Authors: Sazzad Hossain, Touhidul Alam Seyam, Avijit Chowdhury, Munis Xamidov, Rajib Ghose, Abhijit Pathak
Abstract: This paper presents a comparative study aimed at optimizing Llama2 inference, a critical aspect of machine learning and natural language processing (NLP). We evaluate various programming languages and frameworks, including TensorFlow, PyTorch, Python, Mojo, C++, and Java, analyzing their performance in terms of speed, memory consumption, and ease of implementation through extensive benchmarking. Strengths and limitations of each approach are highlighted, along with proposed optimization strategies for parallel processing and hardware utilization. Furthermore, we investigate the Mojo SDK, a novel framework designed for large language model (LLM) inference on Apple Silicon, benchmarking its performance against implementations in C, C++, Rust, Zig, Go, and Julia. Our experiments, conducted on an Apple M1 Max, demonstrate Mojo SDK's competitive performance, ease of use, and seamless Python compatibility, positioning it as a strong alternative for LLM inference on Apple Silicon. We also discuss broader implications for LLM deployment on resource-constrained hardware and identify potential directions for future research.
Authors: Soham Sane
Abstract: Hybrid Group Relative Policy Optimization (Hybrid GRPO) is a reinforcement learning framework that extends Proximal Policy Optimization (PPO) and Group Relative Policy Optimization (GRPO) by incorporating empirical multi-sample action evaluation while preserving the stability of value function-based learning. Unlike DeepSeek GRPO, which eliminates the value function in favor of purely empirical reward estimation, Hybrid GRPO introduces a structured advantage computation method that balances empirical action sampling with bootstrapped value estimation. This approach enhances sample efficiency, improves learning stability, and mitigates variance amplification observed in purely empirical methods. A detailed mathematical comparison between PPO, DeepSeek GRPO, and Hybrid GRPO is presented, highlighting key differences in advantage estimation and policy updates. Experimental validation in a controlled reinforcement learning environment demonstrates that Hybrid GRPO achieves superior convergence speed, more stable policy updates, and improved sample efficiency compared to existing methods. Several extensions to Hybrid GRPO are explored, including entropy-regularized sampling, hierarchical multi-step sub-sampling, adaptive reward normalization, and value-based action selection. Beyond reinforcement learning in simulated environments, Hybrid GRPO provides a scalable framework for bridging the gap between large language models (LLMs) and real-world agent-based decision-making. By integrating structured empirical sampling with reinforcement learning stability mechanisms, Hybrid GRPO has potential applications in autonomous robotics, financial modeling, and AI-driven control systems. These findings suggest that Hybrid GRPO serves as a robust and adaptable reinforcement learning methodology, paving the way for further advancements in policy optimization.
Authors: Iat Hang Fong, Tengyue Li, Simon Fong, Raymond K. Wong, Antonio J. Tall\'on-Ballesteros
Abstract: Air pollution (AP) poses a great threat to human health, and people are paying more attention than ever to its prediction. Accurate prediction of AP helps people to plan for their outdoor activities and aids protecting human health. In this paper, long-short term memory (LSTM) recurrent neural networks (RNNs) have been used to predict the future concentration of air pollutants (APS) in Macau. Additionally, meteorological data and data on the concentration of APS have been utilized. Moreover, in Macau, some air quality monitoring stations (AQMSs) have less observed data in quantity, and, at the same time, some AQMSs recorded less observed data of certain types of APS. Therefore, the transfer learning and pre-trained neural networks have been employed to assist AQMSs with less observed data to build a neural network with high prediction accuracy. The experimental sample covers a period longer than 12-year and includes daily measurements from several APS as well as other more classical meteorological values. Records from five stations, four out of them are AQMSs and the remaining one is an automatic weather station, have been prepared from the aforesaid period and eventually underwent to computational intelligence techniques to build and extract a prediction knowledge-based system. As shown by experimentation, LSTM RNNs initialized with transfer learning methods have higher prediction accuracy; it incurred shorter training time than randomly initialized recurrent neural networks.
Authors: Jinyan Li, Yaoyang Wu, Simon Fong, Antonio J. Tall\'on-Ballesteros, Xin-she Yang, Sabah Mohammed, Feng Wu
Abstract: Ensemble technique and under-sampling technique are both effective tools used for imbalanced dataset classification problems. In this paper, a novel ensemble method combining the advantages of both ensemble learning for biasing classifiers and a new under-sampling method is proposed. The under-sampling method is named Binary PSO instance selection; it gathers with ensemble classifiers to find the most suitable length and combination of the majority class samples to build a new dataset with minority class samples. The proposed method adopts multi-objective strategy, and contribution of this method is a notable improvement of the performances of imbalanced classification, and in the meantime guaranteeing a best integrity possible for the original dataset. We experimented the proposed method and compared its performance of processing imbalanced datasets with several other conventional basic ensemble methods. Experiment is also conducted on these imbalanced datasets using an improved version where ensemble classifiers are wrapped in the Binary PSO instance selection. According to experimental results, our proposed methods outperform single ensemble methods, state-of-the-art under-sampling methods, and also combinations of these methods with the traditional PSO instance selection algorithm.
Authors: Varun Dhanraj, Chris Eliasmith
Abstract: Large language models (LLMs) continue to face challenges in reliably solving reasoning tasks, particularly tasks that involve precise rule following, as often found in mathematical reasoning tasks. This paper introduces a novel neurosymbolic method that improves LLM reasoning by encoding hidden states into neurosymbolic vectors, allowing for problem-solving within a neurosymbolic vector space. The results are decoded and combined with the original hidden state, boosting the model's performance on numerical reasoning tasks. By offloading computation through neurosymbolic representations, this method improves efficiency, reliability, and interpretability. Our experimental results demonstrate an average of $82.86\%$ lower cross entropy loss and $24.50$ times more problems correctly solved on a suite of mathematical reasoning problems compared to chain-of-thought prompting and supervised fine-tuning (LoRA), while at the same time not hindering the performance of the LLM on other tasks.
Authors: Nathaniel Tomczak, Sanmukh Kuppannagari
Abstract: Transformers have demonstrated great success in numerous domains including natural language processing and bioinformatics. This success stems from the use of the attention mechanism by these models in order to represent and propagate pairwise interactions between individual tokens of sequential data. However, the primary limitation of this operation is its quadratic memory and time complexity in relation to the input's context length - the length of a sequence over which the interactions need to be captured. This significantly limits the length of sequences that can be inferred upon by these models. Extensive research has been conducted to reduce the number of pairwise interactions to sub-quadratic in relation to the context length by introducing sparsity into the attention mechanism through the development of sparse attention masks. However, efficient implementations that achieve "true sparsity" are lacking. In this work, we address this issue by proposing a graph computing view of attention where tokens are perceived as nodes of the graph and the attention mask determines the edges of the graph. Using this view, we develop graph processing algorithms to implement the attention mechanism. Both theoretically and empirically, we demonstrate that our algorithms only perform the needed computations, i.e., they are work optimal. We also perform extensive experimentation using popular attention masks to explore the impact of sparsity on execution time and achievable context length. Our experiments demonstrate significant speedups in execution times compared to state-of-the-art attention implementations such as FlashAttention for large sequence lengths. We also demonstrate that our algorithms are able to achieve extremely long sequence lengths of as high as 160 million on a single NVIDIA A100 GPU (SXM4 80GB).
