theta,\gamma)\Vert_2^2 \leq \epsilon$) of the performance (Lagrange) function $L(\theta,\gamma)$, with a sample complexity of $\mathcal{\tilde{O}}(\epsilon^{-2.5})$ in the case of both Constrained Actor Critic (C-AC) and Constrained Natural Actor Critic (C-NAC) algorithms. We also show the results of experiments on three different Safety-Gym environments.
Authors: Zeyuan Zhao, Qingqing Ge, Anfeng Cheng, Yiding Liu, Xiang Li, Shuaiqiang Wang
Abstract: Heterogeneous graph neural networks(HGNNs) have recently shown impressive capability in modeling heterogeneous graphs that are ubiquitous in real-world applications. Most existing methods for heterogeneous graphs mainly learn node embeddings by stacking multiple convolutional or attentional layers, which can be considered as capturing the high-order information from node-level aspect. However, different types of nodes in heterogeneous graphs have diverse features, it is also necessary to capture interactions among node features, namely the high-order information from feature-level aspect. In addition, most methods first align node features by mapping them into one same low-dimensional space, while they may lose some type information of nodes in this way. To address these problems, in this paper, we propose a novel Heterogeneous graph Cascade Attention Network (HetCAN) composed of multiple cascade blocks. Each cascade block includes two components, the type-aware encoder and the dimension-aware encoder. Specifically, the type-aware encoder compensates for the loss of node type information and aims to make full use of graph heterogeneity. The dimension-aware encoder is able to learn the feature-level high-order information by capturing the interactions among node features. With the assistance of these components, HetCAN can comprehensively encode information of node features, graph heterogeneity and graph structure in node embeddings. Extensive experiments demonstrate the superiority of HetCAN over advanced competitors and also exhibit its efficiency and robustness.
Authors: Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik R Narasimhan, Ameet Deshpande
Abstract: The advent of large language models (LLMs) has ushered in a new paradigm of search engines that use generative models to gather and summarize information to answer user queries. This emerging technology, which we formalize under the unified framework of generative engines (GEs), can generate accurate and personalized responses, rapidly replacing traditional search engines like Google and Bing. Generative Engines typically satisfy queries by synthesizing information from multiple sources and summarizing them using LLMs. While this shift significantly improves \textit{user} utility and \textit{generative search engine} traffic, it poses a huge challenge for the third stakeholder - website and content creators. Given the black-box and fast-moving nature of generative engines, content creators have little to no control over \textit{when} and \textit{how} their content is displayed. With generative engines here to stay, we must ensure the creator economy is not disadvantaged. To address this, we introduce Generative Engine Optimization (GEO), the first novel paradigm to aid content creators in improving their content visibility in GE responses through a flexible black-box optimization framework for optimizing and defining visibility metrics. We facilitate systematic evaluation by introducing GEO-bench, a large-scale benchmark of diverse user queries across multiple domains, along with relevant web sources to answer these queries. Through rigorous evaluation, we demonstrate that GEO can boost visibility by up to 40\% in GE responses. Moreover, we show the efficacy of these strategies varies across domains, underscoring the need for domain-specific optimization methods. Our work opens a new frontier in information discovery systems, with profound implications for both developers of GEs and content creators.
Authors: Junping Hong, Ercan Engin Kuruoglu
Abstract: This paper is a preliminary study of the robustness and noise analysis of deep neural networks via a game theory formulation Bayesian Neural Networks (BNN) and the maximal coding rate distortion loss. BNN has been shown to provide some robustness to deep learning, and the minimax method used to be a natural conservative way to assist the Bayesian method. Inspired by the recent closed-loop transcription neural network, we formulate the BNN via game theory between the deterministic neural network $f$ and the sampling network $f + \xi$ or $f + r*\xi$. Compared with previous BNN, BNN via game theory learns a solution space within a certain gap between the center $f$ and the sampling point $f + r*\xi$, and is a conservative choice with a meaningful prior setting compared with previous BNN. Furthermore, the minimum points between $f$ and $f + r*\xi$ become stable when the subspace dimension is large enough with a well-trained model $f$. With these, the model $f$ can have a high chance of recognizing the out-of-distribution data or noise data in the subspace rather than the prediction level, even if $f$ is in online training after a few iterations of true data. So far, our experiments are limited to MNIST and Fashion MNIST data sets, more experiments with realistic data sets and complicated neural network models should be implemented to validate the above arguments.
Authors: Prin Phunyaphibarn, Junghyun Lee, Bohan Wang, Huishuai Zhang, Chulhee Yun
Abstract: Although gradient descent with Polyak's momentum is widely used in modern machine and deep learning, a concrete understanding of its effects on the training trajectory remains elusive. In this work, we empirically show that for linear diagonal networks and nonlinear neural networks, momentum gradient descent with a large learning rate displays large catapults, driving the iterates towards much flatter minima than those found by gradient descent. We hypothesize that the large catapult is caused by momentum "prolonging" the self-stabilization effect (Damian et al., 2023). We provide theoretical and empirical support for our hypothesis in a simple toy example and empirical evidence supporting our hypothesis for linear diagonal networks.
Authors: Pengchao Han, Xingyan Shi, Jianwei Huang
Abstract: Knowledge distillation (KD) can enable collaborative learning among distributed clients that have different model architectures and do not share their local data and model parameters with others. Each client updates its local model using the average model output/feature of all client models as the target, known as federated KD. However, existing federated KD methods often do not perform well when clients' local models are trained with heterogeneous local datasets. In this paper, we propose Federated knowledge distillation enabled by Adversarial Learning (FedAL) to address the data heterogeneity among clients. First, to alleviate the local model output divergence across clients caused by data heterogeneity, the server acts as a discriminator to guide clients' local model training to achieve consensus model outputs among clients through a min-max game between clients and the discriminator. Moreover, catastrophic forgetting may happen during the clients' local training and global knowledge transfer due to clients' heterogeneous local data. Towards this challenge, we design the less-forgetting regularization for both local training and global knowledge transfer to guarantee clients' ability to transfer/learn knowledge to/from others. Experimental results show that FedAL and its variants achieve higher accuracy than other federated KD baselines.
Authors: Tong Nie, Guoyang Qin, Wei Ma, Yuewen Mei, Jian Sun
Abstract: Missing data is a pervasive issue in both scientific and engineering tasks, especially for the modeling of spatiotemporal data. This problem attracts many studies to contribute to data-driven solutions. Existing imputation solutions mainly include low-rank models and deep learning models. The former assumes general structural priors but has limited model capacity. The latter possesses salient features of expressivity but lacks prior knowledge of the underlying spatiotemporal structures. Leveraging the strengths of both two paradigms, we demonstrate a low rankness-induced Transformer to achieve a balance between strong inductive bias and high model expressivity. The exploitation of the inherent structures of spatiotemporal data enables our model to learn balanced signal-noise representations, making it generalizable for a variety of imputation problems. We demonstrate its superiority in terms of accuracy, efficiency, and versatility in heterogeneous datasets, including traffic flow, solar energy, smart meters, and air quality. Promising empirical results provide strong conviction that incorporating time series primitives, such as low-rankness, can substantially facilitate the development of a generalizable model to approach a wide range of spatiotemporal imputation problems.
Authors: Anthony Zhou, Amir Barati Farimani
Abstract: The growth of global consumption has motivated important applications of deep learning to smart manufacturing and machine health monitoring. In particular, analyzing vibration data offers great potential to extract meaningful insights into predictive maintenance by the detection of bearing faults. Deep learning can be a powerful method to predict these mechanical failures; however, they lack generalizability to new tasks or datasets and require expensive, labeled mechanical data. We address this by presenting a novel self-supervised pretraining and fine-tuning framework based on transformer models. In particular, we investigate different tokenization and data augmentation strategies to reach state-of-the-art accuracies using transformer models. Furthermore, we demonstrate self-supervised masked pretraining for vibration signals and its application to low-data regimes, task adaptation, and dataset adaptation. Pretraining is able to improve performance on scarce, unseen training samples, as well as when fine-tuning on fault classes outside of the pretraining distribution. Furthermore, pretrained transformers are shown to be able to generalize to a different dataset in a few-shot manner. This introduces a new paradigm where models can be pretrained on unlabeled data from different bearings, faults, and machinery and quickly deployed to new, data-scarce applications to suit specific manufacturing needs.
Authors: Jingyao Wang, Yi Ren, Zeen Song, Jianqi Zhang, Changwen Zheng, Wenwen Qiang
Abstract: Meta-learning enables rapid generalization to new tasks by learning knowledge from various tasks. It is intuitively assumed that as the training progresses, a model will acquire richer knowledge, leading to better generalization performance. However, our experiments reveal an unexpected result: there is negative knowledge transfer between tasks, affecting generalization performance. To explain this phenomenon, we conduct Structural Causal Models (SCMs) for causal analysis. Our investigation uncovers the presence of spurious correlations between task-specific causal factors and labels in meta-learning. Furthermore, the confounding factors differ across different batches. We refer to these confounding factors as "Task Confounders". Based on these findings, we propose a plug-and-play Meta-learning Causal Representation Learner (MetaCRL) to eliminate task confounders. It encodes decoupled generating factors from multiple tasks and utilizes an invariant-based bi-level optimization mechanism to ensure their causality for meta-learning. Extensive experiments on various benchmark datasets demonstrate that our work achieves state-of-the-art (SOTA) performance.
Authors: Avinash Amballa, Gayathri Akkinapalli, Manas Madine, Naga Pavana Priya Yarrabolu, Przemyslaw A. Grabowicz
Abstract: Structured data in the form of tabular datasets contain features that are distinct and discrete, with varying individual and relative importances to the target. Combinations of one or more features may be more predictive and meaningful than simple individual feature contributions. R's mixed effect linear models library allows users to provide such interactive feature combinations in the model design. However, given many features and possible interactions to select from, model selection becomes an exponentially difficult task. We aim to automate the model selection process for predictions on tabular datasets incorporating feature interactions while keeping computational costs small. The framework includes two distinct approaches for feature selection: a Priority-based Random Grid Search and a Greedy Search method. The Priority-based approach efficiently explores feature combinations using prior probabilities to guide the search. The Greedy method builds the solution iteratively by adding or removing features based on their impact. Experiments on synthetic demonstrate the ability to effectively capture predictive feature combinations.
Authors: Naveen Raman, Mateo Espinosa Zarlenga, Juyeon Heo, Mateja Jamnik
Abstract: Concept-based methods explain model predictions using human-understandable concepts. These models require accurate concept predictors, yet the faithfulness of existing concept predictors to their underlying concepts is unclear. In this paper, we investigate the faithfulness of Concept Bottleneck Models (CBMs), a popular family of concept-based architectures, by looking at whether they respect "localities" in datasets. Localities involve using only relevant features when predicting a concept's value. When localities are not considered, concepts may be predicted based on spuriously correlated features, degrading performance and robustness. This work examines how CBM predictions change when perturbing model inputs, and reveals that CBMs may not capture localities, even when independent concepts are localised to non-overlapping feature subsets. Our empirical and theoretical results demonstrate that datasets with correlated concepts may lead to accurate but uninterpretable models that fail to learn localities. Overall, we find that CBM interpretability is fragile, as CBMs occasionally rely upon spurious features, necessitating further research into the robustness of concept predictors.
Authors: Ming-Yi Hong, Yi-Hsiang Huang, Shao-En Lin, You-Chen Teng, Chih-Yu Wang, Che Lin
Abstract: Graph Neural Networks (GNNs) excel in delineating graph structures in diverse domains, including community analysis and recommendation systems. As the interpretation of GNNs becomes increasingly important, the demand for robust baselines and expansive graph datasets is accentuated, particularly in the context of Heterogeneous Information Networks (HIN). Addressing this, we introduce SynHING, a novel framework for Synthetic Heterogeneous Information Network Generation aimed at enhancing graph learning and explanation. SynHING systematically identifies major motifs in a target HIN and employs a bottom-up generation process with intra-cluster and inter-cluster merge modules. This process, supplemented by post-pruning techniques, ensures the synthetic HIN closely mirrors the original graph's structural and statistical properties. Crucially, SynHING provides ground-truth motifs for evaluating GNN explainer models, setting a new standard for explainable, synthetic HIN generation and contributing to the advancement of interpretable machine learning in complex networks.
Authors: Emanuele La Malfa, Christoph Weinhuber, Orazio Torre, Fangru Lin, Samuele Marro, Anthony Cohn, Nigel Shadbolt, Michael Wooldridge
Abstract: Many reasoning, planning, and problem-solving tasks share an intrinsic algorithmic nature: correctly simulating each step is a sufficient condition to solve them correctly. This work studies to what extent Large Language Models (LLMs) can simulate coding and algorithmic tasks to provide insights into general capabilities in such algorithmic reasoning tasks. We introduce benchmarks for straight-line programs, code that contains critical paths, and approximate and redundant instructions. We further assess the simulation capabilities of LLMs with sorting algorithms and nested loops and show that a routine's computational complexity directly affects an LLM's ability to simulate its execution. While the most powerful LLMs exhibit relatively strong simulation capabilities, the process is fragile, seems to rely heavily on pattern recognition, and is affected by memorisation. We propose a novel off-the-shelf prompting method, Chain of Simulation (CoSm), which instructs LLMs to simulate code execution line by line/follow the computation pattern of compilers. CoSm efficiently helps LLMs reduce memorisation and shallow pattern recognition while improving simulation performance. We consider the success of CoSm in code simulation to be inspirational for other general routine simulation reasoning tasks.
Authors: Benao Dai, Baolin Ye, Lingxi Li
Abstract: In order to overcome these challenges, we have proposed a novel hybrid time-varying graph neural network (HTVGNN) for traffic flow prediction. Firstly, a novel time-aware multi-attention mechanism based on time-varying mask enhancement was reported to more accurately model the dynamic temporal dependencies among distinct traffic nodes in the traffic network. Secondly, we have proposed a novel graph learning strategy to concurrently learn both static and dynamic spatial associations between different traffic nodes in road networks. Meanwhile, in order to enhance the learning ability of time-varying graphs, a coupled graph learning mechanism was designed to couple the graphs learned at each time step. Finally, the effectiveness of the proposed method HTVGNN was demonstrated with four real data sets. Simulation results revealed that HTVGNN achieves superior prediction accuracy compared to the state of the art space-time graph neural network models. Additionally, the ablation experiment verifies that the coupled graph learning mechanism can effectively improve the long-term prediction performance of HTVGNN.
Authors: Theodore Papamarkou, Maria Skoularidou, Konstantina Palla, Laurence Aitchison, Julyan Arbel, David Dunson, Maurizio Filippone, Vincent Fortuin, Philipp Hennig, Jos\'e Miguel Hern\'andez-Lobato, Aliaksandr Hubin, Alexander Immer, Theofanis Karaletsos, Mohammad Emtiyaz Khan, Agustinus Kristiadi, Yingzhen Li, Stephan Mandt, Christopher Nemeth, Michael A. Osborne, Tim G. J. Rudner, David R\"ugamer, Yee Whye Teh, Max Welling, Andrew Gordon Wilson, Ruqi Zhang
Abstract: In the current landscape of deep learning research, there is a predominant emphasis on achieving high predictive accuracy in supervised tasks involving large image and language datasets. However, a broader perspective reveals a multitude of overlooked metrics, tasks, and data types, such as uncertainty, active and continual learning, and scientific data, that demand attention. Bayesian deep learning (BDL) constitutes a promising avenue, offering advantages across these diverse settings. This paper posits that BDL can elevate the capabilities of deep learning. It revisits the strengths of BDL, acknowledges existing challenges, and highlights some exciting research avenues aimed at addressing these obstacles. Looking ahead, the discussion focuses on possible ways to combine large-scale foundation models with BDL to unlock their full potential.
Authors: Zhuomin Chen, Jiaxing Zhang, Jingchao Ni, Xiaoting Li, Yuchen Bian, Md Mezbahul Islam, Ananda Mohan Mondal, Hua Wei, Dongsheng Luo
Abstract: Graph Neural Networks (GNNs) have become a building block in graph data processing, with wide applications in critical domains. The growing needs to deploy GNNs in high-stakes applications necessitate explainability for users in the decision-making processes. A popular paradigm for the explainability of GNNs is to identify explainable subgraphs by comparing their labels with the ones of original graphs. This task is challenging due to the substantial distributional shift from the original graphs in the training set to the set of explainable subgraphs, which prevents accurate prediction of labels with the subgraphs. To address it, in this paper, we propose a novel method that generates proxy graphs for explainable subgraphs that are in the distribution of training data. We introduce a parametric method that employs graph generators to produce proxy graphs. A new training objective based on information theory is designed to ensure that proxy graphs not only adhere to the distribution of training data but also preserve explanatory factors. Such generated proxy graphs can be reliably used to approximate the predictions of the labels of explainable subgraphs. Empirical evaluations across various datasets demonstrate our method achieves more accurate explanations for GNNs.
Authors: Zhangyang Gao, Daize Dong, Cheng Tan, Jun Xia, Bozhen Hu, Stan Z. Li
Abstract: Can we model Non-Euclidean graphs as pure language or even Euclidean vectors while retaining their inherent information? The Non-Euclidean property have posed a long term challenge in graph modeling. Despite recent graph neural networks and graph transformers efforts encoding graphs as Euclidean vectors, recovering the original graph from vectors remains a challenge. In this paper, we introduce GraphsGPT, featuring an Graph2Seq encoder that transforms Non-Euclidean graphs into learnable Graph Words in the Euclidean space, along with a GraphGPT decoder that reconstructs the original graph from Graph Words to ensure information equivalence. We pretrain GraphsGPT on $100$M molecules and yield some interesting findings: (1) The pretrained Graph2Seq excels in graph representation learning, achieving state-of-the-art results on $8/9$ graph classification and regression tasks. (2) The pretrained GraphGPT serves as a strong graph generator, demonstrated by its strong ability to perform both few-shot and conditional graph generation. (3) Graph2Seq+GraphGPT enables effective graph mixup in the Euclidean space, overcoming previously known Non-Euclidean challenges. (4) The edge-centric pretraining framework GraphsGPT demonstrates its efficacy in graph domain tasks, excelling in both representation and generation. Code is available at \href{https://github.com/A4Bio/GraphsGPT}{GitHub}.