Authors: Hao Liu (Deakin University), Yong Ge (The University of Arizona)
Abstract: Employee turnover refers to an individual's termination of employment from the current organization. It is one of the most persistent challenges for firms, especially those ones in Information Technology (IT) industry that confront high turnover rates. Effective prediction of potential employee turnovers benefits multiple stakeholders such as firms and online recruiters. Prior studies have focused on either the turnover prediction within a single firm or the aggregated employee movement among firms. How to predict the individual employees' turnovers among multiple firms has gained little attention in literature, and thus remains a great research challenge. In this study, we propose a novel deep learning approach based on job embeddedness theory to predict the turnovers of individual employees across different firms. Through extensive experimental evaluations using a real-world dataset, our developed method demonstrates superior performance over several state-of-the-art benchmark methods. Additionally, we estimate the cost saving for recruiters by using our turnover prediction solution and interpret the attributions of various driving factors to employee's turnover to showcase its practical business value.
Authors: Jie Ren, Yuhang Zhang, Dongrui Liu, Xiaopeng Zhang, Qi Tian
Abstract: Direct preference optimization (DPO) has shown success in aligning diffusion models with human preference. Previous approaches typically assume a consistent preference label between final generations and noisy samples at intermediate steps, and directly apply DPO to these noisy samples for fine-tuning. However, we theoretically identify inherent issues in this assumption and its impacts on the effectiveness of preference alignment. We first demonstrate the inherent issues from two perspectives: gradient direction and preference order, and then propose a Tailored Preference Optimization (TailorPO) framework for aligning diffusion models with human preference, underpinned by some theoretical insights. Our approach directly ranks intermediate noisy samples based on their step-wise reward, and effectively resolves the gradient direction issues through a simple yet efficient design. Additionally, we incorporate the gradient guidance of diffusion models into preference alignment to further enhance the optimization effectiveness. Experimental results demonstrate that our method significantly improves the model's ability to generate aesthetically pleasing and human-preferred images.
Authors: Chenlu Ding, Jiancan Wu, Yancheng Yuan, Junfeng Fang, Cunchun Li, Xiang Wang, Xiangnan He
Abstract: In the realm of online digital advertising, conversion rate (CVR) prediction plays a pivotal role in maximizing revenue under cost-per-conversion (CPA) models, where advertisers are charged only when users complete specific actions, such as making a purchase. A major challenge in CVR prediction lies in the delayed feedback problem-conversions may occur hours or even weeks after initial user interactions. This delay complicates model training, as recent data may be incomplete, leading to biases and diminished performance. Although existing methods attempt to address this issue, they often fall short in adapting to evolving user behaviors and depend on auxiliary models, which introduces computational inefficiencies and the risk of model inconsistency. In this work, we propose an Influence Function-empowered framework for Delayed Feedback Modeling (IF-DFM). IF-DFM leverages influence functions to estimate how newly acquired and delayed conversion data impact model parameters, enabling efficient parameter updates without the need for full retraining. Additionally, we present a scalable algorithm that efficiently computes parameter updates by reframing the inverse Hessian-vector product as an optimization problem, striking a balance between computational efficiency and effectiveness. Extensive experiments on benchmark datasets demonstrate that IF-DFM consistently surpasses state-of-the-art methods, significantly enhancing both prediction accuracy and model adaptability.
Authors: Yunke Wang, Yanxi Li, Chang Xu
Abstract: AI Scaling has traditionally been synonymous with Scaling Up, which builds larger and more powerful models. However, the growing demand for efficiency, adaptability, and collaboration across diverse applications necessitates a broader perspective. This position paper presents a holistic framework for AI scaling, encompassing Scaling Up, Scaling Down, and Scaling Out. It argues that while Scaling Up of models faces inherent bottlenecks, the future trajectory of AI scaling lies in Scaling Down and Scaling Out. These paradigms address critical technical and societal challenges, such as reducing carbon footprint, ensuring equitable access, and enhancing cross-domain collaboration. We explore transformative applications in healthcare, smart manufacturing, and content creation, demonstrating how AI Scaling can enable breakthroughs in efficiency, personalization, and global connectivity. Additionally, we highlight key challenges, including balancing model complexity with interpretability, managing resource constraints, and fostering ethical development. By synthesizing these approaches, we propose a unified roadmap that redefines the future of AI research and application, paving the way for advancements toward Artificial General Intelligence (AGI).
Authors: Yihe Wang, Nan Huang, Nadia Mammone, Marco Cecchi, Xiang Zhang
Abstract: Electroencephalogram (EEG) provides a non-invasive, highly accessible, and cost-effective solution for Alzheimer's Disease (AD) detection. However, existing methods, whether based on manual feature extraction or deep learning, face two major challenges: the lack of large-scale datasets for robust feature learning and evaluation, and poor detection performance due to inter-subject variations. To address these challenges, we curate an EEG-AD corpus containing 813 subjects, which forms the world's largest EEG-AD dataset to the best of our knowledge. Using this unique dataset, we propose LEAD, the first large foundation model for EEG-based AD detection. Our method encompasses an entire pipeline, from data selection and preprocessing to self-supervised contrastive pretraining, fine-tuning, and key setups such as subject-independent evaluation and majority voting for subject-level detection. We pre-train the model on 11 EEG datasets and unified fine-tune it on 5 AD datasets. Our self-supervised pre-training design includes sample-level and subject-level contrasting to extract useful general EEG features. Fine-tuning is performed on 5 channel-aligned datasets together. The backbone encoder incorporates temporal and channel embeddings to capture features across both temporal and spatial dimensions. Our method demonstrates outstanding AD detection performance, achieving up to a 9.86% increase in F1 score at the sample-level and up to a 9.31% at the subject-level compared to state-of-the-art methods. The results of our model strongly confirm the effectiveness of contrastive pre-training and channel-aligned unified fine-tuning for addressing inter-subject variation. The source code is at https://github.com/DL4mHealth/LEAD.
Authors: Kamal Acharya, Mehul Lad, Liang Sun, Houbing Song
Abstract: Travel demand prediction is crucial for optimizing transportation planning, resource allocation, and infrastructure development, ensuring efficient mobility and economic sustainability. This study introduces a Neurosymbolic Artificial Intelligence (Neurosymbolic AI) framework that integrates decision tree (DT)-based symbolic rules with neural networks (NNs) to predict travel demand, leveraging the interpretability of symbolic reasoning and the predictive power of neural learning. The framework utilizes data from diverse sources, including geospatial, economic, and mobility datasets, to build a comprehensive feature set. DTs are employed to extract interpretable if-then rules that capture key patterns, which are then incorporated as additional features into a NN to enhance its predictive capabilities. Experimental results show that the combined dataset, enriched with symbolic rules, consistently outperforms standalone datasets across multiple evaluation metrics, including Mean Absolute Error (MAE), \(R^2\), and Common Part of Commuters (CPC). Rules selected at finer variance thresholds (e.g., 0.0001) demonstrate superior effectiveness in capturing nuanced relationships, reducing prediction errors, and aligning with observed commuter patterns. By merging symbolic and neural learning paradigms, this Neurosymbolic approach achieves both interpretability and accuracy.