Authors: Vinitra Swamy, Syrielle Montariol, Julian Blackwell, Jibril Frej, Martin Jaggi, Tanja K\"aser
Abstract: Interpretability for neural networks is a trade-off between three key requirements: 1) faithfulness of the explanation (i.e., how perfectly it explains the prediction), 2) understandability of the explanation by humans, and 3) model performance. Most existing methods compromise one or more of these requirements; e.g., post-hoc approaches provide limited faithfulness, automatically identified feature masks compromise understandability, and intrinsically interpretable methods such as decision trees limit model performance. These shortcomings are unacceptable for sensitive applications such as education and healthcare, which require trustworthy explanations, actionable interpretations, and accurate predictions. In this work, we present InterpretCC (interpretable conditional computation), a family of interpretable-by-design neural networks that guarantee human-centric interpretability, while maintaining comparable performance to state-of-the-art models by adaptively and sparsely activating features before prediction. We extend this idea into an interpretable, global mixture-of-experts (MoE) model that allows humans to specify topics of interest, discretely separates the feature space for each data point into topical subnetworks, and adaptively and sparsely activates these topical subnetworks for prediction. We apply variations of the InterpretCC architecture for text, time series and tabular data across several real-world benchmarks, demonstrating comparable performance with non-interpretable baselines, outperforming interpretable-by-design baselines, and showing higher actionability and usefulness according to a user study.
Authors: Yikun Bai, Rocio Diaz Martin, Abihith Kothapalli, Hengrong Du, Xinran Liu, Soheil Kolouri
Abstract: The Gromov-Wasserstein (GW) distance has gained increasing interest in the machine learning community in recent years, as it allows for the comparison of measures in different metric spaces. To overcome the limitations imposed by the equal mass requirements of the classical GW problem, researchers have begun exploring its application in unbalanced settings. However, Unbalanced GW (UGW) can only be regarded as a discrepancy rather than a rigorous metric/distance between two metric measure spaces (mm-spaces). In this paper, we propose a particular case of the UGW problem, termed Partial Gromov-Wasserstein (PGW). We establish that PGW is a well-defined metric between mm-spaces and discuss its theoretical properties, including the existence of a minimizer for the PGW problem and the relationship between PGW and GW, among others. We then propose two variants of the Frank-Wolfe algorithm for solving the PGW problem and show that they are mathematically and computationally equivalent. Moreover, based on our PGW metric, we introduce the analogous concept of barycenters for mm-spaces. Finally, we validate the effectiveness of our PGW metric and related solvers in applications such as shape matching, shape retrieval, and shape interpolation, comparing them against existing baselines.
Authors: Hossein Zakerinia, Amin Behjati, Christoph H. Lampert
Abstract: We introduce a new framework for studying meta-learning methods using PAC-Bayesian theory. Its main advantage over previous work is that it allows for more flexibility in how the transfer of knowledge between tasks is realized. For previous approaches, this could only happen indirectly, by means of learning prior distributions over models. In contrast, the new generalization bounds that we prove express the process of meta-learning much more directly as learning the learning algorithm that should be used for future tasks. The flexibility of our framework makes it suitable to analyze a wide range of meta-learning mechanisms and even design new mechanisms. Other than our theoretical contributions we also show empirically that our framework improves the prediction quality in practical meta-learning mechanisms.
Authors: Agustinus Kristiadi, Felix Strieth-Kalthoff, Marta Skreta, Pascal Poupart, Al\'an Aspuru-Guzik, Geoff Pleiss
Abstract: Automation is one of the cornerstones of contemporary material discovery. Bayesian optimization (BO) is an essential part of such workflows, enabling scientists to leverage prior domain knowledge into efficient exploration of a large molecular space. While such prior knowledge can take many forms, there has been significant fanfare around the ancillary scientific knowledge encapsulated in large language models (LLMs). However, existing work thus far has only explored LLMs for heuristic materials searches. Indeed, recent work obtains the uncertainty estimate -- an integral part of BO -- from point-estimated, non-Bayesian LLMs. In this work, we study the question of whether LLMs are actually useful to accelerate principled Bayesian optimization in the molecular space. We take a sober, dispassionate stance in answering this question. This is done by carefully (i) viewing LLMs as fixed feature extractors for standard but principled BO surrogate models and by (ii) leveraging parameter-efficient finetuning methods and Bayesian neural networks to obtain the posterior of the LLM surrogate. Our extensive experiments with real-world chemistry problems show that LLMs can be useful for BO over molecules, but only if they have been pretrained or finetuned with domain-specific data.
Authors: Wenjie Xu, Wenbin Wang, Yuning Jiang, Bratislav Svetozarevic, Colin N. Jones
Abstract: We study the problem of preferential Bayesian optimization (BO), where we aim to optimize a black-box function with only preference feedback over a pair of candidate solutions. Inspired by the likelihood ratio idea, we construct a confidence set of the black-box function using only the preference feedback. An optimistic algorithm with an efficient computational method is then developed to solve the problem, which enjoys an information-theoretic bound on the total cumulative regret, a first-of-its-kind for preferential BO. This bound further allows us to design a scheme to report an estimated best solution, with a guaranteed convergence rate. Experimental results on sampled instances from Gaussian processes, standard test functions, and a thermal comfort optimization problem all show that our method stably achieves better or competitive performance as compared to the existing state-of-the-art heuristics, which, however, do not have theoretical guarantees on regret bounds or convergence.
Authors: Lior Cohen, Kaixin Wang, Bingyi Kang, Shie Mannor
Abstract: Motivated by the success of Transformers when applied to sequences of discrete symbols, token-based world models (TBWMs) were recently proposed as sample-efficient methods. In TBWMs, the world model consumes agent experience as a language-like sequence of tokens, where each observation constitutes a sub-sequence. However, during imagination, the sequential token-by-token generation of next observations results in a severe bottleneck, leading to long training times, poor GPU utilization, and limited representations. To resolve this bottleneck, we devise a novel Parallel Observation Prediction (POP) mechanism. POP augments a Retentive Network (RetNet) with a novel forward mode tailored to our reinforcement learning setting. We incorporate POP in a novel TBWM agent named REM (Retentive Environment Model), showcasing a 15.4x faster imagination compared to prior TBWMs. REM attains superhuman performance on 12 out of 26 games of the Atari 100K benchmark, while training in less than 12 hours. Our code is available at \url{https://github.com/leor-c/REM}.
Authors: Yunhao Tang, Zhaohan Daniel Guo, Zeyu Zheng, Daniele Calandriello, R\'emi Munos, Mark Rowland, Pierre Harvey Richemond, Michal Valko, Bernardo \'Avila Pires, Bilal Piot
Abstract: Offline preference optimization allows fine-tuning large models directly from offline data, and has proved effective in recent alignment practices. We propose generalized preference optimization (GPO), a family of offline losses parameterized by a general class of convex functions. GPO enables a unified view over preference optimization, encompassing existing algorithms such as DPO, IPO and SLiC as special cases, while naturally introducing new variants. The GPO framework also sheds light on how offline algorithms enforce regularization, through the design of the convex function that defines the loss. Our analysis and experiments reveal the connections and subtle differences between the offline regularization and the KL divergence regularization intended by the canonical RLHF formulation. In a controlled setting akin to Gao et al 2023, we also show that different GPO variants achieve similar trade-offs between regularization and performance, though the optimal values of hyper-parameter might differ as predicted by theory. In all, our results present new algorithmic toolkits and empirical insights to alignment practitioners.
Authors: Muning Wen, Cheng Deng, Jun Wang, Weinan Zhang, Ying Wen
Abstract: Large Language Models (LLMs) have shown promise as intelligent agents in interactive decision-making tasks. Traditional approaches often depend on meticulously designed prompts, high-quality examples, or additional reward models for in-context learning, supervised fine-tuning, or RLHF. Reinforcement learning (RL) presents a dynamic alternative for LLMs to overcome these dependencies by engaging directly with task-specific environments. Nonetheless, it faces significant hurdles: 1) instability stemming from the exponentially vast action space requiring exploration; 2) challenges in assigning token-level credit based on action-level reward signals, resulting in discord between maximizing rewards and accurately modeling corpus data. In response to these challenges, we introduce Entropy-Regularized Token-level Policy Optimization (ETPO), an entropy-augmented RL method tailored for optimizing LLMs at the token level. At the heart of ETPO is our novel per-token soft Bellman update, designed to harmonize the RL process with the principles of language modeling. This methodology decomposes the Q-function update from a coarse action-level view to a more granular token-level perspective, backed by theoretical proof of optimization consistency. Crucially, this decomposition renders linear time complexity in action exploration. We assess the effectiveness of ETPO within a simulated environment that models data science code generation as a series of multi-step interactive tasks; results underline ETPO's potential as a robust method for refining the interactive decision-making capabilities of language agents. For a more detailed preliminary work describing our motivation for token-level decomposition and applying it in PPO methods, please refer to arXiv:2405.15821.
Authors: Wenzhi Gao, Chunlin Sun, Chenyu Xue, Dongdong Ge, Yinyu Ye
Abstract: Online linear programming plays an important role in both revenue management and resource allocation, and recent research has focused on developing efficient first-order online learning algorithms. Despite the empirical success of first-order methods, they typically achieve a regret no better than $\mathcal{O}(\sqrt{T})$, which is suboptimal compared to the $\mathcal{O}(\log T)$ bound guaranteed by the state-of-the-art linear programming (LP)-based online algorithms. This paper establishes several important facts about online linear programming, which unveils the challenge for first-order-method-based online algorithms to achieve beyond $\mathcal{O}(\sqrt{T})$ regret. To address the challenge, we introduce a new algorithmic framework that decouples learning from decision-making. For the first time, we show that first-order methods can attain regret $\mathcal{O}(T^{1/3})$ with this new framework.
Authors: Mohamed Ghanem, Frederik Schmitt, Julian Siber, Bernd Finkbeiner
Abstract: Training neural networks on NP-complete problems typically demands very large amounts of training data and often needs to be coupled with computationally expensive symbolic verifiers to ensure output correctness. In this paper, we present NeuRes, a neuro-symbolic approach to address both challenges for propositional satisfiability, being the quintessential NP-complete problem. By combining certificate-driven training and expert iteration, our model learns better representations than models trained for classification only, with a much higher data efficiency -- requiring orders of magnitude less training data. NeuRes employs propositional resolution as a proof system to generate proofs of unsatisfiability and to accelerate the process of finding satisfying truth assignments, exploring both possibilities in parallel. To realize this, we propose an attention-based architecture that autoregressively selects pairs of clauses from a dynamic formula embedding to derive new clauses. Furthermore, we employ expert iteration whereby model-generated proofs progressively replace longer teacher proofs as the new ground truth. This enables our model to reduce a dataset of proofs generated by an advanced solver by ~32% after training on it with no extra guidance. This shows that NeuRes is not limited by the optimality of the teacher algorithm owing to its self-improving workflow. We show that our model achieves far better performance than NeuroSAT in terms of both correctly classified and proven instances.
Authors: Jose Pablo Folch, Calvin Tsay, Robert M Lee, Behrang Shafei, Weronika Ormaniec, Andreas Krause, Mark van der Wilk, Ruth Misener, Mojm\'ir Mutn\'y
Abstract: Bayesian optimization is a methodology to optimize black-box functions. Traditionally, it focuses on the setting where you can arbitrarily query the search space. However, many real-life problems do not offer this flexibility; in particular, the search space of the next query may depend on previous ones. Example challenges arise in the physical sciences in the form of local movement constraints, required monotonicity in certain variables, and transitions influencing the accuracy of measurements. Altogether, such transition constraints necessitate a form of planning. This work extends classical Bayesian optimization via the framework of Markov Decision Processes. We iteratively solve a tractable linearization of our utility function using reinforcement learning to obtain a policy that plans ahead for the entire horizon. This is a parallel to the optimization of an acquisition function in policy space. The resulting policy is potentially history-dependent and non-Markovian. We showcase applications in chemical reactor optimization, informative path planning, machine calibration, and other synthetic examples.
Authors: Shayan Shirahmad Gale Bagi, Zahra Gharaee, Oliver Schulte, Mark Crowley
Abstract: Learning causal representations from observational and interventional data in the absence of known ground-truth graph structures necessitates implicit latent causal representation learning. Implicit learning of causal mechanisms typically involves two categories of interventional data: hard and soft interventions. In real-world scenarios, soft interventions are often more realistic than hard interventions, as the latter require fully controlled environments. Unlike hard interventions, which directly force changes in a causal variable, soft interventions exert influence indirectly by affecting the causal mechanism. However, the subtlety of soft interventions impose several challenges for learning causal models. One challenge is that soft intervention's effects are ambiguous, since parental relations remain intact. In this paper, we tackle the challenges of learning causal models using soft interventions while retaining implicit modeling. Our approach models the effects of soft interventions by employing a \textit{causal mechanism switch variable} designed to toggle between different causal mechanisms. In our experiments, we consistently observe improved learning of identifiable, causal representations, compared to baseline approaches.
Authors: Josep Lumbreras, Marco Tomamichel
Abstract: We study a noise model for linear stochastic bandits for which the subgaussian noise parameter vanishes linearly as we select actions on the unit sphere closer and closer to the unknown vector. We introduce an algorithm for this problem that exhibits a minimax regret scaling as $\log^3(T)$ in the time horizon $T$, in stark contrast the square root scaling of this regret for typical bandit algorithms. Our strategy, based on weighted least-squares estimation, achieves the eigenvalue relation $\lambda_{\min} ( V_t ) = \Omega (\sqrt{\lambda_{\max}(V_t ) })$ for the design matrix $V_t$ at each time step $t$ through geometrical arguments that are independent of the noise model and might be of independent interest. This allows us to tightly control the expected regret in each time step to be of the order $O(\frac1{t})$, leading to the logarithmic scaling of the cumulative regret.
Authors: Futa Waseda, Ching-Chun Chang, Isao Echizen
Abstract: Although adversarial training has been the state-of-the-art approach to defend against adversarial examples (AEs), it suffers from a robustness-accuracy trade-off, where high robustness is achieved at the cost of clean accuracy. In this work, we leverage invariance regularization on latent representations to learn discriminative yet adversarially invariant representations, aiming to mitigate this trade-off. We analyze two key issues in representation learning with invariance regularization: (1) a "gradient conflict" between invariance loss and classification objectives, leading to suboptimal convergence, and (2) the mixture distribution problem arising from diverged distributions of clean and adversarial inputs. To address these issues, we propose Asymmetrically Representation-regularized Adversarial Training (AR-AT), which incorporates asymmetric invariance loss with stop-gradient operation and a predictor to improve the convergence, and a split-BatchNorm (BN) structure to resolve the mixture distribution problem. Our method significantly improves the robustness-accuracy trade-off by learning adversarially invariant representations without sacrificing discriminative ability. Furthermore, we discuss the relevance of our findings to knowledge-distillation-based defense methods, contributing to a deeper understanding of their relative successes.
Authors: Cong Zhang, Zhiguang Cao, Yaoxin Wu, Wen Song, Jing Sun
Abstract: Existing learning-based methods for solving job shop scheduling problems (JSSP) usually use off-the-shelf GNN models tailored to undirected graphs and neglect the rich and meaningful topological structures of disjunctive graphs (DGs). This paper proposes the topology-aware bidirectional graph attention network (TBGAT), a novel GNN architecture based on the attention mechanism, to embed the DG for solving JSSP in a local search framework. Specifically, TBGAT embeds the DG from a forward and a backward view, respectively, where the messages are propagated by following the different topologies of the views and aggregated via graph attention. Then, we propose a novel operator based on the message-passing mechanism to calculate the forward and backward topological sorts of the DG, which are the features for characterizing the topological structures and exploited by our model. In addition, we theoretically and experimentally show that TBGAT has linear computational complexity to the number of jobs and machines, respectively, strengthening our method's practical value. Besides, extensive experiments on five synthetic datasets and seven classic benchmarks show that TBGAT achieves new SOTA results by outperforming a wide range of neural methods by a large margin. All the code and data are publicly available online at https://github.com/zcaicaros/TBGAT.
Authors: Yutong Feng, Qiongyan Wang, Yutong Xia, Junlin Huang, Siru Zhong, Kun Wang, Shifen Cheng, Yuxuan Liang
Abstract: The air quality inference problem aims to utilize historical data from a limited number of observation sites to infer the air quality index at an unknown location. Considering the sparsity of data due to the high maintenance cost of the stations, good inference algorithms can effectively save the cost and refine the data granularity. While spatio-temporal graph neural networks have made excellent progress on this problem, their non-Euclidean and discrete data structure modeling of reality limits its potential. In this work, we make the first attempt to combine two different spatio-temporal perspectives, fields and graphs, by proposing a new model, Spatio-Temporal Field Neural Network, and its corresponding new framework, Pyramidal Inference. Extensive experiments validate that our model achieves state-of-the-art performance in nationwide air quality inference in the Chinese Mainland, demonstrating the superiority of our proposed model and framework.
Authors: Katie Kang, Eric Wallace, Claire Tomlin, Aviral Kumar, Sergey Levine
Abstract: Large language models are known to hallucinate when faced with unfamiliar queries, but the underlying mechanism that govern how models hallucinate are not yet fully understood. In this work, we find that unfamiliar examples in the models' finetuning data -- those that introduce concepts beyond the base model's scope of knowledge -- are crucial in shaping these errors. In particular, we find that an LLM's hallucinated predictions tend to mirror the responses associated with its unfamiliar finetuning examples. This suggests that by modifying how unfamiliar finetuning examples are supervised, we can influence a model's responses to unfamiliar queries (e.g., say ``I don't know''). We empirically validate this observation in a series of controlled experiments involving SFT, RL, and reward model finetuning on TriviaQA and MMLU. Our work further investigates RL finetuning strategies for improving the factuality of long-form model generations. We find that, while hallucinations from the reward model can significantly undermine the effectiveness of RL factuality finetuning, strategically controlling how reward models hallucinate can minimize these negative effects. Leveraging our previous observations on controlling hallucinations, we propose an approach for learning more reliable reward models, and show that they improve the efficacy of RL factuality finetuning in long-form biography and book/movie plot generation tasks.
Authors: Chuning Zhu, Xinqi Wang, Tyler Han, Simon S. Du, Abhishek Gupta
Abstract: Intelligent agents must be generalists, capable of quickly adapting to various tasks. In reinforcement learning (RL), model-based RL learns a dynamics model of the world, in principle enabling transfer to arbitrary reward functions through planning. However, autoregressive model rollouts suffer from compounding error, making model-based RL ineffective for long-horizon problems. Successor features offer an alternative by modeling a policy's long-term state occupancy, reducing policy evaluation under new tasks to linear reward regression. Yet, policy improvement with successor features can be challenging. This work proposes a novel class of models, i.e., generalized occupancy models (GOMs), that learn a distribution of successor features from a stationary dataset, along with a policy that acts to realize different successor features. These models can quickly select the optimal action for arbitrary new tasks. By directly modeling long-term outcomes in the dataset, GOMs avoid compounding error while enabling rapid transfer across reward functions. We present a practical instantiation of GOMs using diffusion models and show their efficacy as a new class of transferable models, both theoretically and empirically across various simulated robotics problems. Videos and code at https://weirdlabuw.github.io/gom/.
Authors: Mohamed elShehaby, Aditya Kotha, Ashraf Matrawy
Abstract: Adversarial training enhances the robustness of Machine Learning (ML) models against adversarial attacks. However, obtaining labeled training and adversarial training data in network/cybersecurity domains is challenging and costly. Therefore, this letter introduces Adaptive Continuous Adversarial Training (ACAT), a method that integrates adversarial training samples into the model during continuous learning sessions using real-world detected adversarial data. Experimental results with a SPAM detection dataset demonstrate that ACAT reduces the time required for adversarial sample detection compared to traditional processes. Moreover, the accuracy of the under-attack ML-based SPAM filter increased from 69% to over 88% after just three retraining sessions.
Authors: Yifei Shen, Xinyang Jiang, Yezhen Wang, Yifan Yang, Dongqi Han, Dongsheng Li
Abstract: Adding additional control to pretrained diffusion models has become an increasingly popular research area, with extensive applications in computer vision, reinforcement learning, and AI for science. Recently, several studies have proposed training-free loss-based guidance by using off-the-shelf networks pretrained on clean images. This approach enables zero-shot conditional generation for universal control formats, which appears to offer a free lunch in diffusion guidance. In this paper, we aim to develop a deeper understanding of training-free guidance, as well as overcome its limitations. We offer a theoretical analysis that supports training-free guidance from the perspective of optimization, distinguishing it from classifier-based (or classifier-free) guidance. To elucidate their drawbacks, we theoretically demonstrate that training-free guidance is more susceptible to adversarial gradients and exhibits slower convergence rates compared to classifier guidance. We then introduce a collection of techniques designed to overcome the limitations, accompanied by theoretical rationale and empirical evidence. Our experiments in image and motion generation confirm the efficacy of these techniques.
Authors: Anthony Zhou, Amir Barati Farimani
Abstract: Neural solvers for partial differential equations (PDEs) have great potential to generate fast and accurate physics solutions, yet their practicality is currently limited by their generalizability. PDEs evolve over broad scales and exhibit diverse behaviors; predicting these phenomena will require learning representations across a wide variety of inputs which may encompass different coefficients, boundary conditions, resolutions, or even equations. As a step towards generalizable PDE modeling, we adapt masked pretraining for physics problems. Through self-supervised learning across PDEs, masked autoencoders can consolidate heterogeneous physics to learn meaningful latent representations and perform latent PDE arithmetic in this space. Furthermore, we demonstrate that masked pretraining can improve PDE coefficient regression and the classification of PDE features. Lastly, conditioning neural solvers on learned latent representations can improve time-stepping and super-resolution performance across a variety of coefficients, discretizations, or boundary conditions, as well as on unseen PDEs. We hope that masked pretraining can emerge as a unifying method across large, unlabeled, and heterogeneous datasets to learn latent physics at scale.
Authors: Jeremy E. Cohen, Valentin Leplat
Abstract: Regularized nonnegative low-rank approximations such as sparse Nonnegative Matrix Factorization or sparse Nonnegative Tucker Decomposition are an important branch of dimensionality reduction models with enhanced interpretability. However, from a practical perspective, the choice of regularizers and regularization coefficients, as well as the design of efficient algorithms, is challenging because of the multifactor nature of these models and the lack of theory to back these choices. This paper aims at improving upon these issues. By studying a more general model called the Homogeneous Regularized Scale-Invariant, we prove that the scale-invariance inherent to low-rank approximation models causes an implicit regularization with both unexpected beneficial and detrimental effects. This observation allows to better understand the effect of regularization functions in low-rank approximation models, to guide the choice of the regularization hyperparameters, and to design balancing strategies to enhance the convergence speed of dedicated optimization algorithms. Some of these results were already known but restricted to specific instances of regularized low-rank approximations. We also derive a generic Majorization Minimization algorithm that handles many regularized nonnegative low-rank approximations, with convergence guarantees. We showcase our contributions on sparse Nonnegative Matrix Factorization, ridge-regularized Canonical Polyadic decomposition and sparse Nonnegative Tucker Decomposition.
Authors: Shuaicheng Niu, Chunyan Miao, Guohao Chen, Pengcheng Wu, Peilin Zhao
Abstract: Test-time adaptation has proven effective in adapting a given trained model to unseen test samples with potential distribution shifts. However, in real-world scenarios, models are usually deployed on resource-limited devices, e.g., FPGAs, and are often quantized and hard-coded with non-modifiable parameters for acceleration. In light of this, existing methods are often infeasible since they heavily depend on computation-intensive backpropagation for model updating that may be not supported. To address this, we propose a test-time Forward-Optimization Adaptation (FOA) method. In FOA, we seek to solely learn a newly added prompt (as model's input) via a derivative-free covariance matrix adaptation evolution strategy. To make this strategy work stably under our online unsupervised setting, we devise a novel fitness function by measuring test-training statistic discrepancy and model prediction entropy. Moreover, we design an activation shifting scheme that directly tunes the model activations for shifted test samples, making them align with the source training domain, thereby further enhancing adaptation performance. Without using any backpropagation and altering model weights, FOA runs on quantized 8-bit ViT outperforms gradient-based TENT on full-precision 32-bit ViT, while achieving an up to 24-fold memory reduction on ImageNet-C.
Authors: Xiwei Zhang, Tao Li
Abstract: We study the convergence of recursive regularized learning algorithms in the reproducing kernel Hilbert space (RKHS) with dependent and non-stationary online data streams. Firstly, we study the mean square asymptotic stability of a class of random difference equations in RKHS, whose non-homogeneous terms are martingale difference sequences dependent on the homogeneous ones. Secondly, we introduce the concept of random Tikhonov regularization path, and show that if the regularization path is slowly time-varying in some sense, then the output of the algorithm is consistent with the regularization path in mean square. Furthermore, if the data streams also satisfy the RKHS persistence of excitation condition, i.e. there exists a fixed length of time period, such that each eigenvalue of the conditional expectation of the operators induced by the input data accumulated over every time period has a uniformly positive lower bound with respect to time, then the output of the algorithm is consistent with the unknown function in mean square. Finally, for the case with independent and non-identically distributed data streams, the algorithm achieves the mean square consistency provided the marginal probability measures induced by the input data are slowly time-varying and the average measure over each fixed-length time period has a uniformly strictly positive lower bound.
Authors: Namasivayam Kalithasan, Sachit Sachdeva, Himanshu Gaurav Singh, Vishal Bindal, Arnav Tuli, Gurarmaan Singh Panjeta, Divyanshu Aggarwal, Rohan Paul, Parag Singla
Abstract: Our goal is to enable embodied agents to learn inductively generalizable spatial concepts, e.g., learning staircase as an inductive composition of towers of increasing height. Given a human demonstration, we seek a learning architecture that infers a succinct ${program}$ representation that explains the observed instance. Additionally, the approach should generalize inductively to novel structures of different sizes or complex structures expressed as a hierarchical composition of previously learned concepts. Existing approaches that use code generation capabilities of pre-trained large (visual) language models, as well as purely neural models, show poor generalization to a-priori unseen complex concepts. Our key insight is to factor inductive concept learning as (i) ${\it Sketch:}$ detecting and inferring a coarse signature of a new concept (ii) ${\it Plan:}$ performing MCTS search over grounded action sequences (iii) ${\it Generalize:}$ abstracting out grounded plans as inductive programs. Our pipeline facilitates generalization and modular reuse, enabling continual concept learning. Our approach combines the benefits of the code generation ability of large language models (LLM) along with grounded neural representations, resulting in neuro-symbolic programs that show stronger inductive generalization on the task of constructing complex structures in relation to LLM-only and neural-only approaches. Furthermore, we demonstrate reasoning and planning capabilities with learned concepts for embodied instruction following.
Authors: Ren\'e Heinrich, Bernhard Sick, Christoph Scholz
Abstract: Recently, scientists have proposed several deep learning models to monitor the diversity of bird species. These models can detect bird species with high accuracy by analyzing acoustic signals. However, traditional deep learning algorithms are black-box models that provide no insight into their decision-making process. For domain experts, such as ornithologists, it is crucial that these models are not only efficient, but also interpretable in order to be used as assistive tools. In this study, we present an adaption of the Prototypical Part Network (ProtoPNet) for audio classification that provides inherent interpretability through its model architecture. Our approach is based on a ConvNeXt backbone architecture for feature extraction and learns prototypical patterns for each bird species using spectrograms of the training data. Classification of new data is done by comparison with these prototypes in latent space, which simultaneously serve as easily understandable explanations for the model's decisions. We evaluated the performance of our model on seven different datasets representing bird species from different geographical regions. In our experiments, the model showed excellent results, achieving an average AUROC of 0.82 and an average cmAP of 0.37 across the seven datasets, making it comparable to state-of-the-art black-box models for bird sound classification. Thus, this work demonstrates that even for the challenging task of bioacoustic bird classification, powerful yet interpretable deep learning models can be developed to provide valuable insights to domain experts.
Authors: Zhaolin Gao, Jonathan D. Chang, Wenhao Zhan, Owen Oertell, Gokul Swamy, Kiant\'e Brantley, Thorsten Joachims, J. Andrew Bagnell, Jason D. Lee, Wen Sun
Abstract: While originally developed for continuous control problems, Proximal Policy Optimization (PPO) has emerged as the work-horse of a variety of reinforcement learning (RL) applications, including the fine-tuning of generative models. Unfortunately, PPO requires multiple heuristics to enable stable convergence (e.g. value networks, clipping), and is notorious for its sensitivity to the precise implementation of these components. In response, we take a step back and ask what a minimalist RL algorithm for the era of generative models would look like. We propose REBEL, an algorithm that cleanly reduces the problem of policy optimization to regressing the relative reward between two completions to a prompt in terms of the policy, enabling strikingly lightweight implementation. In theory, we prove that fundamental RL algorithms like Natural Policy Gradient can be seen as variants of REBEL, which allows us to match the strongest known theoretical guarantees in terms of convergence and sample complexity in the RL literature. REBEL can also cleanly incorporate offline data and be extended to handle the intransitive preferences we frequently see in practice. Empirically, we find that REBEL provides a unified approach to language modeling and image generation with stronger or similar performance as PPO and DPO, all while being simpler to implement and more computationally efficient than PPO. When fine-tuning Llama-3-8B-Instruct, REBEL achieves strong performance in AlpacaEval 2.0, MT-Bench, and Open LLM Leaderboard.
Authors: Xingli Fang, Jung-Eun Kim
Abstract: Membership inference attacks (MIAs) are currently considered one of the main privacy attack strategies, and their defense mechanisms have also been extensively explored. However, there is still a gap between the existing defense approaches and ideal models in performance and deployment costs. In particular, we observed that the privacy vulnerability of the model is closely correlated with the gap between the model's data-memorizing ability and generalization ability. To address this, we propose a new architecture-agnostic training paradigm called center-based relaxed learning (CRL), which is adaptive to any classification model and provides privacy preservation by sacrificing a minimal or no loss of model generalizability. We emphasize that CRL can better maintain the model's consistency between member and non-member data. Through extensive experiments on standard classification datasets, we empirically show that this approach exhibits comparable performance without requiring additional model capacity or data costs.
Authors: Kang Du, Yu Xiang
Abstract: Causal inference from observational data following the restricted structural causal models (SCM) framework hinges largely on the asymmetry between cause and effect from the data generating mechanisms, such as non-Gaussianity or non-linearity. This methodology can be adapted to stationary time series, yet inferring causal relationships from nonstationary time series remains a challenging task. In this work, we propose a new class of restricted SCM, via a time-varying filter and stationary noise, and exploit the asymmetry from nonstationarity for causal identification in both bivariate and network settings. We propose efficient procedures by leveraging powerful estimates of the bivariate evolutionary spectra for slowly varying processes. Various synthetic and real datasets that involve high-order and non-smooth filters are evaluated to demonstrate the effectiveness of our proposed methodology.
Authors: Hanshu Yan, Xingchao Liu, Jiachun Pan, Jun Hao Liew, Qiang Liu, Jiashi Feng
Abstract: We present Piecewise Rectified Flow (PeRFlow), a flow-based method for accelerating diffusion models. PeRFlow divides the sampling process of generative flows into several time windows and straightens the trajectories in each interval via the reflow operation, thereby approaching piecewise linear flows. PeRFlow achieves superior performance in a few-step generation. Moreover, through dedicated parameterizations, the PeRFlow models inherit knowledge from the pretrained diffusion models. Thus, the training converges fast and the obtained models show advantageous transfer ability, serving as universal plug-and-play accelerators that are compatible with various workflows based on the pre-trained diffusion models. Codes for training and inference are publicly released. https://github.com/magic-research/piecewise-rectified-flow
URLs: https://github.com/magic-research/piecewise-rectified-flow
Authors: Francisco Eiras, Aleksandar Petrov, Bertie Vidgen, Christian Schroeder, Fabio Pizzati, Katherine Elkins, Supratik Mukhopadhyay, Adel Bibi, Aaron Purewal, Csaba Botos, Fabro Steibel, Fazel Keshtkar, Fazl Barez, Genevieve Smith, Gianluca Guadagni, Jon Chun, Jordi Cabot, Joseph Imperial, Juan Arturo Nolazco, Lori Landay, Matthew Jackson, Phillip H. S. Torr, Trevor Darrell, Yong Lee, Jakob Foerster
Abstract: Applications of Generative AI (Gen AI) are expected to revolutionize a number of different areas, ranging from science & medicine to education. The potential for these seismic changes has triggered a lively debate about the potential risks of the technology, and resulted in calls for tighter regulation, in particular from some of the major tech companies who are leading in AI development. This regulation is likely to put at risk the budding field of open-source generative AI. Using a three-stage framework for Gen AI development (near, mid and long-term), we analyze the risks and opportunities of open-source generative AI models with similar capabilities to the ones currently available (near to mid-term) and with greater capabilities (long-term). We argue that, overall, the benefits of open-source Gen AI outweigh its risks. As such, we encourage the open sourcing of models, training and evaluation data, and provide a set of recommendations and best practices for managing risks associated with open-source generative AI.
Authors: Charikleia Iakovidou, Kibaek Kim
Abstract: Federated learning (FL) was recently proposed to securely train models with data held over multiple locations ("clients") under the coordination of a central server. Two major challenges hindering the performance of FL algorithms are long training times caused by straggling clients, and a decline in model accuracy under non-iid local data distributions ("client drift"). In this work, we propose and analyze Asynchronous Exact Averaging (AREA), a new stochastic (sub)gradient algorithm that utilizes asynchronous communication to speed up convergence and enhance scalability, and employs client memory to correct the client drift caused by variations in client update frequencies. Moreover, AREA is, to the best of our knowledge, the first method that is guaranteed to converge under arbitrarily long delays, without the use of delay-adaptive stepsizes, and (i) for strongly convex, smooth functions, asymptotically converges to an error neighborhood whose size depends only on the variance of the stochastic gradients used with respect to the number of iterations, and (ii) for convex, non-smooth functions, matches the convergence rate of the centralized stochastic subgradient method up to a constant factor, which depends on the average of the individual client update frequencies instead of their minimum (or maximum). Our numerical results validate our theoretical analysis and indicate AREA outperforms state-of-the-art methods when local data are highly non-iid, especially as the number of clients grows.
Authors: Cheng Chen, Lei Fan
Abstract: Landslide susceptibility mapping (LSM) is crucial for identifying high-risk areas and informing prevention strategies. This study investigates the interpretability of statistical, machine learning (ML), and deep learning (DL) models in predicting landslide susceptibility. This is achieved by incorporating various relevant interpretation methods and two types of input factors: a comprehensive set of 19 contributing factors that are statistically relevant to landslides, as well as a dedicated set of 9 triggering factors directly associated with triggering landslides. Given that model performance is a crucial metric in LSM, our investigations into interpretability naturally involve assessing and comparing LSM accuracy across different models considered. In our investigation, the convolutional neural network model achieved the highest accuracy (0.8447 with 19 factors; 0.8048 with 9 factors), while Extreme Gradient Boosting and Support Vector Machine also demonstrated strong predictive capabilities, outperforming conventional statistical models. These findings indicate that DL and sophisticated ML algorithms can effectively capture the complex relationships between input factors and landslide occurrence. However, the interpretability of predictions varied among different models, particularly when using the broader set of 19 contributing factors. Explanation methods like SHAP, LIME, and DeepLIFT also led to variations in interpretation results. Using a comprehensive set of 19 contributing factors improved prediction accuracy but introduced complexities and inconsistency in model interpretations. Focusing on a dedicated set of 9 triggering factors sacrificed some predictive power but enhanced interpretability, as evidenced by more consistent key factors identified across various models and alignment with the findings of field investigation reports....