Authors: Ziyang Zheng, Shan Huang, Jianyuan Zhong, Zhengyuan Shi, Guohao Dai, Ningyi Xu, Qiang Xu
Abstract: Circuit representation learning has become pivotal in electronic design automation, enabling critical tasks such as testability analysis, logic reasoning, power estimation, and SAT solving. However, existing models face significant challenges in scaling to large circuits due to limitations like over-squashing in graph neural networks and the quadratic complexity of transformer-based models. To address these issues, we introduce DeepGate4, a scalable and efficient graph transformer specifically designed for large-scale circuits. DeepGate4 incorporates several key innovations: (1) an update strategy tailored for circuit graphs, which reduce memory complexity to sub-linear and is adaptable to any graph transformer; (2) a GAT-based sparse transformer with global and local structural encodings for AIGs; and (3) an inference acceleration CUDA kernel that fully exploit the unique sparsity patterns of AIGs. Our extensive experiments on the ITC99 and EPFL benchmarks show that DeepGate4 significantly surpasses state-of-the-art methods, achieving 15.5% and 31.1% performance improvements over the next-best models. Furthermore, the Fused-DeepGate4 variant reduces runtime by 35.1% and memory usage by 46.8%, making it highly efficient for large-scale circuit analysis. These results demonstrate the potential of DeepGate4 to handle complex EDA tasks while offering superior scalability and efficiency.
Authors: Srinitish Srinivasan, Omkumar CU
Abstract: Graph representation learning has emerged as a cornerstone for tasks like node classification and link prediction, yet prevailing self-supervised learning (SSL) methods face challenges such as computational inefficiency, reliance on contrastive objectives, and representation collapse. Existing approaches often depend on feature reconstruction, negative sampling, or complex decoders, which introduce training overhead and hinder generalization. Further, current techniques which address such limitations fail to account for the contribution of node embeddings to a certain prediction in the absence of labeled nodes. To address these limitations, we propose a novel joint embedding predictive framework for graph SSL that eliminates contrastive objectives and negative sampling while preserving semantic and structural information. Additionally, we introduce a semantic-aware objective term that incorporates pseudo-labels derived from Gaussian Mixture Models (GMMs), enhancing node discriminability by evaluating latent feature contributions. Extensive experiments demonstrate that our framework outperforms state-of-the-art graph SSL methods across benchmarks, achieving superior performance without contrastive loss or complex decoders. Key innovations include (1) a non-contrastive, view-invariant joint embedding predictive architecture, (2) Leveraging single context and multiple targets relationship between subgraphs, and (3) GMM-based pseudo-label scoring to capture semantic contributions. This work advances graph SSL by offering a computationally efficient, collapse-resistant paradigm that bridges spatial and semantic graph features for downstream tasks. The code for our paper can be found at https://github.com/Deceptrax123/JPEB-GSSL
Authors: Jiaxing Xu, Yongqiang Chen, Xia Dong, Mengcheng Lan, Tiancheng Huang, Qingtian Bian, James Cheng, Yiping Ke
Abstract: In neuroscience, identifying distinct patterns linked to neurological disorders, such as Alzheimer's and Autism, is critical for early diagnosis and effective intervention. Graph Neural Networks (GNNs) have shown promising in analyzing brain networks, but there are two major challenges in using GNNs: (1) distribution shifts in multi-site brain network data, leading to poor Out-of-Distribution (OOD) generalization, and (2) limited interpretability in identifying key brain regions critical to neurological disorders. Existing graph OOD methods, while effective in other domains, struggle with the unique characteristics of brain networks. To bridge these gaps, we introduce BrainOOD, a novel framework tailored for brain networks that enhances GNNs' OOD generalization and interpretability. BrainOOD framework consists of a feature selector and a structure extractor, which incorporates various auxiliary losses including an improved Graph Information Bottleneck (GIB) objective to recover causal subgraphs. By aligning structure selection across brain networks and filtering noisy features, BrainOOD offers reliable interpretations of critical brain regions. Our approach outperforms 16 existing methods and improves generalization to OOD subjects by up to 8.5%. Case studies highlight the scientific validity of the patterns extracted, which aligns with the findings in known neuroscience literature. We also propose the first OOD brain network benchmark, which provides a foundation for future research in this field. Our code is available at https://github.com/AngusMonroe/BrainOOD.
Authors: Kim Yong Tan, Yueming Lyu, Ivor Tsang, Yew-Soon Ong
Abstract: Guided diffusion-model generation is a promising direction for customizing the generation process of a pre-trained diffusion-model to address the specific downstream tasks. Existing guided diffusion models either rely on training of the guidance model with pre-collected datasets or require the objective functions to be differentiable. However, for most real-world tasks, the offline datasets are often unavailable, and their objective functions are often not differentiable, such as image generation with human preferences, molecular generation for drug discovery, and material design. Thus, we need an \textbf{online} algorithm capable of collecting data during runtime and supporting a \textbf{black-box} objective function. Moreover, the \textbf{query efficiency} of the algorithm is also critical because the objective evaluation of the query is often expensive in the real-world scenarios. In this work, we propose a novel and simple algorithm, \textbf{Fast Direct}, for query-efficient online black-box target generation. Our Fast Direct builds a pseudo-target on the data manifold to update the noise sequence of the diffusion model with a universal direction, which is promising to perform query-efficient guided generation. Extensive experiments on twelve high-resolution ($\small {1024 \times 1024}$) image target generation tasks and six 3D-molecule target generation tasks show $\textbf{6}\times$ up to $\textbf{10}\times$ query efficiency improvement and $\textbf{11}\times$ up to $\textbf{44}\times$ query efficiency improvement, respectively. Our implementation is publicly available at: https://github.com/kimyong95/guide-stable-diffusion/tree/fast-direct
URLs: https://github.com/kimyong95/guide-stable-diffusion/tree/fast-direct
Authors: Priodyuti Pradhan, Amit Reza
Abstract: In complex systems, information propagation can be defined as diffused or delocalized, weakly localized, and strongly localized. Can a machine learning model learn the behavior of a linear dynamical system on networks? In this work, we develop a graph neural network framework for identifying the steady-state behavior of the linear dynamical system. We reveal that our model learns the different states with high accuracy. To understand the explainability of our model, we provide an analytical derivation for the forward and backward propagation of our framework. Finally, we use the real-world graphs in our model for validation.
Authors: Tianlin Zhang, En Yu, Yi Shao, Shuai Li, Sujuan Hou, Jiande Sun
Abstract: Multimodal fake news detection has garnered significant attention due to its profound implications for social security. While existing approaches have contributed to understanding cross-modal consistency, they often fail to leverage modal-specific representations and explicit discrepant features. To address these limitations, we propose a Multimodal Inverse Attention Network (MIAN), a novel framework that explores intrinsic discriminative features based on news content to advance fake news detection. Specifically, MIAN introduces a hierarchical learning module that captures diverse intra-modal relationships through local-to-global and local-to-local interactions, thereby generating enhanced unimodal representations to improve the identification of fake news at the intra-modal level. Additionally, a cross-modal interaction module employs a co-attention mechanism to establish and model dependencies between the refined unimodal representations, facilitating seamless semantic integration across modalities. To explicitly extract inconsistency features, we propose an inverse attention mechanism that effectively highlights the conflicting patterns and semantic deviations introduced by fake news in both intra- and inter-modality. Extensive experiments on benchmark datasets demonstrate that MIAN significantly outperforms state-of-the-art methods, underscoring its pivotal contribution to advancing social security through enhanced multimodal fake news detection.