Authors: Jan-Christoph Klie, Juan Haladjian, Marc Kirchner, Rahul Nair
Abstract: Annotated datasets are an essential ingredient to train, evaluate, compare and productionalize supervised machine learning models. It is therefore imperative that annotations are of high quality. For their creation, good quality management and thereby reliable quality estimates are needed. Then, if quality is insufficient during the annotation process, rectifying measures can be taken to improve it. Quality estimation is often performed by having experts manually label instances as correct or incorrect. But checking all annotated instances tends to be expensive. Therefore, in practice, usually only subsets are inspected; sizes are chosen mostly without justification or regard to statistical power and more often than not, are relatively small. Basing estimates on small sample sizes, however, can lead to imprecise values for the error rate. Using unnecessarily large sample sizes costs money that could be better spent, for instance on more annotations. Therefore, we first describe in detail how to use confidence intervals for finding the minimal sample size needed to estimate the annotation error rate. Then, we propose applying acceptance sampling as an alternative to error rate estimation We show that acceptance sampling can reduce the required sample sizes up to 50% while providing the same statistical guarantees.
Authors: Chang Zhou, Yang Zhao, Jin Cao, Yi Shen, Xiaoling Cui, Chiyu Cheng
Abstract: This paper explores the integration of strategic optimization methods in search advertising, focusing on ad ranking and bidding mechanisms within E-commerce platforms. By employing a combination of reinforcement learning and evolutionary strategies, we propose a dynamic model that adjusts to varying user interactions and optimizes the balance between advertiser cost, user relevance, and platform revenue. Our results suggest significant improvements in ad placement accuracy and cost efficiency, demonstrating the model's applicability in real-world scenarios.
Authors: Wanghan Xu, Fenghua Ling, Wenlong Zhang, Tao Han, Hao Chen, Wanli Ouyang, Lei Bai
Abstract: Data-driven artificial intelligence (AI) models have made significant advancements in weather forecasting, particularly in medium-range and nowcasting. However, most data-driven weather forecasting models are black-box systems that focus on learning data mapping rather than fine-grained physical evolution in the time dimension. Consequently, the limitations in the temporal scale of datasets prevent these models from forecasting at finer time scales. This paper proposes a physics-AI hybrid model (i.e., WeatherGFT) which Generalizes weather forecasts to Finer-grained Temporal scales beyond training dataset. Specifically, we employ a carefully designed PDE kernel to simulate physical evolution on a small time scale (e.g., 300 seconds) and use a parallel neural networks with a learnable router for bias correction. Furthermore, we introduce a lead time-aware training framework to promote the generalization of the model at different lead times. The weight analysis of physics-AI modules indicates that physics conducts major evolution while AI performs corrections adaptively. Extensive experiments show that WeatherGFT trained on an hourly dataset, achieves state-of-the-art performance across multiple lead times and exhibits the capability to generalize 30-minute forecasts.
Authors: Oscar Davis, Samuel Kessler, Mircea Petrache, \.Ismail \.Ilkan Ceylan, Michael Bronstein, Avishek Joey Bose
Abstract: Generative modeling over discrete data has recently seen numerous success stories, with applications spanning language modeling, biological sequence design, and graph-structured molecular data. The predominant generative modeling paradigm for discrete data is still autoregressive, with more recent alternatives based on diffusion or flow-matching falling short of their impressive performance in continuous data settings, such as image or video generation. In this work, we introduce Fisher-Flow, a novel flow-matching model for discrete data. Fisher-Flow takes a manifestly geometric perspective by considering categorical distributions over discrete data as points residing on a statistical manifold equipped with its natural Riemannian metric: the $\textit{Fisher-Rao metric}$. As a result, we demonstrate discrete data itself can be continuously reparameterised to points on the positive orthant of the $d$-hypersphere $\mathbb{S}^d_+$, which allows us to define flows that map any source distribution to target in a principled manner by transporting mass along (closed-form) geodesics of $\mathbb{S}^d_+$. Furthermore, the learned flows in Fisher-Flow can be further bootstrapped by leveraging Riemannian optimal transport leading to improved training dynamics. We prove that the gradient flow induced by Fisher-Flow is optimal in reducing the forward KL divergence. We evaluate Fisher-Flow on an array of synthetic and diverse real-world benchmarks, including designing DNA Promoter, and DNA Enhancer sequences. Empirically, we find that Fisher-Flow improves over prior diffusion and flow-matching models on these benchmarks.
Authors: Dimitrios Bachtis, Giulio Biroli, Aur\'elien Decelle, Beatriz Seoane
Abstract: In this paper, we investigate the feature encoding process in a prototypical energy-based generative model, the Restricted Boltzmann Machine (RBM). We start with an analytical investigation using simplified architectures and data structures, and end with numerical analysis of real trainings on real datasets. Our study tracks the evolution of the model's weight matrix through its singular value decomposition, revealing a series of phase transitions associated to a progressive learning of the principal modes of the empirical probability distribution. The model first learns the center of mass of the modes and then progressively resolve all modes through a cascade of phase transitions. We first describe this process analytically in a controlled setup that allows us to study analytically the training dynamics. We then validate our theoretical results by training the Bernoulli-Bernoulli RBM on real data sets. By using data sets of increasing dimension, we show that learning indeed leads to sharp phase transitions in the high-dimensional limit. Moreover, we propose and test a mean-field finite-size scaling hypothesis. This shows that the first phase transition is in the same universality class of the one we studied analytically, and which is reminiscent of the mean-field paramagnetic-to-ferromagnetic phase transition.
Authors: Duke Nguyen, Aditya Joshi, Flora Salim
Abstract: Linearization of attention using various kernel approximation and kernel learning techniques has shown promise. Past methods use a subset of combinations of component functions and weight matrices within the random features paradigm. We identify the need for a systematic comparison of different combinations of weight matrix and component functions for attention learning in Transformer. In this work, we introduce Spectraformer, a unified framework for approximating and learning the kernel function in linearized attention of the Transformer. We experiment with broad classes of component functions and weight matrices for three textual tasks in the LRA benchmark. Our experimentation with multiple combinations of component functions and weight matrices leads us to a novel combination with 23.4% faster training time and 25.2% lower memory consumption over the previous SOTA random feature Transformer, while maintaining the performance, as compared to the Original Transformer. Our code is available at: https://github.com/dukeraphaelng/spectraformer .
Authors: Anthony Gruber, Kookjin Lee, Haksoo Lim, Noseong Park, Nathaniel Trask
Abstract: Metriplectic systems are learned from data in a way that scales quadratically in both the size of the state and the rank of the metriplectic data. Besides being provably energy conserving and entropy stable, the proposed approach comes with approximation results demonstrating its ability to accurately learn metriplectic dynamics from data as well as an error estimate indicating its potential for generalization to unseen timescales when approximation error is low. Examples are provided which illustrate performance in the presence of both full state information as well as when entropic variables are unknown, confirming that the proposed approach exhibits superior accuracy and scalability without compromising on model expressivity.
Authors: Zechun Liu, Changsheng Zhao, Igor Fedorov, Bilge Soran, Dhruv Choudhary, Raghuraman Krishnamoorthi, Vikas Chandra, Yuandong Tian, Tijmen Blankevoort
Abstract: Post-training quantization (PTQ) techniques applied to weights, activations, and the KV cache greatly reduce memory usage, latency, and power consumption of Large Language Models (LLMs), but may lead to large quantization errors when outliers are present. Recent findings suggest that rotating activation or weight matrices helps remove outliers and benefits quantization. In this work, we identify a collection of applicable rotation parameterizations that lead to identical outputs in full-precision Transformer architectures, and find that some random rotations lead to much better quantization than others, with an up to 13 points difference in downstream zero-shot reasoning performance. As a result, we propose SpinQuant that optimizes (or learns) the rotation matrices with Cayley optimization on a small validation set. With 4-bit quantization of weight, activation, and KV-cache, SpinQuant narrows the accuracy gap on zero-shot reasoning tasks with full precision to merely 2.9 points on the LLaMA-2 7B model, surpassing LLM-QAT by 19.1 points and SmoothQuant by 25.0 points. SpinQuant also outperforms concurrent work QuaRot, which applies random rotations to remove outliers. In particular, for LLaMA-2 7B/LLaMA-3 8B models that are hard to quantize, SpinQuant reduces the gap to full precision by 30.2%/34.1% relative to QuaRot.
Authors: Wei Qian, Aobo Chen, Chenxu Zhao, Yangyi Li, Mengdi Huai
Abstract: In education data mining (EDM) communities, machine learning has achieved remarkable success in discovering patterns and structures to tackle educational challenges. Notably, fairness and algorithmic bias have gained attention in learning analytics of EDM. With the increasing demand for the right to be forgotten, there is a growing need for machine learning models to forget sensitive data and its impact, particularly within the realm of EDM. The paradigm of selective forgetting, also known as machine unlearning, has been extensively studied to address this need by eliminating the influence of specific data from a pre-trained model without complete retraining. However, existing research assumes that interactive data removal operations are conducted in secure and reliable environments, neglecting potential malicious unlearning requests to undermine the fairness of machine learning systems. In this paper, we introduce a novel class of selective forgetting attacks designed to compromise the fairness of learning models while maintaining their predictive accuracy, thereby preventing the model owner from detecting the degradation in model performance. Additionally, we propose an innovative optimization framework for selective forgetting attacks, capable of generating malicious unlearning requests across various attack scenarios. We validate the effectiveness of our proposed selective forgetting attacks on fairness through extensive experiments using diverse EDM datasets.
Authors: Chenhao Lu, Ruizhe Shi, Yuyao Liu, Kaizhe Hu, Simon S. Du, Huazhe Xu
Abstract: Sequential decision-making algorithms such as reinforcement learning (RL) in real-world scenarios inevitably face environments with partial observability. This paper scrutinizes the effectiveness of a popular architecture, namely Transformers, in Partially Observable Markov Decision Processes (POMDPs) and reveals its theoretical limitations. We establish that regular languages, which Transformers struggle to model, are reducible to POMDPs. This poses a significant challenge for Transformers in learning POMDP-specific inductive biases, due to their lack of inherent recurrence found in other models like RNNs. This paper casts doubt on the prevalent belief in Transformers as sequence models for RL and proposes to introduce a point-wise recurrent structure. The Deep Linear Recurrent Unit (LRU) emerges as a well-suited alternative for Partially Observable RL, with empirical results highlighting the sub-optimal performance of the Transformer and considerable strength of LRU.
Authors: Hanlin Gu, WinKent Ong, Chee Seng Chan, Lixin Fan
Abstract: The advent of Federated Learning (FL) highlights the practical necessity for the 'right to be forgotten' for all clients, allowing them to request data deletion from the machine learning model's service provider. This necessity has spurred a growing demand for Federated Unlearning (FU). Feature unlearning has gained considerable attention due to its applications in unlearning sensitive features, backdoor features, and bias features. Existing methods employ the influence function to achieve feature unlearning, which is impractical for FL as it necessitates the participation of other clients in the unlearning process. Furthermore, current research lacks an evaluation of the effectiveness of feature unlearning. To address these limitations, we define feature sensitivity in the evaluation of feature unlearning according to Lipschitz continuity. This metric characterizes the rate of change or sensitivity of the model output to perturbations in the input feature. We then propose an effective federated feature unlearning framework called Ferrari, which minimizes feature sensitivity. Extensive experimental results and theoretical analysis demonstrate the effectiveness of Ferrari across various feature unlearning scenarios, including sensitive, backdoor, and biased features.
Authors: Sheng Yue, Xingyuan Hua, Lili Chen, Ju Ren
Abstract: Federated Reinforcement Learning (FRL) has garnered increasing attention recently. However, due to the intrinsic spatio-temporal non-stationarity of data distributions, the current approaches typically suffer from high interaction and communication costs. In this paper, we introduce a new FRL algorithm, named $\texttt{MFPO}$, that utilizes momentum, importance sampling, and additional server-side adjustment to control the shift of stochastic policy gradients and enhance the efficiency of data utilization. We prove that by proper selection of momentum parameters and interaction frequency, $\texttt{MFPO}$ can achieve $\tilde{\mathcal{O}}(H N^{-1}\epsilon^{-3/2})$ and $\tilde{\mathcal{O}}(\epsilon^{-1})$ interaction and communication complexities ($N$ represents the number of agents), where the interaction complexity achieves linear speedup with the number of agents, and the communication complexity aligns the best achievable of existing first-order FL algorithms. Extensive experiments corroborate the substantial performance gains of $\texttt{MFPO}$ over existing methods on a suite of complex and high-dimensional benchmarks.
Authors: Sheng Yue, Zerui Qin, Xingyuan Hua, Yongheng Deng, Ju Ren
Abstract: Federated Reinforcement Learning (FRL) has been deemed as a promising solution for intelligent decision-making in the era of Artificial Internet of Things. However, existing FRL approaches often entail repeated interactions with the environment during local updating, which can be prohibitively expensive or even infeasible in many real-world domains. To overcome this challenge, this paper proposes a novel offline federated policy optimization algorithm, named $\texttt{DRPO}$, which enables distributed agents to collaboratively learn a decision policy only from private and static data without further environmental interactions. $\texttt{DRPO}$ leverages dual regularization, incorporating both the local behavioral policy and the global aggregated policy, to judiciously cope with the intrinsic two-tier distributional shifts in offline FRL. Theoretical analysis characterizes the impact of the dual regularization on performance, demonstrating that by achieving the right balance thereof, $\texttt{DRPO}$ can effectively counteract distributional shifts and ensure strict policy improvement in each federative learning round. Extensive experiments validate the significant performance gains of $\texttt{DRPO}$ over baseline methods.
Authors: Sheng Yue, Jiani Liu, Xingyuan Hua, Ju Ren, Sen Lin, Junshan Zhang, Yaoxue Zhang
Abstract: Offline Imitation Learning (IL) with imperfect demonstrations has garnered increasing attention owing to the scarcity of expert data in many real-world domains. A fundamental problem in this scenario is how to extract positive behaviors from noisy data. In general, current approaches to the problem select data building on state-action similarity to given expert demonstrations, neglecting precious information in (potentially abundant) $\textit{diverse}$ state-actions that deviate from expert ones. In this paper, we introduce a simple yet effective data selection method that identifies positive behaviors based on their resultant states -- a more informative criterion enabling explicit utilization of dynamics information and effective extraction of both expert and beneficial diverse behaviors. Further, we devise a lightweight behavior cloning algorithm capable of leveraging the expert and selected data correctly. In the experiments, we evaluate our method on a suite of complex and high-dimensional offline IL benchmarks, including continuous-control and vision-based tasks. The results demonstrate that our method achieves state-of-the-art performance, outperforming existing methods on $\textbf{20/21}$ benchmarks, typically by $\textbf{2-5x}$, while maintaining a comparable runtime to Behavior Cloning ($\texttt{BC}$).
Authors: Sheng Yue, Xingyuan Hua, Ju Ren, Sen Lin, Junshan Zhang, Yaoxue Zhang
Abstract: In this paper, we study offline-to-online Imitation Learning (IL) that pretrains an imitation policy from static demonstration data, followed by fast finetuning with minimal environmental interaction. We find the na\"ive combination of existing offline IL and online IL methods tends to behave poorly in this context, because the initial discriminator (often used in online IL) operates randomly and discordantly against the policy initialization, leading to misguided policy optimization and $\textit{unlearning}$ of pretraining knowledge. To overcome this challenge, we propose a principled offline-to-online IL method, named $\texttt{OLLIE}$, that simultaneously learns a near-expert policy initialization along with an $\textit{aligned discriminator initialization}$, which can be seamlessly integrated into online IL, achieving smooth and fast finetuning. Empirically, $\texttt{OLLIE}$ consistently and significantly outperforms the baseline methods in $\textbf{20}$ challenging tasks, from continuous control to vision-based domains, in terms of performance, demonstration efficiency, and convergence speed. This work may serve as a foundation for further exploration of pretraining and finetuning in the context of IL.
Authors: Ju-Seung Byun, Andrew Perrault
Abstract: Reinforcement learning (RL) training is inherently unstable due to factors such as moving targets and high gradient variance. Reinforcement Learning from Human Feedback (RLHF) and Reinforcement Learning from AI Feedback (RLAIF) can introduce additional difficulty. Differing preferences can complicate the alignment process, and prediction errors in a trained reward model can become more severe as the LLM generates unseen outputs. To enhance training robustness, RL has adopted techniques from supervised learning, such as ensembles and layer normalization. In this work, we improve the stability of RL training by adapting the reverse cross entropy (RCE) from supervised learning for noisy data to define a symmetric RL loss. We demonstrate performance improvements across various tasks and scales. We conduct experiments in discrete action tasks (Atari games) and continuous action space tasks (MuJoCo benchmark and Box2D) using Symmetric A2C (SA2C) and Symmetric PPO (SPPO), with and without added noise with especially notable performance in SPPO across different hyperparameters. Furthermore, we validate the benefits of the symmetric RL loss when using SPPO for large language models through improved performance in RLHF tasks, such as IMDB positive sentiment sentiment and TL;DR summarization tasks.
Authors: Youlong Ding, Xueyang Wu, Yining Meng, Yonggang Luo, Hao Wang, Weike Pan
Abstract: Deep learning with differential privacy (DP) has garnered significant attention over the past years, leading to the development of numerous methods aimed at enhancing model accuracy and training efficiency. This paper delves into the problem of training Transformer models with differential privacy. Our treatment is modular: the logic is to `reduce' the problem of training DP Transformer to the more basic problem of training DP vanilla neural nets. The latter is better understood and amenable to many model-agnostic methods. Such `reduction' is done by first identifying the hardness unique to DP Transformer training: the attention distraction phenomenon and a lack of compatibility with existing techniques for efficient gradient clipping. To deal with these two issues, we propose the Re-Attention Mechanism and Phantom Clipping, respectively. We believe that our work not only casts new light on training DP Transformers but also promotes a modular treatment to advance research in the field of differentially private deep learning.