Authors: Mohammad Amin Hasanpour, Mikkel Kirkegaard, Xenofon Fafoutis
Abstract: The integration of artificial intelligence (AI) into embedded devices, a paradigm known as embedded artificial intelligence (eAI) or tiny machine learning (TinyML), is transforming industries by enabling intelligent data processing at the edge. However, the many tools available in this domain leave researchers and developers wondering which one is best suited to their needs. This paper provides a review of existing eAI tools, highlighting their features, trade-offs, and limitations. Additionally, we introduce EdgeMark, an open-source automation system designed to streamline the workflow for deploying and benchmarking machine learning (ML) models on embedded platforms. EdgeMark simplifies model generation, optimization, conversion, and deployment while promoting modularity, reproducibility, and scalability. Experimental benchmarking results showcase the performance of widely used eAI tools, including TensorFlow Lite Micro (TFLM), Edge Impulse, Ekkono, and Renesas eAI Translator, across a wide range of models, revealing insights into their relative strengths and weaknesses. The findings provide guidance for researchers and developers in selecting the most suitable tools for specific application requirements, while EdgeMark lowers the barriers to adoption of eAI technologies.
Authors: Cl\'ement Lalanne (IMT, ANITI), Jean-Michel Loubes (IMT, ANITI), David Rodr\'iguez-V\'itores (UVa, IMUVA)
Abstract: In this work, we introduce a novel framework for privately optimizing objectives that rely on Wasserstein distances between data-dependent empirical measures. Our main theoretical contribution is, based on an explicit formulation of the Wasserstein gradient in a fully discrete setting, a control on the sensitivity of this gradient to individual data points, allowing strong privacy guarantees at minimal utility cost. Building on these insights, we develop a deep learning approach that incorporates gradient and activations clipping, originally designed for DP training of problems with a finite-sum structure. We further demonstrate that privacy accounting methods extend to Wasserstein-based objectives, facilitating large-scale private training. Empirical results confirm that our framework effectively balances accuracy and privacy, offering a theoretically sound solution for privacy-preserving machine learning tasks relying on optimal transport distances such as Wasserstein distance or sliced-Wasserstein distance.
Authors: Christopher E. Mower, Haitham Bou-Ammar
Abstract: Inferring physical laws from data is a central challenge in science and engineering, including but not limited to healthcare, physical sciences, biosciences, social sciences, sustainability, climate, and robotics. Deep networks offer high-accuracy results but lack interpretability, prompting interest in models built from simple components. The Sparse Identification of Nonlinear Dynamics (SINDy) method has become the go-to approach for building such modular and interpretable models. SINDy leverages sparse regression with L1 regularization to identify key terms from a library of candidate functions. However, SINDy's choice of candidate library and optimization method requires significant technical expertise, limiting its widespread applicability. This work introduces Al-Khwarizmi, a novel agentic framework for physical law discovery from data, which integrates foundational models with SINDy. Leveraging LLMs, VLMs, and Retrieval-Augmented Generation (RAG), our approach automates physical law discovery, incorporating prior knowledge and iteratively refining candidate solutions via reflection. Al-Khwarizmi operates in two steps: it summarizes system observations-comprising textual descriptions, raw data, and plots-followed by a secondary step that generates candidate feature libraries and optimizer configurations to identify hidden physics laws correctly. Evaluating our algorithm on over 198 models, we demonstrate state-of-the-art performance compared to alternatives, reaching a 20 percent increase against the best-performing alternative.
Authors: Moses Ananta, Muhammad Farid Adilazuarda, Zayd Muhammad Kawakibi Zuhri, Ayu Purwarianti, Alham Fikri Aji
Abstract: Fine-tuning large language models (LLMs) is often constrained by the computational costs of processing massive datasets. We propose \textbf{QLESS} (Quantized Low-rank Gradient Similarity Search), which integrates gradient quantization with the LESS framework to enable memory-efficient data valuation and selection. QLESS employs a two-step compression process: first, it obtains low-dimensional gradient representations through LoRA-based random projection; then, it quantizes these gradients to low-bitwidth representations. Experiments on multiple LLM architectures (LLaMA, Mistral, Qwen) and benchmarks (MMLU, BBH, TyDiQA) show that QLESS achieves comparable data selection performance to LESS while reducing memory usage by up to 16x. Even 1-bit gradient quantization preserves data valuation quality. These findings underscore QLESS as a practical, scalable approach to identifying informative examples within strict memory constraints.
Authors: Xianglong Yan, Tianao Zhang, Zhiteng Li, Yulun Zhang
Abstract: Large language models (LLMs) have achieved remarkable success in natural language processing tasks, but their high computational and memory demands pose challenges for deployment on resource-constrained devices. Binarization, as an efficient compression method that reduces model weights to just 1 bit, significantly lowers both computational and memory requirements. Despite this, the binarized LLM still contains redundancy, which can be further compressed. Semi-structured pruning provides a promising approach to achieve this, which offers a better trade-off between model performance and hardware efficiency. However, simply combining binarization with semi-structured pruning can lead to a significant performance drop. To address this issue, we propose a Progressive Binarization with Semi-Structured Pruning (PBS$^2$P) method for LLM compression. We first propose a Stepwise semi-structured Pruning with Binarization Optimization (SPBO). Our optimization strategy significantly reduces the total error caused by pruning and binarization, even below that of the no-pruning scenario. Furthermore, we design a Coarse-to-Fine Search (CFS) method to select pruning elements more effectively. Extensive experiments demonstrate that PBS$^2$P achieves superior accuracy across various LLM families and evaluation metrics, noticeably outperforming state-of-the-art (SOTA) binary PTQ methods. The code and models will be available at https://github.com/XIANGLONGYAN/PBS2P.
Authors: Xiuzhan Guo
Abstract: In this paper, we study the machine learning elements which we are interested in together as a machine learning system, consisting of a collection of machine learning elements and a collection of relations between the elements. The relations we concern are algebraic operations, binary relations, and binary relations with composition that can be reasoned categorically. A machine learning system transformation between two systems is a map between the systems, which preserves the relations we concern. The system transformations given by quotient or clustering, representable functor, and Yoneda embedding are highlighted and discussed by machine learning examples. An adjunction between machine learning systems, a special machine learning system transformation loop, provides the optimal way of solving problems. Machine learning system transformations are linked and compared by their maps at 2-cell, natural transformations. New insights and structures can be obtained from universal properties and algebraic structures given by monads, which are generated from adjunctions.
Authors: Martin Genzel, Patrick Putzky, Pengfei Zhao, Sebastian Schulze, Mattes Mollenhauer, Robert Seidel, Stefan Dietzel, Thomas Wollmann
Abstract: The adoption of Foundation Models in resource-constrained environments remains challenging due to their large size and inference costs. A promising way to overcome these limitations is post-training compression, which aims to balance reduced model size against performance degradation. This work presents Any Compression via Iterative Pruning (ACIP), a novel algorithmic approach to determine a compression-performance trade-off from a single stochastic gradient descent run. To ensure parameter efficiency, we use an SVD-reparametrization of linear layers and iteratively prune their singular values with a sparsity-inducing penalty. The resulting pruning order gives rise to a global parameter ranking that allows us to materialize models of any target size. Importantly, the compressed models exhibit strong predictive downstream performance without the need for costly fine-tuning. We evaluate ACIP on a large selection of open-weight LLMs and tasks, and demonstrate state-of-the-art results compared to existing factorisation-based compression methods. We also show that ACIP seamlessly complements common quantization-based compression techniques.
Authors: Dmitry Manning-Coe, Jacopo Gliozzi, Alexander G. Stapleton, Edward Hirst, Giuseppe De Tomasi, Barry Bradlyn, David S. Berman
Abstract: Grokking typically achieves similar loss to ordinary, "steady", learning. We ask whether these different learning paths - grokking versus ordinary training - lead to fundamental differences in the learned models. To do so we compare the features, compressibility, and learning dynamics of models trained via each path in two tasks. We find that grokked and steadily trained models learn the same features, but there can be large differences in the efficiency with which these features are encoded. In particular, we find a novel "compressive regime" of steady training in which there emerges a linear trade-off between model loss and compressibility, and which is absent in grokking. In this regime, we can achieve compression factors 25x times the base model, and 5x times the compression achieved in grokking. We then track how model features and compressibility develop through training. We show that model development in grokking is task-dependent, and that peak compressibility is achieved immediately after the grokking plateau. Finally, novel information-geometric measures are introduced which demonstrate that models undergoing grokking follow a straight path in information space.
Authors: Shuangyi Chen, Yuanxin Guo, Yue Ju, Harik Dalal, Ashish Khisti
Abstract: Parameter-Efficient Fine-Tuning (PEFT) methods like Low-Rank Adaptation (LoRA) optimize federated training by reducing computational and communication costs. We propose RoLoRA, a federated framework using alternating optimization to fine-tune LoRA adapters. Our approach emphasizes the importance of learning up and down projection matrices to enhance expressiveness and robustness. We use both theoretical analysis and extensive experiments to demonstrate the advantages of RoLoRA over prior approaches that either generate imperfect model updates or limit expressiveness of the model. We present theoretical analysis on a simplified linear model to demonstrate the importance of learning both down-projection and up-projection matrices in LoRA. We provide extensive experimental evaluations on a toy neural network on MNIST as well as large language models including RoBERTa-Large, Llama-2-7B on diverse tasks to demonstrate the advantages of RoLoRA over other methods.
Authors: Thomas T. Zhang, Behrad Moniri, Ansh Nagwekar, Faraz Rahman, Anton Xue, Hamed Hassani, Nikolai Matni
Abstract: Layer-wise preconditioning methods are a family of memory-efficient optimization algorithms that introduce preconditioners per axis of each layer's weight tensors. These methods have seen a recent resurgence, demonstrating impressive performance relative to entry-wise ("diagonal") preconditioning methods such as Adam(W) on a wide range of neural network optimization tasks. Complementary to their practical performance, we demonstrate that layer-wise preconditioning methods are provably necessary from a statistical perspective. To showcase this, we consider two prototypical models, linear representation learning and single-index learning, which are widely used to study how typical algorithms efficiently learn useful features to enable generalization. In these problems, we show SGD is a suboptimal feature learner when extending beyond ideal isotropic inputs $\mathbf{x} \sim \mathsf{N}(\mathbf{0}, \mathbf{I})$ and well-conditioned settings typically assumed in prior work. We demonstrate theoretically and numerically that this suboptimality is fundamental, and that layer-wise preconditioning emerges naturally as the solution. We further show that standard tools like Adam preconditioning and batch-norm only mildly mitigate these issues, supporting the unique benefits of layer-wise preconditioning.
Authors: Mark Horton, Tergel Molom-Ochir, Peter Liu, Bhavna Gopal, Chiyue Wei, Cong Guo, Brady Taylor, Deliang Fan, Shan X. Wang, Hai Li, Yiran Chen
Abstract: Pre-trained transformer models with extended context windows are notoriously expensive to run at scale, often limiting real-world deployment due to their high computational and memory requirements. In this paper, we introduce Hamming Attention Distillation (HAD), a novel framework that binarizes keys and queries in the attention mechanism to achieve significant efficiency gains. By converting keys and queries into {-1, +1} vectors and replacing dot-product operations with efficient Hamming distance computations, our method drastically reduces computational overhead. Additionally, we incorporate attention matrix sparsification to prune low-impact activations, which further reduces the cost of processing long-context sequences. \par Despite these aggressive compression strategies, our distilled approach preserves a high degree of representational power, leading to substantially improved accuracy compared to prior transformer binarization methods. We evaluate HAD on a range of tasks and models, including the GLUE benchmark, ImageNet, and QuALITY, demonstrating state-of-the-art performance among binarized Transformers while drastically reducing the computational costs of long-context inference. \par We implement HAD in custom hardware simulations, demonstrating superior performance characteristics compared to a custom hardware implementation of standard attention. HAD achieves just $\mathbf{1.78}\%$ performance losses on GLUE compared to $9.08\%$ in state-of-the-art binarization work, and $\mathbf{2.5}\%$ performance losses on ImageNet compared to $12.14\%$, all while targeting custom hardware with a $\mathbf{79}\%$ area reduction and $\mathbf{87}\%$ power reduction compared to its standard attention counterpart.
Authors: Breno W. Carvalho, Artur S. d'Avila Garcez, Lu\'is C. Lamb, Em\'ilio Vital Brazil
Abstract: Grokking, or delayed generalization, is an intriguing learning phenomenon where test set loss decreases sharply only after a model's training set loss has converged. This challenges conventional understanding of the training dynamics in deep learning networks. In this paper, we formalize and investigate grokking, highlighting that a key factor in its emergence is a distribution shift between training and test data. We introduce two synthetic datasets specifically designed to analyze grokking. One dataset examines the impact of limited sampling, and the other investigates transfer learning's role in grokking. By inducing distribution shifts through controlled imbalanced sampling of sub-categories, we systematically reproduce the phenomenon, demonstrating that while small-sampling is strongly associated with grokking, it is not its cause. Instead, small-sampling serves as a convenient mechanism for achieving the necessary distribution shift. We also show that when classes form an equivariant map, grokking can be explained by the model's ability to learn from similar classes or sub-categories. Unlike earlier work suggesting that grokking primarily arises from high regularization and sparse data, we demonstrate that it can also occur with dense data and minimal hyper-parameter tuning. Our findings deepen the understanding of grokking and pave the way for developing better stopping criteria in future training processes.
Authors: Martijn Bartelds, Ananjan Nandi, Moussa Koulako Bala Doumbouya, Dan Jurafsky, Tatsunori Hashimoto, Karen Livescu
Abstract: Modern deep learning models often achieve high overall performance, but consistently fail on specific subgroups. Group distributionally robust optimization (group DRO) addresses this problem by minimizing the worst-group loss, but it fails when group losses misrepresent performance differences between groups. This is common in domains like speech, where the widely used connectionist temporal classification (CTC) loss scales with input length and varies with linguistic and acoustic properties, leading to spurious differences between group losses. We present CTC-DRO, which addresses the shortcomings of the group DRO objective by smoothing the group weight update to prevent overemphasis on consistently high-loss groups, while using input length-matched batching to mitigate CTC's scaling issues. We evaluate CTC-DRO on the task of multilingual automatic speech recognition (ASR) across five language sets from the ML-SUPERB 2.0 benchmark. CTC-DRO consistently outperforms group DRO and CTC-based baseline models, reducing the worst-language error by up to 65.9% and the average error by up to 47.7%. CTC-DRO can be applied to ASR with minimal computational costs, and offers the potential for reducing group disparities in other domains with similar challenges.