Authors: George Ma, Yifei Wang, Derek Lim, Stefanie Jegelka, Yisen Wang
Abstract: In many applications, we desire neural networks to exhibit invariance or equivariance to certain groups due to symmetries inherent in the data. Recently, frame-averaging methods emerged to be a unified framework for attaining symmetries efficiently by averaging over input-dependent subsets of the group, i.e., frames. What we currently lack is a principled understanding of the design of frames. In this work, we introduce a canonization perspective that provides an essential and complete view of the design of frames. Canonization is a classic approach for attaining invariance by mapping inputs to their canonical forms. We show that there exists an inherent connection between frames and canonical forms. Leveraging this connection, we can efficiently compare the complexity of frames as well as determine the optimality of certain frames. Guided by this principle, we design novel frames for eigenvectors that are strictly superior to existing methods -- some are even optimal -- both theoretically and empirically. The reduction to the canonization perspective further uncovers equivalences between previous methods. These observations suggest that canonization provides a fundamental understanding of existing frame-averaging methods and unifies existing equivariant and invariant learning methods.
Authors: Alexander H\"agele, Elie Bakouch, Atli Kosson, Loubna Ben Allal, Leandro Von Werra, Martin Jaggi
Abstract: Scale has become a main ingredient in obtaining strong machine learning models. As a result, understanding a model's scaling properties is key to effectively designing both the right training setup as well as future generations of architectures. In this work, we argue that scale and training research has been needlessly complex due to reliance on the cosine schedule, which prevents training across different lengths for the same model size. We investigate the training behavior of a direct alternative - constant learning rate and cooldowns - and find that it scales predictably and reliably similar to cosine. Additionally, we show that stochastic weight averaging yields improved performance along the training trajectory, without additional training costs, across different scales. Importantly, with these findings we demonstrate that scaling experiments can be performed with significantly reduced compute and GPU hours by utilizing fewer but reusable training runs. Our code is available at https://github.com/epfml/schedules-and-scaling.
Authors: JinZhe Jiang, Xin Zhang, Chen Li, YaQian Zhao, RenGang Li
Abstract: Generalization is an important feature of neural network, and there have been many studies on it. Recently, with the development of quantum compu-ting, it brings new opportunities. In this paper, we studied a class of quantum neural network constructed by quantum gate. In this model, we mapped the feature data to a quantum state in Hilbert space firstly, and then implement unitary evolution on it, in the end, we can get the classification result by im-plement measurement on the quantum state. Since all the operations in quan-tum neural networks are unitary, the parameters constitute a hypersphere of Hilbert space. Compared with traditional neural network, the parameter space is flatter. Therefore, it is not easy to fall into local optimum, which means the quantum neural networks have better generalization. In order to validate our proposal, we evaluated our model on three public datasets, the results demonstrated that our model has better generalization than the classical neu-ral network with the same structure.
Authors: Kwonjoon Lee, Huiwen Chang, Lu Jiang, Han Zhang, Zhuowen Tu, Ce Liu
Abstract: Recently, Vision Transformers (ViTs) have shown competitive performance on image recognition while requiring less vision-specific inductive biases. In this paper, we investigate if such performance can be extended to image generation. To this end, we integrate the ViT architecture into generative adversarial networks (GANs). For ViT discriminators, we observe that existing regularization methods for GANs interact poorly with self-attention, causing serious instability during training. To resolve this issue, we introduce several novel regularization techniques for training GANs with ViTs. For ViT generators, we examine architectural choices for latent and pixel mapping layers to facilitate convergence. Empirically, our approach, named ViTGAN, achieves comparable performance to the leading CNN-based GAN models on three datasets: CIFAR-10, CelebA, and LSUN bedroom.
Authors: Kartheek Bondugula, Ver\'onica \'Alvarez, Jos\'e I. Segovia-Mart\'in, Aritz P\'erez, Santiago Mazuelas
Abstract: Libraries for supervised classification have enabled the wide-spread usage of machine learning methods. Existing libraries, such as scikit-learn, caret, and mlpack, implement techniques based on the classical empirical risk minimization (ERM) approach. We present a Python library, MRCpy, that implements minimax risk classifiers (MRCs) based on the robust risk minimization (RRM) approach. The library offers multiple variants of MRCs that can provide performance guarantees, enable efficient learning in high dimensions, and adapt to distribution shifts. MRCpy follows an object-oriented approach and adheres to the standards of popular Python libraries, such as scikit-learn, facilitating readability and easy usage together with a seamless integration with other libraries. The source code is available under the GPL-3.0 license at https://github.com/MachineLearningBCAM/MRCpy.
Authors: Fei Zheng, Chaochao Chen, Lingjuan Lyu, Xinyi Fu, Xing Fu, Weiqiang Wang, Xiaolin Zheng, Jianwei Yin
Abstract: As a practical privacy-preserving learning method, split learning has drawn much attention in academia and industry. However, its security is constantly being questioned since the intermediate results are shared during training and inference. In this paper, we focus on the privacy leakage from the forward embeddings of split learning. Specifically, since the forward embeddings contain too much information about the label, the attacker can either use a few labeled samples to fine-tune the top model or perform unsupervised attacks such as clustering to infer the true labels from the forward embeddings. To prevent such kind of privacy leakage, we propose the potential energy loss to make the forward embeddings become more 'complicated', by pushing embeddings of the same class towards the decision boundary. Therefore, it is hard for the attacker to learn from the forward embeddings. Experiment results show that our method significantly lowers the performance of both fine-tuning attacks and clustering attacks.
Authors: Yuantong Li, Guang Cheng, Xiaowu Dai
Abstract: In this paper, we propose a new recommendation algorithm for addressing the problem of two-sided online matching markets with complementary preferences and quota constraints, where agents' preferences are unknown a priori and must be learned from data. The presence of mixed quota and complementary preferences constraints can lead to instability in the matching process, making this problem challenging to solve. To overcome this challenge, we formulate the problem as a bandit learning framework and propose the Multi-agent Multi-type Thompson Sampling (MMTS) algorithm. The algorithm combines the strengths of Thompson Sampling for exploration with a new double matching technique to provide a stable matching outcome. Our theoretical analysis demonstrates the effectiveness of MMTS as it can achieve stability and has a total $\widetilde{\mathcal{O}}(Q{\sqrt{K_{\max}T}})$-Bayesian regret with high probability, which exhibits linearity with respect to the total firm's quota $Q$, the square root of the maximum size of available type workers $\sqrt{K_{\max}}$ and time horizon $T$. In addition, simulation studies also demonstrate MMTS's effectiveness in various settings. We provide code used in our experiments \url{https://github.com/Likelyt/Double-Matching}.
Authors: Ren\'e Haas, Inbar Huberman-Spiegelglas, Rotem Mulayoff, Stella Gra{\ss}hof, Sami S. Brandt, Tomer Michaeli
Abstract: Denoising Diffusion Models (DDMs) have emerged as a strong competitor to Generative Adversarial Networks (GANs). However, despite their widespread use in image synthesis and editing applications, their latent space is still not as well understood. Recently, a semantic latent space for DDMs, coined `$h$-space', was shown to facilitate semantic image editing in a way reminiscent of GANs. The $h$-space is comprised of the bottleneck activations in the DDM's denoiser across all timesteps of the diffusion process. In this paper, we explore the properties of h-space and propose several novel methods for finding meaningful semantic directions within it. We start by studying unsupervised methods for revealing interpretable semantic directions in pretrained DDMs. Specifically, we show that global latent directions emerge as the principal components in the latent space. Additionally, we provide a novel method for discovering image-specific semantic directions by spectral analysis of the Jacobian of the denoiser w.r.t. the latent code. Next, we extend the analysis by finding directions in a supervised fashion in unconditional DDMs. We demonstrate how such directions can be found by relying on either a labeled data set of real images or by annotating generated samples with a domain-specific attribute classifier. We further show how to semantically disentangle the found direction by simple linear projection. Our approaches are applicable without requiring any architectural modifications, text-based guidance, CLIP-based optimization, or model fine-tuning.
Authors: Songwei Ge, Taesung Park, Jun-Yan Zhu, Jia-Bin Huang
Abstract: Plain text has become a prevalent interface for text-to-image synthesis. However, its limited customization options hinder users from accurately describing desired outputs. For example, plain text makes it hard to specify continuous quantities, such as the precise RGB color value or importance of each word. Furthermore, creating detailed text prompts for complex scenes is tedious for humans to write and challenging for text encoders to interpret. To address these challenges, we propose using a rich-text editor supporting formats such as font style, size, color, and footnote. We extract each word's attributes from rich text to enable local style control, explicit token reweighting, precise color rendering, and detailed region synthesis. We achieve these capabilities through a region-based diffusion process. We first obtain each word's region based on attention maps of a diffusion process using plain text. For each region, we enforce its text attributes by creating region-specific detailed prompts and applying region-specific guidance, and maintain its fidelity against plain-text generation through region-based injections. We present various examples of image generation from rich text and demonstrate that our method outperforms strong baselines with quantitative evaluations.
Authors: Wenyue Hua, Yingqiang Ge, Shuyuan Xu, Jianchao Ji, Yongfeng Zhang
Abstract: Recent advances in Foundation Models such as Large Language Models (LLMs) have propelled them to the forefront of Recommender Systems (RS). Despite their utility, there is a growing concern that LLMs might inadvertently perpetuate societal stereotypes, resulting in unfair recommendations. Since fairness is critical for RS as many users take it for decision-making and demand fulfillment, this paper focuses on user-side fairness for LLM-based recommendation where the users may require a recommender system to be fair on specific sensitive features such as gender or age. In this paper, we dive into the extent of unfairness exhibited by LLM-based recommender models based on both T5 and LLaMA backbones, and discuss appropriate methods for promoting equitable treatment of users in LLM-based recommendation models. We introduce a novel Counterfactually-Fair-Prompt (CFP) method towards Unbiased Foundation mOdels (UFO) for fairness-aware LLM-based recommendation. Experiments are conducted on two real-world datasets, MovieLens-1M and Insurance, and compared with both matching-based and sequential-based fairness-aware recommendation models. Results show that CFP achieves better recommendation performance with a high level of fairness. Data and code are open-sourced at https://github.com/agiresearch/UP5.
Authors: Tim Tsz-Kit Lau, Han Liu, Thomas Pock
Abstract: We study the problem of approximate sampling from non-log-concave distributions, e.g., Gaussian mixtures, which is often challenging even in low dimensions due to their multimodality. We focus on performing this task via Markov chain Monte Carlo (MCMC) methods derived from discretizations of the overdamped Langevin diffusions, which are commonly known as Langevin Monte Carlo algorithms. Furthermore, we are also interested in two nonsmooth cases for which a large class of proximal MCMC methods have been developed: (i) a nonsmooth prior is considered with a Gaussian mixture likelihood; (ii) a Laplacian mixture distribution. Such nonsmooth and non-log-concave sampling tasks arise from a wide range of applications to Bayesian inference and imaging inverse problems such as image deconvolution. We perform numerical simulations to compare the performance of most commonly used Langevin Monte Carlo algorithms.
Authors: Kouroche Bouchiat, Alexander Immer, Hugo Y\`eche, Gunnar R\"atsch, Vincent Fortuin
Abstract: Neural additive models (NAMs) enhance the transparency of deep neural networks by handling input features in separate additive sub-networks. However, they lack inherent mechanisms that provide calibrated uncertainties and enable selection of relevant features and interactions. Approaching NAMs from a Bayesian perspective, we augment them in three primary ways, namely by a) providing credible intervals for the individual additive sub-networks; b) estimating the marginal likelihood to perform an implicit selection of features via an empirical Bayes procedure; and c) facilitating the ranking of feature pairs as candidates for second-order interaction in fine-tuned models. In particular, we develop Laplace-approximated NAMs (LA-NAMs), which show improved empirical performance on tabular datasets and challenging real-world medical tasks.
Authors: Filip Lemic, Gerard Calvo Bartra, Arnau Brosa L\'opez, Jorge Torres G\'omez, Jakob Struye, Falko Dressler, Sergi Abadal, Xavier Costa Perez
Abstract: Nanodevices with Terahertz (THz)-based wireless communication capabilities are providing a primer for flow-guided localization within the human bloodstreams. Such localization is allowing for assigning the locations of sensed events with the events themselves, providing benefits in precision medicine along the lines of early and precise diagnostics, and reduced costs and invasiveness. Flow-guided localization is still in a rudimentary phase, with only a handful of works targeting the problem. Nonetheless, the performance assessments of the proposed solutions are already carried out in a non-standardized way, usually along a single performance metric, and ignoring various aspects that are relevant at such a scale (e.g., nanodevices' limited energy) and for such a challenging environment (e.g., extreme attenuation of in-body THz propagation). As such, these assessments feature low levels of realism and cannot be compared in an objective way. Toward addressing this issue, we account for the environmental and scale-related peculiarities of the scenario and assess the performance of two state-of-the-art flow-guided localization approaches along a set of heterogeneous performance metrics such as the accuracy and reliability of localization.
Authors: Hubert Siuzdak
Abstract: Recent advancements in neural vocoding are predominantly driven by Generative Adversarial Networks (GANs) operating in the time-domain. While effective, this approach neglects the inductive bias offered by time-frequency representations, resulting in reduntant and computionally-intensive upsampling operations. Fourier-based time-frequency representation is an appealing alternative, aligning more accurately with human auditory perception, and benefitting from well-established fast algorithms for its computation. Nevertheless, direct reconstruction of complex-valued spectrograms has been historically problematic, primarily due to phase recovery issues. This study seeks to close this gap by presenting Vocos, a new model that directly generates Fourier spectral coefficients. Vocos not only matches the state-of-the-art in audio quality, as demonstrated in our evaluations, but it also substantially improves computational efficiency, achieving an order of magnitude increase in speed compared to prevailing time-domain neural vocoding approaches. The source code and model weights have been open-sourced at https://github.com/gemelo-ai/vocos.
Authors: Snehal Jauhri, Ishikaa Lunawat, Georgia Chalvatzaki
Abstract: A significant challenge for real-world robotic manipulation is the effective 6DoF grasping of objects in cluttered scenes from any single viewpoint without the need for additional scene exploration. This work reinterprets grasping as rendering and introduces NeuGraspNet, a novel method for 6DoF grasp detection that leverages advances in neural volumetric representations and surface rendering. It encodes the interaction between a robot's end-effector and an object's surface by jointly learning to render the local object surface and learning grasping functions in a shared feature space. The approach uses global (scene-level) features for grasp generation and local (grasp-level) neural surface features for grasp evaluation. This enables effective, fully implicit 6DoF grasp quality prediction, even in partially observed scenes. NeuGraspNet operates on random viewpoints, common in mobile manipulation scenarios, and outperforms existing implicit and semi-implicit grasping methods. The real-world applicability of the method has been demonstrated with a mobile manipulator robot, grasping in open, cluttered spaces. Project website at https://sites.google.com/view/neugraspnet
Authors: Chandramouli Sastry, Sri Harsha Dumpala, Sageev Oore
Abstract: We introduce DiffAug, a simple and efficient diffusion-based augmentation technique to train image classifiers for the crucial yet challenging goal of improved classifier robustness. Applying DiffAug to a given example consists of one forward-diffusion step followed by one reverse-diffusion step. Using both ResNet-50 and Vision Transformer architectures, we comprehensively evaluate classifiers trained with DiffAug and demonstrate the surprising effectiveness of single-step reverse diffusion in improving robustness to covariate shifts, certified adversarial accuracy and out of distribution detection. When we combine DiffAug with other augmentations such as AugMix and DeepAugment we demonstrate further improved robustness. Finally, building on this approach, we also improve classifier-guided diffusion wherein we observe improvements in: (i) classifier-generalization, (ii) gradient quality (i.e., improved perceptual alignment) and (iii) image generation performance. We thus introduce a computationally efficient technique for training with improved robustness that does not require any additional data, and effectively complements existing augmentation approaches.
Authors: Jinhao Duan, Hao Cheng, Shiqi Wang, Alex Zavalny, Chenan Wang, Renjing Xu, Bhavya Kailkhura, Kaidi Xu
Abstract: Large Language Models (LLMs) show promising results in language generation and instruction following but frequently "hallucinate", making their outputs less reliable. Despite Uncertainty Quantification's (UQ) potential solutions, implementing it accurately within LLMs is challenging. Our research introduces a simple heuristic: not all tokens in auto-regressive LLM text equally represent the underlying meaning, as "linguistic redundancy" often allows a few keywords to convey the essence of long sentences. However, current methods underestimate this inequality when assessing uncertainty, causing tokens with limited semantics to be equally or excessively weighted in UQ. To correct this, we propose Shifting Attention to more Relevant (SAR) components at both token- and sentence-levels for better UQ. We conduct extensive experiments involving a range of popular "off-the-shelf" LLMs, such as Vicuna, WizardLM, and LLaMA-2-chat, with model sizes extending up to 33B parameters. We evaluate various free-form question-answering tasks, encompassing domains such as reading comprehension, science Q&A, and medical Q&A. Our experimental results, coupled with a comprehensive demographic analysis, demonstrate the superior performance of SAR. The code is available at https://github.com/jinhaoduan/SAR.
Authors: Christophe Bonneville, Youngsoo Choi, Debojyoti Ghosh, Jonathan L. Belof
Abstract: Numerically solving partial differential equations (PDEs) can be challenging and computationally expensive. This has led to the development of reduced-order models (ROMs) that are accurate but faster than full order models (FOMs). Recently, machine learning advances have enabled the creation of non-linear projection methods, such as Latent Space Dynamics Identification (LaSDI). LaSDI maps full-order PDE solutions to a latent space using autoencoders and learns the system of ODEs governing the latent space dynamics. By interpolating and solving the ODE system in the reduced latent space, fast and accurate ROM predictions can be made by feeding the predicted latent space dynamics into the decoder. In this paper, we introduce GPLaSDI, a novel LaSDI-based framework that relies on Gaussian process (GP) for latent space ODE interpolations. Using GPs offers two significant advantages. First, it enables the quantification of uncertainty over the ROM predictions. Second, leveraging this prediction uncertainty allows for efficient adaptive training through a greedy selection of additional training data points. This approach does not require prior knowledge of the underlying PDEs. Consequently, GPLaSDI is inherently non-intrusive and can be applied to problems without a known PDE or its residual. We demonstrate the effectiveness of our approach on the Burgers equation, Vlasov equation for plasma physics, and a rising thermal bubble problem. Our proposed method achieves between 200 and 100,000 times speed-up, with up to 7% relative error.
Authors: Wenjun Jiang, Tianlong Fan, Changhao Li, Chuanfu Zhang, Tao Zhang, Zong-fu Luo
Abstract: Connectivity robustness, a crucial aspect for understanding, optimizing, and repairing complex networks, has traditionally been evaluated through time-consuming and often impractical simulations. Fortunately, machine learning provides a new avenue for addressing this challenge. However, several key issues remain unresolved, including the performance in more general edge removal scenarios, capturing robustness through attack curves instead of directly training for robustness, scalability of predictive tasks, and transferability of predictive capabilities. In this paper, we address these challenges by designing a convolutional neural networks (CNN) model with spatial pyramid pooling networks (SPP-net), adapting existing evaluation metrics, redesigning the attack modes, introducing appropriate filtering rules, and incorporating the value of robustness as training data. The results demonstrate the thoroughness of the proposed CNN framework in addressing the challenges of high computational time across various network types, failure component types and failure scenarios. However, the performance of the proposed CNN model varies: for evaluation tasks that are consistent with the trained network type, the proposed CNN model consistently achieves accurate evaluations of both attack curves and robustness values across all removal scenarios. When the predicted network type differs from the trained network, the CNN model still demonstrates favorable performance in the scenario of random node failure, showcasing its scalability and performance transferability. Nevertheless, the performance falls short of expectations in other removal scenarios. This observed scenario-sensitivity in the evaluation of network features has been overlooked in previous studies and necessitates further attention and optimization. Lastly, we discuss important unresolved questions and further investigation.
Authors: Jiyuan Tan, Chenyu Xue, Chuwen Zhang, Qi Deng, Dongdong Ge, Yinyu Ye
Abstract: Gradient dominance property is a condition weaker than strong convexity, yet sufficiently ensures global convergence even in non-convex optimization. This property finds wide applications in machine learning, reinforcement learning (RL), and operations management. In this paper, we propose the stochastic homogeneous second-order descent method (SHSODM) for stochastic functions enjoying gradient dominance property based on a recently proposed homogenization approach. Theoretically, we provide its sample complexity analysis, and further present an enhanced result by incorporating variance reduction techniques. Our findings show that SHSODM matches the best-known sample complexity achieved by other second-order methods for gradient-dominated stochastic optimization but without cubic regularization. Empirically, since the homogenization approach only relies on solving extremal eigenvector problem at each iteration instead of Newton-type system, our methods gain the advantage of cheaper computational cost and robustness in ill-conditioned problems. Numerical experiments on several RL tasks demonstrate the better performance of SHSODM compared to other off-the-shelf methods.
Authors: Philipp Scholl, Katharina Bieker, Hillary Hauger, Gitta Kutyniok
Abstract: The problem of symbolic regression (SR) arises in many different applications, such as identifying physical laws or deriving mathematical equations describing the behavior of financial markets from given data. Various methods exist to address the problem of SR, often based on genetic programming. However, these methods are usually complicated and involve various hyperparameters. In this paper, we present our new approach ParFam that utilizes parametric families of suitable symbolic functions to translate the discrete symbolic regression problem into a continuous one, resulting in a more straightforward setup compared to current state-of-the-art methods. In combination with a global optimizer, this approach results in a highly effective method to tackle the problem of SR. We theoretically analyze the expressivity of ParFam and demonstrate its performance with extensive numerical experiments based on the common SR benchmark suit SRBench, showing that we achieve state-of-the-art results. Moreover, we present an extension incorporating a pre-trained transformer network DL-ParFam to guide ParFam, accelerating the optimization process by up to two magnitudes. Our code and results can be found at https://github.com/Philipp238/parfam.
Authors: Boxin Wang, Wei Ping, Lawrence McAfee, Peng Xu, Bo Li, Mohammad Shoeybi, Bryan Catanzaro
Abstract: Pretraining auto-regressive large language models~(LLMs) with retrieval demonstrates better perplexity and factual accuracy by leveraging external databases. However, the size of existing pretrained retrieval-augmented LLM is still limited (e.g., Retro has 7.5B parameters), which limits the effectiveness of instruction tuning and zero-shot generalization. In this work, we introduce Retro 48B, the largest LLM pretrained with retrieval. Specifically, we continue to pretrain a 43B GPT model on additional 100 billion tokens using the Retro augmentation method by retrieving from 1.2 trillion tokens. Notably, the obtained foundation model, Retro 48B, largely outperforms the counterpart GPT 43B trained on 1.2T tokens in terms of perplexity with only 2.58% additional GPU hours, demonstrating the significant scaling potential of the method. After instruction tuning on Retro, InstructRetro demonstrates significant improvement over the instruction tuned GPT on a wide range of zero-shot tasks. Specifically, the average improvement of InstructRetro is 7% over its GPT counterpart across 8 short-form QA and reading comprehension tasks, 10% over GPT across 4 challenging long-form QA tasks, and 16% over GPT across 3 summarization tasks. Surprisingly, we find that one can ablate the encoder from InstructRetro architecture and directly use its decoder backbone, while achieving comparable results. Our results highlight the promising direction to obtain a better GPT decoder through continued pretraining with retrieval before instruction tuning. Our code and checkpoints are publicly available at: https://huggingface.co/nvidia/retro-48b-instruct-4k.
Authors: Pawe{\l} Czy\.z, Frederic Grabowski, Julia E. Vogt, Niko Beerenwinkel, Alexander Marx
Abstract: The pointwise mutual information profile, or simply profile, is the distribution of pointwise mutual information for a given pair of random variables. One of its important properties is that its expected value is precisely the mutual information between these random variables. In this paper, we analytically describe the profiles of multivariate normal distributions and introduce a novel family of distributions, Bend and Mix Models, for which the profile can be accurately estimated using Monte Carlo methods. We then show how Bend and Mix Models can be used to study the limitations of existing mutual information estimators, investigate the behavior of neural critics used in variational estimators, and understand the effect of experimental outliers on mutual information estimation. Finally, we show how Bend and Mix Models can be used to obtain model-based Bayesian estimates of mutual information, suitable for problems with available domain expertise in which uncertainty quantification is necessary.
Authors: Danny Wood, Theodore Papamarkou, Matt Benatan, Richard Allmendinger
Abstract: In order to trust the predictions of a machine learning algorithm, it is necessary to understand the factors that contribute to those predictions. In the case of probabilistic and uncertainty-aware models, it is necessary to understand not only the reasons for the predictions themselves, but also the reasons for the model's level of confidence in those predictions. In this paper, we show how existing methods in explainability can be extended to uncertainty-aware models and how such extensions can be used to understand the sources of uncertainty in a model's predictive distribution. In particular, by adapting permutation feature importance, partial dependence plots, and individual conditional expectation plots, we demonstrate that novel insights into model behaviour may be obtained and that these methods can be used to measure the impact of features on both the entropy of the predictive distribution and the log-likelihood of the ground truth labels under that distribution. With experiments using both synthetic and real-world data, we demonstrate the utility of these approaches to understand both the sources of uncertainty and their impact on model performance.
Authors: Chunyi Sun, Junlin Han, Weijian Deng, Xinlong Wang, Zishan Qin, Stephen Gould
Abstract: In the pursuit of efficient automated content creation, procedural generation, leveraging modifiable parameters and rule-based systems, emerges as a promising approach. Nonetheless, it could be a demanding endeavor, given its intricate nature necessitating a deep understanding of rules, algorithms, and parameters. To reduce workload, we introduce 3D-GPT, a framework utilizing large language models~(LLMs) for instruction-driven 3D modeling. 3D-GPT positions LLMs as proficient problem solvers, dissecting the procedural 3D modeling tasks into accessible segments and appointing the apt agent for each task. 3D-GPT integrates three core agents: the task dispatch agent, the conceptualization agent, and the modeling agent. They collaboratively achieve two objectives. First, it enhances concise initial scene descriptions, evolving them into detailed forms while dynamically adapting the text based on subsequent instructions. Second, it integrates procedural generation, extracting parameter values from enriched text to effortlessly interface with 3D software for asset creation. Our empirical investigations confirm that 3D-GPT not only interprets and executes instructions, delivering reliable results but also collaborates effectively with human designers. Furthermore, it seamlessly integrates with Blender, unlocking expanded manipulation possibilities. Our work highlights the potential of LLMs in 3D modeling, offering a basic framework for future advancements in scene generation and animation.
Authors: Runze Cheng, Yao Sun, Dusit Niyato, Lan Zhang, Lei Zhang, Muhammad Ali Imran
Abstract: Generative AI applications have been recently catering to a vast user base by creating diverse and high-quality AI-generated content (AIGC). With the proliferation of mobile devices and rapid growth of mobile traffic, providing ubiquitous access to high-quality AIGC services via wireless communication networks is becoming the future direction. However, it is challenging to provide qualified AIGC services in wireless networks with unstable channels, limited bandwidth resources, and unevenly distributed computational resources. To tackle these challenges, we propose a semantic communication (SemCom)-empowered AIGC (SemAIGC) generation and transmission framework, where only semantic information of the content rather than all the binary bits should be generated and transmitted by using SemCom. Specifically, SemAIGC integrates diffusion models within the semantic encoder and decoder to design a workload-adjustable transceiver thereby allowing adjustment of computational resource utilization in edge and local. In addition, a Resource-aware wOrk lOad Trade-off (ROOT) scheme is devised to intelligently make workload adaptation decisions for the transceiver, thus efficiently generating, transmitting, and fine-tuning content as per dynamic wireless channel conditions and service requirements. Simulations verify the superiority of our proposed SemAIGC framework in terms of latency and content quality compared to conventional approaches.
Authors: Jiacheng Li, Ninghui Li, Bruno Ribeiro
Abstract: In Member Inference (MI) attacks, the adversary try to determine whether an instance is used to train a machine learning (ML) model. MI attacks are a major privacy concern when using private data to train ML models. Most MI attacks in the literature take advantage of the fact that ML models are trained to fit the training data well, and thus have very low loss on training instances. Most defenses against MI attacks therefore try to make the model fit the training data less well. Doing so, however, generally results in lower accuracy. We observe that training instances have different degrees of vulnerability to MI attacks. Most instances will have low loss even when not included in training. For these instances, the model can fit them well without concerns of MI attacks. An effective defense only needs to (possibly implicitly) identify instances that are vulnerable to MI attacks and avoids overfitting them. A major challenge is how to achieve such an effect in an efficient training process. Leveraging two distinct recent advancements in representation learning: counterfactually-invariant representations and subspace learning methods, we introduce a novel Membership-Invariant Subspace Training (MIST) method to defend against MI attacks. MIST avoids overfitting the vulnerable instances without significant impact on other instances. We have conducted extensive experimental studies, comparing MIST with various other state-of-the-art (SOTA) MI defenses against several SOTA MI attacks. We find that MIST outperforms other defenses while resulting in minimal reduction in testing accuracy.
Authors: Ravil Mussabayev, Rustam Mussabayev
Abstract: This paper introduces a novel formulation of the clustering problem, namely the Minimum Sum-of-Squares Clustering of Infinitely Tall Data (MSSC-ITD), and presents HPClust, an innovative set of hybrid parallel approaches for its effective solution. By utilizing modern high-performance computing techniques, HPClust enhances key clustering metrics: effectiveness, computational efficiency, and scalability. In contrast to vanilla data parallelism, which only accelerates processing time through the MapReduce framework, our approach unlocks superior performance by leveraging the multi-strategy competitive-cooperative parallelism and intricate properties of the objective function landscape. Unlike other available algorithms that struggle to scale, our algorithm is inherently parallel in nature, improving solution quality through increased scalability and parallelism, and outperforming even advanced algorithms designed for small and medium-sized datasets. Our evaluation of HPClust, featuring four parallel strategies, demonstrates its superiority over traditional and cutting-edge methods by offering better performance in the key metrics. These results also show that parallel processing not only enhances the clustering efficiency, but the accuracy as well. Additionally, we explore the balance between computational efficiency and clustering quality, providing insights into optimal parallel strategies based on dataset specifics and resource availability. This research advances our understanding of parallelism in clustering algorithms, demonstrating that a judicious hybridization of advanced parallel approaches yields optimal results for MSSC-ITD. Experiments on synthetic data further confirm HPClust's exceptional scalability and robustness to noise.
Authors: Yongqi Dong, Xingmin Lu, Ruohan Li, Wei Song, Bart van Arem, Haneen Farah
Abstract: The burgeoning navigation services using digital maps provide great convenience to drivers. Nevertheless, the presence of anomalies in lane rendering map images occasionally introduces potential hazards, as such anomalies can be misleading to human drivers and consequently contribute to unsafe driving conditions. In response to this concern and to accurately and effectively detect the anomalies, this paper transforms lane rendering image anomaly detection into a classification problem and proposes a four-phase pipeline consisting of data pre-processing, self-supervised pre-training with the masked image modeling (MiM) method, customized fine-tuning using cross-entropy based loss with label smoothing, and post-processing to tackle it leveraging state-of-the-art deep learning techniques, especially those involving Transformer models. Various experiments verify the effectiveness of the proposed pipeline. Results indicate that the proposed pipeline exhibits superior performance in lane rendering image anomaly detection, and notably, the self-supervised pre-training with MiM can greatly enhance the detection accuracy while significantly reducing the total training time. For instance, employing the Swin Transformer with Uniform Masking as self-supervised pretraining (Swin-Trans-UM) yielded a heightened accuracy at 94.77% and an improved Area Under The Curve (AUC) score of 0.9743 compared with the pure Swin Transformer without pre-training (Swin-Trans) with an accuracy of 94.01% and an AUC of 0.9498. The fine-tuning epochs were dramatically reduced to 41 from the original 280. In conclusion, the proposed pipeline, with its incorporation of self-supervised pre-training using MiM and other advanced deep learning techniques, emerges as a robust solution for enhancing the accuracy and efficiency of lane rendering image anomaly detection in digital navigation systems.
Authors: Thomas W. Mitchel, Carlos Esteves, Ameesh Makadia
Abstract: We introduce a framework for intrinsic latent diffusion models operating directly on the surfaces of 3D shapes, with the goal of synthesizing high-quality textures. Our approach is underpinned by two contributions: field latents, a latent representation encoding textures as discrete vector fields on the mesh vertices, and field latent diffusion models, which learn to denoise a diffusion process in the learned latent space on the surface. We consider a single-textured-mesh paradigm, where our models are trained to generate variations of a given texture on a mesh. We show the synthesized textures are of superior fidelity compared those from existing single-textured-mesh generative models. Our models can also be adapted for user-controlled editing tasks such as inpainting and label-guided generation. The efficacy of our approach is due in part to the equivariance of our proposed framework under isometries, allowing our models to seamlessly reproduce details across locally similar regions and opening the door to a notion of generative texture transfer.
Authors: Majd Al Aawar, Srikar Mutnuri, Mansooreh Montazerin, Ajitesh Srivastava
Abstract: During the COVID-19 pandemic, a major driver of new surges has been the emergence of new variants. When a new variant emerges in one or more countries, other nations monitor its spread in preparation for its potential arrival. The impact of the new variant and the timings of epidemic peaks in a country highly depend on when the variant arrives. The current methods for predicting the spread of new variants rely on statistical modeling, however, these methods work only when the new variant has already arrived in the region of interest and has a significant prevalence. Can we predict when a variant existing elsewhere will arrive in a given region? To address this question, we propose a variant-dynamics-informed Graph Neural Network (GNN) approach. First, we derive the dynamics of variant prevalence across pairs of regions (countries) that apply to a large class of epidemic models. The dynamics motivate the introduction of certain features in the GNN. We demonstrate that our proposed dynamics-informed GNN outperforms all the baselines, including the currently pervasive framework of Physics-Informed Neural Networks (PINNs). To advance research in this area, we introduce a benchmarking tool to assess a user-defined model's prediction performance across 87 countries and 36 variants.
Authors: Konstantin A. Maslov, Claudio Persello, Thomas Schellenberger, Alfred Stein
Abstract: Accurate global glacier mapping is critical for understanding climate change impacts. Despite its importance, automated glacier mapping at a global scale remains largely unexplored. Here we address this gap and propose Glacier-VisionTransformer-U-Net (GlaViTU), a convolutional-transformer deep learning model, and five strategies for multitemporal global-scale glacier mapping using open satellite imagery. Assessing the spatial, temporal and cross-sensor generalisation shows that our best strategy achieves intersection over union >0.85 on previously unobserved images in most cases, which drops to >0.75 for debris-rich areas such as High-Mountain Asia and increases to >0.90 for regions dominated by clean ice. A comparative validation against human expert uncertainties in terms of area and distance deviations underscores GlaViTU performance, approaching or matching expert-level delineation. Adding synthetic aperture radar data, namely, backscatter and interferometric coherence, increases the accuracy in all regions where available. The calibrated confidence for glacier extents is reported making the predictions more reliable and interpretable. We also release a benchmark dataset that covers 9% of glaciers worldwide. Our results support efforts towards automated multitemporal and global glacier mapping.