Authors: Stavros Orfanoudakis, Nanda Kishor Panda, Peter Palensky, Pedro P. Vergara
Abstract: Reinforcement Learning (RL) methods used for solving real-world optimization problems often involve dynamic state-action spaces, larger scale, and sparse rewards, leading to significant challenges in convergence, scalability, and efficient exploration of the solution space. This study introduces GNN-DT, a novel Decision Transformer (DT) architecture that integrates Graph Neural Network (GNN) embedders with a novel residual connection between input and output tokens crucial for handling dynamic environments. By learning from previously collected trajectories, GNN-DT reduces dependence on accurate simulators and tackles the sparse rewards limitations of online RL algorithms. We evaluate GNN-DT on the complex electric vehicle (EV) charging optimization problem and prove that its performance is superior and requires significantly fewer training trajectories, thus improving sample efficiency compared to existing DT baselines. Furthermore, GNN-DT exhibits robust generalization to unseen environments and larger action spaces, addressing a critical gap in prior DT-based approaches
Authors: Shuchen Wu, Stephan Alaniz, Eric Schulz, Zeynep Akata
Abstract: Understanding neural networks is challenging due to their high-dimensional, interacting components. Inspired by human cognition, which processes complex sensory data by chunking it into recurring entities, we propose leveraging this principle to interpret artificial neural population activities. Biological and artificial intelligence share the challenge of learning from structured, naturalistic data, and we hypothesize that the cognitive mechanism of chunking can provide insights into artificial systems. We first demonstrate this concept in recurrent neural networks (RNNs) trained on artificial sequences with imposed regularities, observing that their hidden states reflect these patterns, which can be extracted as a dictionary of chunks that influence network responses. Extending this to large language models (LLMs) like LLaMA, we identify similar recurring embedding states corresponding to concepts in the input, with perturbations to these states activating or inhibiting the associated concepts. By exploring methods to extract dictionaries of identifiable chunks across neural embeddings of varying complexity, our findings introduce a new framework for interpreting neural networks, framing their population activity as structured reflections of the data they process.
Authors: Pierre Ablin, Angelos Katharopoulos, Skyler Seto, David Grangier
Abstract: Machine learning models are routinely trained on a mixture of different data domains. Different domain weights yield very different downstream performances. We propose the Soup-of-Experts, a novel architecture that can instantiate a model at test time for any domain weights with minimal computational cost and without re-training the model. Our architecture consists of a bank of expert parameters, which are linearly combined to instantiate one model. We learn the linear combination coefficients as a function of the input domain weights. To train this architecture, we sample random domain weights, instantiate the corresponding model, and backprop through one batch of data sampled with these domain weights. We demonstrate how our approach obtains small specialized models on several language modeling tasks quickly. Soup-of-Experts are particularly appealing when one needs to ship many different specialist models quickly under a model size constraint.
Authors: Jianming Huang, Hiroyuki Kasai
Abstract: Graph data, with its structurally variable nature, represents complex real-world phenomena like chemical compounds, protein structures, and social networks. Traditional Graph Neural Networks (GNNs) primarily utilize the message-passing mechanism, but their expressive power is limited and their prediction lacks explainability. To address these limitations, researchers have focused on graph substructures. Subgraph neural networks (SGNNs) and GNN explainers have emerged as potential solutions, but each has its limitations. SGNNs computes graph representations based on the bags of subgraphs to enhance the expressive power. However, they often rely on predefined algorithm-based sampling strategies, which is inefficient. GNN explainers adopt data-driven approaches to generate important subgraphs to provide explanation. Nevertheless, their explanation is difficult to be translated into practical improvements on GNNs. To overcome these issues, we propose a novel self-supervised framework that integrates SGNNs with the generation approach of GNN explainers, named the Reinforcement Walk Exploration SGNN (RWE-SGNN). Our approach features a sampling model trained in an explainer fashion, optimizing subgraphs to enhance model performance. To achieve a data-driven sampling approach, unlike traditional subgraph generation approaches, we propose a novel walk exploration process, which efficiently extracts important substructures, simplifying the embedding process and avoiding isomorphism problems. Moreover, we prove that our proposed walk exploration process has equivalent generation capability to the traditional subgraph generation process. Experimental results on various graph datasets validate the effectiveness of our proposed method, demonstrating significant improvements in performance and precision.
Authors: Hanyang Zhao, Haoxian Chen, Ji Zhang, David D. Yao, Wenpin Tang
Abstract: Reinforcement learning from human feedback (RLHF), which aligns a diffusion model with input prompt, has become a crucial step in building reliable generative AI models. Most works in this area use a discrete-time formulation, which is prone to induced errors, and often not applicable to models with higher-order/black-box solvers. The objective of this study is to develop a disciplined approach to fine-tune diffusion models using continuous-time RL, formulated as a stochastic control problem with a reward function that aligns the end result (terminal state) with input prompt. The key idea is to treat score matching as controls or actions, and thereby making connections to policy optimization and regularization in continuous-time RL. To carry out this idea, we lay out a new policy optimization framework for continuous-time RL, and illustrate its potential in enhancing the value networks design space via leveraging the structural property of diffusion models. We validate the advantages of our method by experiments in downstream tasks of fine-tuning large-scale Text2Image models of Stable Diffusion v1.5.
Authors: Eric Zhao, Pranjal Awasthi, Sreenivas Gollapudi
Abstract: Sampling-based search, a simple paradigm for utilizing test-time compute, involves generating multiple candidate responses and selecting the best one -- typically by verifying each response for correctness. In this paper, we study the scaling trends governing sampling-based search. Among our findings is that simply scaling up a minimalist implementation that uses only random sampling and direct self-verification results in sustained performance improvements that, for example, elevate the Gemini v1.5 Pro model's reasoning capabilities past that of o1-Preview on popular benchmarks. We partially attribute the scalability of sampling-based search to a phenomenon of implicit scaling, where sampling a larger pool of responses in turn improves verification accuracy. We further identify two useful principles for improving self-verification capabilities with test-time compute: (1) comparing across responses provides helpful signals about the locations of errors and hallucinations, and (2) different model output styles are useful for different contexts -- chains of thought are useful for reasoning but harder to verify. We also find that, though accurate verification can be elicited, frontier models demonstrate remarkably weak out-of-box verification capabilities and introduce a benchmark to measure progress on these deficiencies.
Authors: Vicky Kouni
Abstract: We present a new ensemble framework for boosting the performance of overparameterized unfolding networks solving the compressed sensing problem. We combine a state-of-the-art overparameterized unfolding network with a continuation technique, to warm-start a crucial quantity of the said network's architecture; we coin the resulting continued network C-DEC. Moreover, for training and evaluating C-DEC, we incorporate the log-cosh loss function, which enjoys both linear and quadratic behavior. Finally, we numerically assess C-DEC's performance on real-world images. Results showcase that the combination of continuation with the overparameterized unfolded architecture, trained and evaluated with the chosen loss function, yields smoother loss landscapes and improved reconstruction and generalization performance of C-DEC, consistently for all datasets.