Authors: Xindi Wang, Mahsa Salmani, Parsa Omidi, Xiangyu Ren, Mehdi Rezagholizadeh, Armaghan Eshaghi
Abstract: Recently, large language models (LLMs) have shown remarkable capabilities including understanding context, engaging in logical reasoning, and generating responses. However, this is achieved at the expense of stringent computational and memory requirements, hindering their ability to effectively support long input sequences. This survey provides an inclusive review of the recent techniques and methods devised to extend the sequence length in LLMs, thereby enhancing their capacity for long-context understanding. In particular, we review and categorize a wide range of techniques including architectural modifications, such as modified positional encoding and altered attention mechanisms, which are designed to enhance the processing of longer sequences while avoiding a proportional increase in computational requirements. The diverse methodologies investigated in this study can be leveraged across different phases of LLMs, i.e., training, fine-tuning and inference. This enables LLMs to efficiently process extended sequences. The limitations of the current methodologies is discussed in the last section along with the suggestions for future research directions, underscoring the importance of sequence length in the continued advancement of LLMs.
Authors: Wenlin Chen, Mingtian Zhang, Brooks Paige, Jos\'e Miguel Hern\'andez-Lobato, David Barber
Abstract: The inadequate mixing of conventional Markov Chain Monte Carlo (MCMC) methods for multi-modal distributions presents a significant challenge in practical applications such as Bayesian inference and molecular dynamics. Addressing this, we propose Diffusive Gibbs Sampling (DiGS), an innovative family of sampling methods designed for effective sampling from distributions characterized by distant and disconnected modes. DiGS integrates recent developments in diffusion models, leveraging Gaussian convolution to create an auxiliary noisy distribution that bridges isolated modes in the original space and applying Gibbs sampling to alternately draw samples from both spaces. A novel Metropolis-within-Gibbs scheme is proposed to enhance mixing in the denoising sampling step. DiGS exhibits a better mixing property for sampling multi-modal distributions than state-of-the-art methods such as parallel tempering, attaining substantially improved performance across various tasks, including mixtures of Gaussians, Bayesian neural networks and molecular dynamics.
Authors: Christopher Liao, Christian So, Theodoros Tsiligkaridis, Brian Kulis
Abstract: Domain generalization (DG) is an important problem that learns a model which generalizes to unseen test domains leveraging one or more source domains, under the assumption of shared label spaces. However, most DG methods assume access to abundant source data in the target label space, a requirement that proves overly stringent for numerous real-world applications, where acquiring the same label space as the target task is prohibitively expensive. For this setting, we tackle the multimodal version of the unsupervised domain generalization (MUDG) problem, which uses a large task-agnostic unlabeled source dataset during finetuning. Our framework does not explicitly assume any relationship between the source dataset and target task. Instead, it relies only on the premise that the source dataset can be accurately and efficiently searched in a joint vision-language space. We make three contributions in the MUDG setting. Firstly, we show theoretically that cross-modal approximate nearest neighbor search suffers from low recall due to the large distance between text queries and the image centroids used for coarse quantization. Accordingly, we propose paired k-means, a simple clustering algorithm that improves nearest neighbor recall by storing centroids in query space instead of image space. Secondly, we propose an adaptive text augmentation scheme for target labels designed to improve zero-shot accuracy and diversify retrieved image data. Lastly, we present two simple but effective components to further improve downstream target accuracy. We compare against state-of-the-art name-only transfer, source-free DG and zero-shot (ZS) methods on their respective benchmarks and show consistent improvement in accuracy on 20 diverse datasets. Code is available: https://github.com/Chris210634/mudg
Authors: Tam Le, Truyen Nguyen, Kenji Fukumizu
Abstract: We study the optimal transport (OT) problem for measures supported on a graph metric space. Recently, Le et al. (2022) leverage the graph structure and propose a variant of OT, namely Sobolev transport (ST), which yields a closed-form expression for a fast computation. However, ST is essentially coupled with the $L^p$ geometric structure within its definition which makes it nontrivial to utilize ST for other prior structures. In contrast, the classic OT has the flexibility to adapt to various geometric structures by modifying the underlying cost function. An important instance is the Orlicz-Wasserstein (OW) which moves beyond the $L^p$ structure by leveraging the \emph{Orlicz geometric structure}. Comparing to the usage of standard $p$-order Wasserstein, OW remarkably helps to advance certain machine learning approaches. Nevertheless, OW brings up a new challenge on its computation due to its two-level optimization formulation. In this work, we leverage a specific class of convex functions for Orlicz structure to propose the generalized Sobolev transport (GST). GST encompasses the ST as its special case, and can be utilized for prior structures beyond the $L^p$ geometry. In connection with the OW, we show that one only needs to simply solve a univariate optimization problem to compute the GST, unlike the complex two-level optimization problem in OW. We empirically illustrate that GST is several-order faster than the OW. Moreover, we provide preliminary evidences on the advantages of GST for document classification and for several tasks in topological data analysis.
Authors: Sahel Iqbal, Adrien Corenflos, Simo S\"arkk\"a, Hany Abdulsamad
Abstract: In this paper, we propose a novel approach to Bayesian experimental design for non-exchangeable data that formulates it as risk-sensitive policy optimization. We develop the Inside-Out SMC$^2$ algorithm, a nested sequential Monte Carlo technique to infer optimal designs, and embed it into a particle Markov chain Monte Carlo framework to perform gradient-based policy amortization. Our approach is distinct from other amortized experimental design techniques, as it does not rely on contrastive estimators. Numerical validation on a set of dynamical systems showcases the efficacy of our method in comparison to other state-of-the-art strategies.
Authors: Jiwon Song, Kyungseok Oh, Taesu Kim, Hyungjun Kim, Yulhwa Kim, Jae-Joon Kim
Abstract: Large language models (LLMs) have proven to be highly effective across various natural language processing tasks. However, their large number of parameters poses significant challenges for practical deployment. Pruning, a technique aimed at reducing the size and complexity of LLMs, offers a potential solution by removing redundant components from the network. Despite the promise of pruning, existing methods often struggle to achieve substantial end-to-end LLM inference speedup. In this paper, we introduce SLEB, a novel approach designed to streamline LLMs by eliminating redundant transformer blocks. We choose the transformer block as the fundamental unit for pruning, because LLMs exhibit block-level redundancy with high similarity between the outputs of neighboring blocks. This choice allows us to effectively enhance the processing speed of LLMs. Our experimental results demonstrate that SLEB outperforms previous LLM pruning methods in accelerating LLM inference while also maintaining superior perplexity and accuracy, making SLEB as a promising technique for enhancing the efficiency of LLMs. The code is available at: https://github.com/jiwonsong-dev/SLEB.
Authors: Hila Manor, Tomer Michaeli
Abstract: Editing signals using large pre-trained models, in a zero-shot manner, has recently seen rapid advancements in the image domain. However, this wave has yet to reach the audio domain. In this paper, we explore two zero-shot editing techniques for audio signals, which use DDPM inversion with pre-trained diffusion models. The first, which we coin ZEro-shot Text-based Audio (ZETA) editing, is adopted from the image domain. The second, named ZEro-shot UnSupervized (ZEUS) editing, is a novel approach for discovering semantically meaningful editing directions without supervision. When applied to music signals, this method exposes a range of musically interesting modifications, from controlling the participation of specific instruments to improvisations on the melody. Samples and code can be found in https://hilamanor.github.io/AudioEditing/ .
Authors: Hualin Zhan, Viqar Ahmad, Azul Mayon, Grace Tabi, Anh Dinh Bui, Zhuofeng Li, Daniel Walter, Hieu Nguyen, Klaus Weber, Thomas White, Kylie Catchpole
Abstract: The ability to extract material parameters of perovskite from quantitative experimental analysis is essential for rational design of photovoltaic and optoelectronic applications. However, the difficulty of this analysis increases significantly with the complexity of the theoretical model and the number of material parameters for perovskite. Here we use Bayesian optimization to develop an analysis platform that can extract up to 8 fundamental material parameters of an organometallic perovskite semiconductor from a transient photoluminescence experiment, based on a complex full physics model that includes drift-diffusion of carriers and dynamic defect occupation. An example study of thermal degradation reveals that the carrier mobility and trap-assisted recombination coefficient are reduced noticeably, while the defect energy level remains nearly unchanged. The reduced carrier mobility can dominate the overall effect on thermal degradation of perovskite solar cells by reducing the fill factor, despite the opposite effect of the reduced trap-assisted recombination coefficient on increasing the fill factor. In future, this platform can be conveniently applied to other experiments or to combinations of experiments, accelerating materials discovery and optimization of semiconductor materials for photovoltaics and other applications.
Authors: Shen Li, Liuyi Yao, Jinyang Gao, Lan Zhang, Yaliang Li
Abstract: To support various applications, a prevalent and efficient approach for business owners is leveraging their valuable datasets to fine-tune a pre-trained LLM through the API provided by LLM owners or cloud servers. However, this process carries a substantial risk of model misuse, potentially resulting in severe economic consequences for business owners. Thus, safeguarding the copyright of these customized models during LLM fine-tuning has become an urgent practical requirement, but there are limited existing solutions to provide such protection. To tackle this pressing issue, we propose a novel watermarking approach named ``Double-I watermark''. Specifically, based on the instruct-tuning data, two types of backdoor data paradigms are introduced with trigger in the instruction and the input, respectively. By leveraging LLM's learning capability to incorporate customized backdoor samples into the dataset, the proposed approach effectively injects specific watermarking information into the customized model during fine-tuning, which makes it easy to inject and verify watermarks in commercial scenarios. We evaluate the proposed "Double-I watermark" under various fine-tuning methods, demonstrating its harmlessness, robustness, uniqueness, imperceptibility, and validity through both quantitative and qualitative analyses.
Authors: Jeffrey G. Wang, Jason Wang, Marvin Li, Seth Neel
Abstract: In this paper we develop state-of-the-art privacy attacks against Large Language Models (LLMs), where an adversary with some access to the model tries to learn something about the underlying training data. Our headline results are new membership inference attacks (MIAs) against pretrained LLMs that perform hundreds of times better than baseline attacks, and a pipeline showing that over 50% (!) of the fine-tuning dataset can be extracted from a fine-tuned LLM in natural settings. We consider varying degrees of access to the underlying model, pretraining and fine-tuning data, and both MIAs and training data extraction. For pretraining data, we propose two new MIAs: a supervised neural network classifier that predicts training data membership on the basis of (dimensionality-reduced) model gradients, as well as a variant of this attack that only requires logit access to the model which leverages recent model-stealing work on LLMs. To our knowledge this is the first MIA that explicitly incorporates model-stealing information. Both attacks outperform existing black-box baselines, and our supervised attack closes the gap between MIA attack success against LLMs and the strongest known attacks for other machine learning models. In fine-tuning, we find that a simple attack based on the ratio of the loss between the base and fine-tuned models is able to achieve near-perfect MIA performance; we then leverage our MIA to extract a large fraction of the fine-tuning dataset from fine-tuned Pythia and Llama models. Taken together, these results represent the strongest existing privacy attacks against both pretrained and fine-tuned LLMs for MIAs and training data extraction, which are of independent scientific interest and have important practical implications for LLM security, privacy, and copyright issues.
Authors: Muhammad Faaiz Taufiq, Jean-Francois Ton, Yang Liu
Abstract: In machine learning fairness, training models that minimize disparity across different sensitive groups often leads to diminished accuracy, a phenomenon known as the fairness-accuracy trade-off. The severity of this trade-off inherently depends on dataset characteristics such as dataset imbalances or biases and therefore, using a uniform fairness requirement across diverse datasets remains questionable. To address this, we present a computationally efficient approach to approximate the fairness-accuracy trade-off curve tailored to individual datasets, backed by rigorous statistical guarantees. By utilizing the You-Only-Train-Once (YOTO) framework, our approach mitigates the computational burden of having to train multiple models when approximating the trade-off curve. Crucially, we introduce a novel methodology for quantifying uncertainty in our estimates, thereby providing practitioners with a robust framework for auditing model fairness while avoiding false conclusions due to estimation errors. Our experiments spanning tabular (e.g., Adult), image (CelebA), and language (Jigsaw) datasets underscore that our approach not only reliably quantifies the optimum achievable trade-offs across various data modalities but also helps detect suboptimality in SOTA fairness methods.
Authors: Tijana Zrnic, Emmanuel J. Cand\`es
Abstract: Inspired by the concept of active learning, we propose active inference$\unicode{x2013}$a methodology for statistical inference with machine-learning-assisted data collection. Assuming a budget on the number of labels that can be collected, the methodology uses a machine learning model to identify which data points would be most beneficial to label, thus effectively utilizing the budget. It operates on a simple yet powerful intuition: prioritize the collection of labels for data points where the model exhibits uncertainty, and rely on the model's predictions where it is confident. Active inference constructs provably valid confidence intervals and hypothesis tests while leveraging any black-box machine learning model and handling any data distribution. The key point is that it achieves the same level of accuracy with far fewer samples than existing baselines relying on non-adaptively-collected data. This means that for the same number of collected samples, active inference enables smaller confidence intervals and more powerful p-values. We evaluate active inference on datasets from public opinion research, census analysis, and proteomics.
Authors: Dingkang Yang, Kun Yang, Mingcheng Li, Shunli Wang, Shuaibing Wang, Lihua Zhang
Abstract: Context-aware emotion recognition (CAER) has recently boosted the practical applications of affective computing techniques in unconstrained environments. Mainstream CAER methods invariably extract ensemble representations from diverse contexts and subject-centred characteristics to perceive the target person's emotional state. Despite advancements, the biggest challenge remains due to context bias interference. The harmful bias forces the models to rely on spurious correlations between background contexts and emotion labels in likelihood estimation, causing severe performance bottlenecks and confounding valuable context priors. In this paper, we propose a counterfactual emotion inference (CLEF) framework to address the above issue. Specifically, we first formulate a generalized causal graph to decouple the causal relationships among the variables in CAER. Following the causal graph, CLEF introduces a non-invasive context branch to capture the adverse direct effect caused by the context bias. During the inference, we eliminate the direct context effect from the total causal effect by comparing factual and counterfactual outcomes, resulting in bias mitigation and robust prediction. As a model-agnostic framework, CLEF can be readily integrated into existing methods, bringing consistent performance gains.
Authors: Xiang Li, Soo Min Kwon, Ismail R. Alkhouri, Saiprasad Ravishankar, Qing Qu
Abstract: Diffusion models have recently gained traction as a powerful class of deep generative priors, excelling in a wide range of image restoration tasks due to their exceptional ability to model data distributions. To solve image restoration problems, many existing techniques achieve data consistency by incorporating additional likelihood gradient steps into the reverse sampling process of diffusion models. However, the additional gradient steps pose a challenge for real-world practical applications as they incur a large computational overhead, thereby increasing inference time. They also present additional difficulties when using accelerated diffusion model samplers, as the number of data consistency steps is limited by the number of reverse sampling steps. In this work, we propose a novel diffusion-based image restoration solver that addresses these issues by decoupling the reverse process from the data consistency steps. Our method involves alternating between a reconstruction phase to maintain data consistency and a refinement phase that enforces the prior via diffusion purification. Our approach demonstrates versatility, making it highly adaptable for efficient problem-solving in latent space. Additionally, it reduces the necessity for numerous sampling steps through the integration of consistency models. The efficacy of our approach is validated through comprehensive experiments across various image restoration tasks, including image denoising, deblurring, inpainting, and super-resolution.
Authors: Alina Petukhova, Joao P. Matos-Carvalho, Nuno Fachada
Abstract: Text clustering is an important approach for organising the growing amount of digital content, helping to structure and find hidden patterns in uncategorised data. In this research, we investigated how different textual embeddings - particularly those used in large language models (LLMs) - and clustering algorithms affect how text datasets are clustered. A series of experiments were conducted to assess how embeddings influence clustering results, the role played by dimensionality reduction through summarisation, and embedding size adjustment. Results reveal that LLM embeddings excel at capturing the nuances of structured language, while BERT leads the lightweight options in performance. In addition, we find that increasing embedding dimensionality and summarisation techniques do not uniformly improve clustering efficiency, suggesting that these strategies require careful analysis to use in real-life models. These results highlight a complex balance between the need for nuanced text representation and computational feasibility in text clustering applications. This study extends traditional text clustering frameworks by incorporating embeddings from LLMs, thereby paving the way for improved methodologies and opening new avenues for future research in various types of textual analysis.
Authors: Leying Zhang, Yao Qian, Long Zhou, Shujie Liu, Dongmei Wang, Xiaofei Wang, Midia Yousefi, Yanmin Qian, Jinyu Li, Lei He, Sheng Zhao, Michael Zeng
Abstract: Recent advancements in zero-shot text-to-speech (TTS) modeling have led to significant strides in generating high-fidelity and diverse speech. However, dialogue generation, along with achieving human-like naturalness in speech, continues to be a challenge. In this paper, we introduce CoVoMix: Conversational Voice Mixture Generation, a novel model for zero-shot, human-like, multi-speaker, multi-round dialogue speech generation. CoVoMix first converts dialogue text into multiple streams of discrete tokens, with each token stream representing semantic information for individual talkers. These token streams are then fed into a flow-matching based acoustic model to generate mixed mel-spectrograms. Finally, the speech waveforms are produced using a HiFi-GAN model. Furthermore, we devise a comprehensive set of metrics for measuring the effectiveness of dialogue modeling and generation. Our experimental results show that CoVoMix can generate dialogues that are not only human-like in their naturalness and coherence but also involve multiple talkers engaging in multiple rounds of conversation. This is exemplified by instances generated in a single channel where one speaker's utterance is seamlessly mixed with another's interjections or laughter, indicating the latter's role as an attentive listener. Audio samples are available at https://aka.ms/covomix.
URLs: https://aka.ms/covomix.