Authors: Ethan Harvey, Mikhail Petrov, Michael C. Hughes
Abstract: When training large flexible models, practitioners often rely on grid search to select hyperparameters that control over-fitting. This grid search has several disadvantages: the search is computationally expensive, requires carving out a validation set that reduces the available data for training, and requires users to specify candidate values. In this paper, we propose an alternative: directly learning regularization hyperparameters on the full training set via the evidence lower bound ("ELBo") objective from variational methods. For deep neural networks with millions of parameters, we recommend a modified ELBo that upweights the influence of the data likelihood relative to the prior. Our proposed technique overcomes all three disadvantages of grid search. In a case study on transfer learning of image classifiers, we show how our method reduces the 88+ hour grid search of past work to under 3 hours while delivering comparable accuracy. We further demonstrate how our approach enables efficient yet accurate approximations of Gaussian processes with learnable length-scale kernels.
Authors: Boyan Gao, Bo Zhao, Shreyank N Gowda, Xingrun Xing, Yibo Yang, Timothy Hospedales, David A. Clifton
Abstract: Dataset condensation aims to synthesize datasets with a few representative samples that can effectively represent the original datasets. This enables efficient training and produces models with performance close to those trained on the original sets. Most existing dataset condensation methods conduct dataset learning under the bilevel (inner- and outer-loop) based optimization. However, the preceding methods perform with limited dataset generalization due to the notoriously complicated loss landscape and expensive time-space complexity of the inner-loop unrolling of bilevel optimization. These issues deteriorate when the datasets are learned via matching the trajectories of networks trained on the real and synthetic datasets with a long horizon inner-loop. To address these issues, we introduce Sharpness-Aware Trajectory Matching (SATM), which enhances the generalization capability of learned synthetic datasets by optimising the sharpness of the loss landscape and objective simultaneously. Moreover, our approach is coupled with an efficient hypergradient approximation that is mathematically well-supported and straightforward to implement along with controllable computational overhead. Empirical evaluations of SATM demonstrate its effectiveness across various applications, including in-domain benchmarks and out-of-domain settings. Moreover, its easy-to-implement properties afford flexibility, allowing it to integrate with other advanced sharpness-aware minimizers. Our code will be released.
Authors: Edoardo Urettini, Antonio Carta
Abstract: Online Continual Learning (OCL) models continuously adapt to nonstationary data streams, usually without task information. These settings are complex and many traditional CL methods fail, while online methods (mainly replay-based) suffer from instabilities after the task shift. To address this issue, we formalize replay-based OCL as a second-order online joint optimization with explicit KL-divergence constraints on replay data. We propose Online Curvature-Aware Replay (OCAR) to solve the problem: a method that leverages second-order information of the loss using a K-FAC approximation of the Fisher Information Matrix (FIM) to precondition the gradient. The FIM acts as a stabilizer to prevent forgetting while also accelerating the optimization in non-interfering directions. We show how to adapt the estimation of the FIM to a continual setting stabilizing second-order optimization for non-iid data, uncovering the role of the Tikhonov regularization in the stability-plasticity tradeoff. Empirical results show that OCAR outperforms state-of-the-art methods in continual metrics achieving higher average accuracy throughout the training process in three different benchmarks.
Authors: Anastasiia Soboleva, Andrey Pudovikov, Roman Snetkov, Alina Babenko, Egor Samosvat, Yuriy Dorn
Abstract: Online advertising platforms often face a common challenge: the cold start problem. Insufficient behavioral data (clicks) makes accurate click-through rate (CTR) forecasting of new ads challenging. CTR for "old" items can also be significantly underestimated due to their early performance influencing their long-term behavior on the platform. The cold start problem has far-reaching implications for businesses, including missed long-term revenue opportunities. To mitigate this issue, we developed a UCB-like algorithm under multi-armed bandit (MAB) setting for positional-based model (PBM), specifically tailored to auction pay-per-click systems. Our proposed algorithm successfully combines theory and practice: we obtain theoretical upper estimates of budget regret, and conduct a series of experiments on synthetic and real-world data that confirm the applicability of the method on the real platform. In addition to increasing the platform's long-term profitability, we also propose a mechanism for maintaining short-term profits through controlled exploration and exploitation of items.
Authors: Yihan Du, Anna Winnicki, Gal Dalal, Shie Mannor, R. Srikant
Abstract: Standard reinforcement learning (RL) assumes that an agent can observe a reward for each state-action pair. However, in practical applications, it is often difficult and costly to collect a reward for each state-action pair. While there have been several works considering RL with trajectory feedback, it is unclear if trajectory feedback is inefficient for learning when trajectories are long. In this work, we consider a model named RL with segment feedback, which offers a general paradigm filling the gap between per-state-action feedback and trajectory feedback. In this model, we consider an episodic Markov decision process (MDP), where each episode is divided into $m$ segments, and the agent observes reward feedback only at the end of each segment. Under this model, we study two popular feedback settings: binary feedback and sum feedback, where the agent observes a binary outcome and a reward sum according to the underlying reward function, respectively. To investigate the impact of the number of segments $m$ on learning performance, we design efficient algorithms and establish regret upper and lower bounds for both feedback settings. Our theoretical and experimental results show that: under binary feedback, increasing the number of segments $m$ decreases the regret at an exponential rate; in contrast, surprisingly, under sum feedback, increasing $m$ does not reduce the regret significantly.
Authors: Yipu Zhang, Likai Wang, Kuan-Jui Su, Aiying Zhang, Hao Zhu, Xiaowen Liu, Hui Shen, Vince D. Calhoun, Yuping Wang, Hongwen Deng
Abstract: Resting-state functional magnetic resonance imaging (rs-fMRI) and its derived functional connectivity networks (FCNs) have become critical for understanding neurological disorders. However, collaborative analyses and the generalizability of models still face significant challenges due to privacy regulations and the non-IID (non-independent and identically distributed) property of multiple data sources. To mitigate these difficulties, we propose Domain Adversarial Federated Learning (DAFed), a novel federated deep learning framework specifically designed for non-IID fMRI data analysis in multi-site settings. DAFed addresses these challenges through feature disentanglement, decomposing the latent feature space into domain-invariant and domain-specific components, to ensure robust global learning while preserving local data specificity. Furthermore, adversarial training facilitates effective knowledge transfer between labeled and unlabeled datasets, while a contrastive learning module enhances the global representation of domain-invariant features. We evaluated DAFed on the diagnosis of ASD and further validated its generalizability in the classification of AD, demonstrating its superior classification accuracy compared to state-of-the-art methods. Additionally, an enhanced Score-CAM module identifies key brain regions and functional connectivity significantly associated with ASD and MCI, respectively, uncovering shared neurobiological patterns across sites. These findings highlight the potential of DAFed to advance multi-site collaborative research in neuroimaging while protecting data confidentiality.
Authors: Peter Chen, Yue Xie, Qingpeng Zhang
Abstract: Optimal Transport (OT) theory seeks to determine the map $T:X \to Y$ that transports a source measure $P$ to a target measure $Q$, minimizing the cost $c(\mathbf{x}, T(\mathbf{x}))$ between $\mathbf{x}$ and its image $T(\mathbf{x})$. Building upon the Input Convex Neural Network OT solver and incorporating the concept of displacement-sparse maps, we introduce a sparsity penalty into the minimax Wasserstein formulation, promote sparsity in displacement vectors $\Delta(\mathbf{x}) := T(\mathbf{x}) - \mathbf{x}$, and enhance the interpretability of the resulting map. However, increasing sparsity often reduces feasibility, causing $T_{\#}(P)$ to deviate more significantly from the target measure. In low-dimensional settings, we propose a heuristic framework to balance the trade-off between sparsity and feasibility by dynamically adjusting the sparsity intensity parameter during training. For high-dimensional settings, we directly constrain the dimensionality of displacement vectors by enforcing $\dim(\Delta(\mathbf{x})) \leq l$, where $l < d$ for $X \subseteq \mathbb{R}^d$. Among maps satisfying this constraint, we aim to identify the most feasible one. This goal can be effectively achieved by adapting our low-dimensional heuristic framework without resorting to dimensionality reduction. We validate our method on both synthesized sc-RNA and real 4i cell perturbation datasets, demonstrating improvements over existing methods.