Authors: Sanat Sharma, Jayant Kumar, Twisha Naik, Zhaoyu Lu, Arvind Srikantan, Tracy Holloway King
Abstract: Accurate explicit and implicit product identification in search queries is critical for enhancing user experiences, especially at a company like Adobe which has over 50 products and covers queries across hundreds of tools. In this work, we present a novel approach to training a product classifier from user behavioral data. Our semantic model led to >25% relative improvement in CTR (click through rate) across the deployed surfaces; a >50% decrease in null rate; a 2x increase in the app cards surfaced, which helps drive product visibility.
Authors: Sanat Sharma, David Seunghyun Yoon, Franck Dernoncourt, Dewang Sultania, Karishma Bagga, Mengjiao Zhang, Trung Bui, Varun Kotte
Abstract: Question answering (QA) has become an important application in the advanced development of large language models. General pre-trained large language models for question-answering are not trained to properly understand the knowledge or terminology for a specific domain, such as finance, healthcare, education, and customer service for a product. To better cater to domain-specific understanding, we build an in-house question-answering system for Adobe products. We propose a novel framework to compile a large question-answer database and develop the approach for retrieval-aware finetuning of a Large Language model. We showcase that fine-tuning the retriever leads to major improvements in the final generation. Our overall approach reduces hallucinations during generation while keeping in context the latest retrieval information for contextual grounding.
Authors: Yueyuan Sui, Minghui Zhao, Junxi Xia, Xiaofan Jiang, Stephen Xia
Abstract: We propose TRAMBA, a hybrid transformer and Mamba architecture for acoustic and bone conduction speech enhancement, suitable for mobile and wearable platforms. Bone conduction speech enhancement has been impractical to adopt in mobile and wearable platforms for several reasons: (i) data collection is labor-intensive, resulting in scarcity; (ii) there exists a performance gap between state of-art models with memory footprints of hundreds of MBs and methods better suited for resource-constrained systems. To adapt TRAMBA to vibration-based sensing modalities, we pre-train TRAMBA with audio speech datasets that are widely available. Then, users fine-tune with a small amount of bone conduction data. TRAMBA outperforms state-of-art GANs by up to 7.3% in PESQ and 1.8% in STOI, with an order of magnitude smaller memory footprint and an inference speed up of up to 465 times. We integrate TRAMBA into real systems and show that TRAMBA (i) improves battery life of wearables by up to 160% by requiring less data sampling and transmission; (ii) generates higher quality voice in noisy environments than over-the-air speech; (iii) requires a memory footprint of less than 20.0 MB.
Authors: Peter Schmitt-F\"orster, Tobias Sutter
Abstract: We propose a new Q-learning variant, called 2RA Q-learning, that addresses some weaknesses of existing Q-learning methods in a principled manner. One such weakness is an underlying estimation bias which cannot be controlled and often results in poor performance. We propose a distributionally robust estimator for the maximum expected value term, which allows us to precisely control the level of estimation bias introduced. The distributionally robust estimator admits a closed-form solution such that the proposed algorithm has a computational cost per iteration comparable to Watkins' Q-learning. For the tabular case, we show that 2RA Q-learning converges to the optimal policy and analyze its asymptotic mean-squared error. Lastly, we conduct numerical experiments for various settings, which corroborate our theoretical findings and indicate that 2RA Q-learning often performs better than existing methods.
Authors: Jonas Katona, Xiuyuan Wang, Andre Wibisono
Abstract: Motivated by understanding the behavior of the Alternating Mirror Descent (AMD) algorithm for bilinear zero-sum games, we study the discretization of continuous-time Hamiltonian flow via the symplectic Euler method. We provide a framework for analysis using results from Hamiltonian dynamics, Lie algebra, and symplectic numerical integrators, with an emphasis on the existence and properties of a conserved quantity, the modified Hamiltonian (MH), for the symplectic Euler method. We compute the MH in closed-form when the original Hamiltonian is a quadratic function, and show that it generally differs from the other conserved quantity known previously in that case. We derive new error bounds on the MH when truncated at orders in the stepsize in terms of the number of iterations, $K$, and use these bounds to show an improved $\mathcal{O}(K^{1/5})$ total regret bound and an $\mathcal{O}(K^{-4/5})$ duality gap of the average iterates for AMD. Finally, we propose a conjecture which, if true, would imply that the total regret for AMD scales as $\mathcal{O}\left(K^{\varepsilon}\right)$ and the duality gap of the average iterates as $\mathcal{O}\left(K^{-1+\varepsilon}\right)$ for any $\varepsilon>0$, and we can take $\varepsilon=0$ upon certain convergence conditions for the MH.
Authors: Chen Shao, Elias Giacoumidis, Syed Moktacim Billah, Shi Li, Jialei Li, Prashasti Sahu, Andre Richter, Tobias Kaefer, Michael Faerber
Abstract: In recent years, extensive research has been conducted to explore the utilization of machine learning algorithms in various direct-detected and self-coherent short-reach communication applications. These applications encompass a wide range of tasks, including bandwidth request prediction, signal quality monitoring, fault detection, traffic prediction, and digital signal processing (DSP)-based equalization. As a versatile approach, machine learning demonstrates the ability to address stochastic phenomena in optical systems networks where deterministic methods may fall short. However, when it comes to DSP equalization algorithms, their performance improvements are often marginal, and their complexity is prohibitively high, especially in cost-sensitive short-reach communications scenarios such as passive optical networks (PONs). They excel in capturing temporal dependencies, handling irregular or nonlinear patterns effectively, and accommodating variable time intervals. Within this extensive survey, we outline the application of machine learning techniques in short-reach communications, specifically emphasizing their utilization in high-bandwidth demanding PONs. Notably, we introduce a novel taxonomy for time-series methods employed in machine learning signal processing, providing a structured classification framework. Our taxonomy categorizes current time series methods into four distinct groups: traditional methods, Fourier convolution-based methods, transformer-based models, and time-series convolutional networks. Finally, we highlight prospective research directions within this rapidly evolving field and outline specific solutions to mitigate the complexity associated with hardware implementations. We aim to pave the way for more practical and efficient deployment of machine learning approaches in short-reach optical communication systems by addressing complexity concerns.
Authors: Andres Hernandez, Zhongqi Miao, Luisa Vargas, Rahul Dodhia, Juan Lavista
Abstract: The alarming decline in global biodiversity, driven by various factors, underscores the urgent need for large-scale wildlife monitoring. In response, scientists have turned to automated deep learning methods for data processing in wildlife monitoring. However, applying these advanced methods in real-world scenarios is challenging due to their complexity and the need for specialized knowledge, primarily because of technical challenges and interdisciplinary barriers. To address these challenges, we introduce Pytorch-Wildlife, an open-source deep learning platform built on PyTorch. It is designed for creating, modifying, and sharing powerful AI models. This platform emphasizes usability and accessibility, making it accessible to individuals with limited or no technical background. It also offers a modular codebase to simplify feature expansion and further development. Pytorch-Wildlife offers an intuitive, user-friendly interface, accessible through local installation or Hugging Face, for animal detection and classification in images and videos. As two real-world applications, Pytorch-Wildlife has been utilized to train animal classification models for species recognition in the Amazon Rainforest and for invasive opossum recognition in the Galapagos Islands. The Opossum model achieves 98% accuracy, and the Amazon model has 92% recognition accuracy for 36 animals in 90% of the data. As Pytorch-Wildlife evolves, we aim to integrate more conservation tasks, addressing various environmental challenges. Pytorch-Wildlife is available at https://github.com/microsoft/CameraTraps.
Authors: Huajie Qian, Donghao Ying, Henry Lam, Wotao Yin
Abstract: Bagging is a popular ensemble technique to improve the accuracy of machine learning models. It hinges on the well-established rationale that, by repeatedly retraining on resampled data, the aggregated model exhibits lower variance and hence higher stability, especially for discontinuous base learners. In this paper, we provide a new perspective on bagging: By suitably aggregating the base learners at the parametrization instead of the output level, bagging improves generalization performances exponentially, a strength that is significantly more powerful than variance reduction. More precisely, we show that for general stochastic optimization problems that suffer from slowly (i.e., polynomially) decaying generalization errors, bagging can effectively reduce these errors to an exponential decay. Moreover, this power of bagging is agnostic to the solution schemes, including common empirical risk minimization, distributionally robust optimization, and various regularizations. We demonstrate how bagging can substantially improve generalization performances in a range of examples involving heavy-tailed data that suffer from intrinsically slow rates.
Authors: Andres Algaba, Carmen Mazijn, Vincent Holst, Floriano Tori, Sylvia Wenmackers, Vincent Ginis
Abstract: Citation practices are crucial in shaping the structure of scientific knowledge, yet they are often influenced by contemporary norms and biases. The emergence of Large Language Models (LLMs) like GPT-4 introduces a new dynamic to these practices. Interestingly, the characteristics and potential biases of references recommended by LLMs that entirely rely on their parametric knowledge, and not on search or retrieval-augmented generation, remain unexplored. Here, we analyze these characteristics in an experiment using a dataset of 166 papers from AAAI, NeurIPS, ICML, and ICLR, published after GPT-4's knowledge cut-off date, encompassing 3,066 references in total. In our experiment, GPT-4 was tasked with suggesting scholarly references for the anonymized in-text citations within these papers. Our findings reveal a remarkable similarity between human and LLM citation patterns, but with a more pronounced high citation bias in GPT-4, which persists even after controlling for publication year, title length, number of authors, and venue. Additionally, we observe a large consistency between the characteristics of GPT-4's existing and non-existent generated references, indicating the model's internalization of citation patterns. By analyzing citation graphs, we show that the references recommended by GPT-4 are embedded in the relevant citation context, suggesting an even deeper conceptual internalization of the citation networks. While LLMs can aid in citation generation, they may also amplify existing biases and introduce new ones, potentially skewing scientific knowledge dissemination. Our results underscore the need for identifying the model's biases and for developing balanced methods to interact with LLMs in general.
Authors: Abhishek Kumar, Robert Morabito, Sanzhar Umbet, Jad Kabbara, Ali Emami
Abstract: As the use of Large Language Models (LLMs) becomes more widespread, understanding their self-evaluation of confidence in generated responses becomes increasingly important as it is integral to the reliability of the output of these models. We introduce the concept of Confidence-Probability Alignment, that connects an LLM's internal confidence, quantified by token probabilities, to the confidence conveyed in the model's response when explicitly asked about its certainty. Using various datasets and prompting techniques that encourage model introspection, we probe the alignment between models' internal and expressed confidence. These techniques encompass using structured evaluation scales to rate confidence, including answer options when prompting, and eliciting the model's confidence level for outputs it does not recognize as its own. Notably, among the models analyzed, OpenAI's GPT-4 showed the strongest confidence-probability alignment, with an average Spearman's $\hat{\rho}$ of 0.42, across a wide range of tasks. Our work contributes to the ongoing efforts to facilitate risk assessment in the application of LLMs and to further our understanding of model trustworthiness.
Authors: Jianqiao Lu, Zhiyang Dou, Hongru Wang, Zeyu Cao, Jianbo Dai, Yingjia Wan, Yinya Huang, Zhijiang Guo
Abstract: In this work, we propose a novel method named \textbf{Auto}mated Process Labeling via \textbf{C}onfidence \textbf{V}ariation (\textbf{\textsc{AutoCV}}) to enhance the reasoning capabilities of large language models (LLMs) by automatically annotating the reasoning steps. Our approach begins by training a verification model on the correctness of final answers, enabling it to generate automatic process annotations. This verification model assigns a confidence score to each reasoning step, indicating the probability of arriving at the correct final answer from that point onward. We detect relative changes in the verification's confidence scores across reasoning steps to automatically annotate the reasoning process. This alleviates the need for numerous manual annotations or the high computational costs associated with model-induced annotation approaches. We experimentally validate that the confidence variations learned by the verification model trained on the final answer correctness can effectively identify errors in the reasoning steps. Subsequently, we demonstrate that the process annotations generated by \textsc{AutoCV} can improve the accuracy of the verification model in selecting the correct answer from multiple outputs generated by LLMs. Notably, we achieve substantial improvements across five datasets in mathematics and commonsense reasoning. The source code of \textsc{AutoCV} is available at \url{https://github.com/rookie-joe/AUTOCV}.
Authors: Fan Wang, Chuan Lin, Yang Cao, Yu Kang
Abstract: In-context learning (ICL) capabilities are becoming increasingly appealing for building general intelligence due to their sample efficiency and independence from artificial optimization skills. To enhance generalization, biological neural systems primarily inherit learning capabilities and subsequently refine their memory, acquiring diverse skills and knowledge through extensive lifelong experiences. This process gives rise to the concept of general-purpose in-context learning (GPICL). Compared to standard ICL, GPICL addresses a broader range of tasks, extends learning horizons, and starts at a lower zero-shot baseline. We introduce two lightweight but insightful benchmarks specifically crafted to train and evaluate GPICL functionalities. Each benchmark includes a vast number of tasks characterized by significant task variance and minimal transferable knowledge among tasks, facilitating lifelong in-context learning through continuous generation and interaction. These features pose significant challenges for models that rely on context or interactions to improve their proficiency, including language models, decision models, and world models. Our experiments reveal that parameter scale alone may not be crucial for ICL or GPICL, suggesting alternative approaches such as increasing the scale of contexts and memory states.
Authors: Franz Motzkus, Georgii Mikriukov, Christian Hellert, Ute Schmid
Abstract: Ensuring the quality of black-box Deep Neural Networks (DNNs) has become ever more significant, especially in safety-critical domains such as automated driving. While global concept encodings generally enable a user to test a model for a specific concept, linking global concept encodings to the local processing of single network inputs reveals their strengths and limitations. Our proposed framework global-to-local Concept Attribution (glCA) uses approaches from local (why a specific prediction originates) and global (how a model works generally) eXplainable Artificial Intelligence (xAI) to test DNNs for a predefined semantical concept locally. The approach allows for conditioning local, post-hoc explanations on predefined semantic concepts encoded as linear directions in the model's latent space. Pixel-exact scoring concerning the global concept usage assists the tester in further understanding the model processing of single data points for the selected concept. Our approach has the advantage of fully covering the model-internal encoding of the semantic concept and allowing the localization of relevant concept-related information. The results show major differences in the local perception and usage of individual global concept encodings and demand for further investigations regarding obtaining thorough semantic concept encodings.
Authors: Ilia Azizi, Marc-Olivier Boldi, Val\'erie Chavez-Demoulin
Abstract: This work introduces the Supervised Expectation-Maximization Framework (SEMF), a versatile and model-agnostic framework that generates prediction intervals for datasets with complete or missing data. SEMF extends the Expectation-Maximization (EM) algorithm, traditionally used in unsupervised learning, to a supervised context, enabling it to extract latent representations for uncertainty estimation. The framework demonstrates robustness through extensive empirical evaluation across 11 tabular datasets, achieving$\unicode{x2013}$in some cases$\unicode{x2013}$narrower normalized prediction intervals and higher coverage than traditional quantile regression methods. Furthermore, SEMF integrates seamlessly with existing machine learning algorithms, such as gradient-boosted trees and neural networks, exemplifying its usefulness for real-world applications. The experimental results highlight SEMF's potential to advance state-of-the-art techniques in uncertainty quantification.
Authors: Yixiao Zhang, Yukara Ikemiya, Woosung Choi, Naoki Murata, Marco A. Mart\'inez-Ram\'irez, Liwei Lin, Gus Xia, Wei-Hsiang Liao, Yuki Mitsufuji, Simon Dixon
Abstract: Recent advances in text-to-music editing, which employ text queries to modify music (e.g.\ by changing its style or adjusting instrumental components), present unique challenges and opportunities for AI-assisted music creation. Previous approaches in this domain have been constrained by the necessity to train specific editing models from scratch, which is both resource-intensive and inefficient; other research uses large language models to predict edited music, resulting in imprecise audio reconstruction. To Combine the strengths and address these limitations, we introduce Instruct-MusicGen, a novel approach that finetunes a pretrained MusicGen model to efficiently follow editing instructions such as adding, removing, or separating stems. Our approach involves a modification of the original MusicGen architecture by incorporating a text fusion module and an audio fusion module, which allow the model to process instruction texts and audio inputs concurrently and yield the desired edited music. Remarkably, Instruct-MusicGen only introduces 8% new parameters to the original MusicGen model and only trains for 5K steps, yet it achieves superior performance across all tasks compared to existing baselines, and demonstrates performance comparable to the models trained for specific tasks. This advancement not only enhances the efficiency of text-to-music editing but also broadens the applicability of music language models in dynamic music production environments.
Authors: Ethan Shen, Alan Fan, Sarah M Pratt, Jae Sung Park, Matthew Wallingford, Sham M. Kakade, Ari Holtzman, Ranjay Krishna, Ali Farhadi, Aditya Kusupati
Abstract: Many applications today provide users with multiple auto-complete drafts as they type, including GitHub's code completion, Gmail's smart compose, and Apple's messaging auto-suggestions. Under the hood, language models support this by running an autoregressive inference pass to provide a draft. Consequently, providing $k$ drafts to the user requires running an expensive language model $k$ times. To alleviate the computation cost of running $k$ inference passes, we propose Superposed Decoding, a new decoding algorithm that generates $k$ drafts at the computation cost of one autoregressive inference pass. We achieve this by feeding a superposition of the most recent token embeddings from the $k$ drafts as input to the next decoding step of the language model. At every inference step we combine the $k$ drafts with the top-$k$ tokens to get $k^2$ new drafts and cache the $k$ most likely options, using an n-gram interpolation with minimal compute overhead to filter out incoherent generations. Our experiments show that $k$ drafts from Superposed Decoding are at least as coherent and factual as Nucleus Sampling and Greedy Decoding respectively, while being at least $2.44\times$ faster for $k\ge3$. In a compute-normalized setting, user evaluations demonstrably favor text generated by Superposed Decoding over Nucleus Sampling. Code and more examples open-sourced at https://github.com/RAIVNLab/SuperposedDecoding.