Authors: Kemal Kurniawan, Meladel Mistica, Timothy Baldwin, Jey Han Lau
Abstract: Human label variation (HLV) challenges the standard assumption that an example has a single ground truth, instead embracing the natural variation in human labelling to train and evaluate models. While various training methods and metrics for HLV have been proposed, there has been no systematic meta-evaluation of HLV evaluation metrics, contributing to the lack of clarity in the best HLV training method. We propose new evaluation metrics and training methods and empirically meta-evaluate HLV evaluation metrics. We find that training on either disaggregated annotations or soft labels often performs best across metrics, and that our proposed soft metric correlates best with human preference.
Authors: Arian Eamaz, Farhang Yeganegi, Mojtaba Soltanalian
Abstract: Recent advancements in Large Language Model (LLM) compression, such as BitNet and BitNet b1.58, have marked significant strides in reducing the computational demands of LLMs through innovative one-bit quantization techniques. We extend this frontier by looking at Large Inference Models (LIMs) that have become indispensable across various applications. However, their scale and complexity often come at a significant computational cost. We introduce a novel approach that leverages one-bit algorithm unrolling, effectively integrating information from the physical world in the model architecture. Our method achieves a bit-per-link rate significantly lower than the 1.58 bits reported in prior work, thanks to the natural sparsity that emerges in our network architectures. We numerically demonstrate that the proposed one-bit algorithm unrolling scheme can improve both training and test outcomes by effortlessly increasing the number of layers while substantially compressing the network. Additionally, we provide theoretical results on the generalization gap, convergence rate, stability, and sensitivity of our proposed one-bit algorithm unrolling.
Authors: Yalin Liao, Austin J. Brockmeier
Abstract: Unsupervised anomaly detection is a challenging task. Autoencoders (AEs) or generative models are often employed to model the data distribution of normal inputs and subsequently identify anomalous, out-of-distribution inputs by high reconstruction error or low likelihood, respectively. However, AEs may generalize and achieve small reconstruction errors on abnormal inputs. We propose a decoupled training approach for anomaly detection that both an AE and a likelihood model trained with noise contrastive estimation (NCE). After training the AE, NCE estimates a probability density function, to serve as the anomaly score, on the joint space of the AE's latent representation combined with features of the reconstruction quality. To further reduce the false negative rate in NCE we systematically varying the reconstruction features to augment the training and optimize the contrastive Gaussian noise distribution. Experimental assessments on multiple benchmark datasets demonstrate that the proposed approach matches the performance of prevalent state-of-the-art anomaly detection algorithms.
Authors: Shubham Gupta, Thibaut Durand, Graham Taylor, Lilian W. Bia{\l}okozowicz
Abstract: We present a novel prompt design for Large Language Models (LLMs) tailored to Asynchronous Time Series. Unlike regular time series, which assume values at evenly spaced time points, asynchronous time series consist of timestamped events occurring at irregular intervals, each described in natural language. Our approach effectively utilizes the rich natural language of event descriptions, allowing LLMs to benefit from their broad world knowledge for reasoning across different domains and tasks. This allows us to extend the scope of asynchronous time series analysis beyond forecasting to include tasks like anomaly detection and data imputation. We further introduce Stochastic Soft Prompting, a novel prompt-tuning mechanism that significantly improves model performance, outperforming existing fine-tuning methods such as QLoRA. Through extensive experiments on real world datasets, we demonstrate that our approach achieves state-of-the-art performance across different tasks and datasets.
Authors: Zaiyan Xu, Sushil Vemuri, Kishan Panaganti, Dileep Kalathil, Rahul Jain, Deepak Ramachandran
Abstract: A major challenge in aligning large language models (LLMs) with human preferences is the issue of distribution shift. LLM alignment algorithms rely on static preference datasets, assuming that they accurately represent real-world user preferences. However, user preferences vary significantly across geographical regions, demographics, linguistic patterns, and evolving cultural trends. This preference distribution shift leads to catastrophic alignment failures in many real-world applications. We address this problem using the principled framework of distributionally robust optimization, and develop two novel distributionally robust direct preference optimization (DPO) algorithms, namely, Wasserstein DPO (WDPO) and Kullback-Leibler DPO (KLDPO). We characterize the sample complexity of learning the optimal policy parameters for WDPO and KLDPO. Moreover, we propose scalable gradient descent-style learning algorithms by developing suitable approximations for the challenging minimax loss functions of WDPO and KLDPO. Our empirical experiments demonstrate the superior performance of WDPO and KLDPO in substantially improving the alignment when there is a preference distribution shift.
Authors: Chang Liu, Hai Huang, Yujie Xing, Xingquan Zuo
Abstract: The robustness of Graph Neural Networks (GNNs) has become an increasingly important topic due to their expanding range of applications. Various attack methods have been proposed to explore the vulnerabilities of GNNs, ranging from Graph Modification Attacks (GMA) to the more practical and flexible Graph Injection Attacks (GIA). However, existing methods face two key challenges: (i) their reliance on surrogate models, which often leads to reduced attack effectiveness due to structural differences and prior biases, and (ii) existing GIA methods often sacrifice attack success rates in undefended settings to bypass certain defense models, thereby limiting their overall effectiveness. To overcome these limitations, we propose QUGIA, a Query-based and Unnoticeable Graph Injection Attack. QUGIA injects nodes by first selecting edges based on victim node connections and then generating node features using a Bayesian framework. This ensures that the injected nodes are similar to the original graph nodes, implicitly preserving homophily and making the attack more unnoticeable. Unlike previous methods, QUGIA does not rely on surrogate models, thereby avoiding performance degradation and achieving better generalization. Extensive experiments on six real-world datasets with diverse characteristics demonstrate that QUGIA achieves unnoticeable attacks and outperforms state-of-the-art attackers. The code will be released upon acceptance.
Authors: Xinyi Wu, Yifei Wang, Stefanie Jegelka, Ali Jadbabaie
Abstract: Recent studies have revealed various manifestations of position bias in transformer architectures, from the "lost-in-the-middle" phenomenon to attention sinks, yet a comprehensive theoretical understanding of how attention masks and positional encodings shape these biases remains elusive. This paper introduces a novel graph-theoretic framework to analyze position bias in multi-layer attention. Modeling attention masks as directed graphs, we quantify how tokens interact with contextual information based on their sequential positions. We uncover two key insights: First, causal masking inherently biases attention toward earlier positions, as tokens in deeper layers attend to increasingly more contextualized representations of earlier tokens. Second, we characterize the competing effects of the causal mask and relative positional encodings, such as the decay mask and rotary positional encoding (RoPE): while both mechanisms introduce distance-based decay within individual attention maps, their aggregate effect across multiple attention layers -- coupled with the causal mask -- leads to a trade-off between the long-term decay effects and the cumulative importance of early sequence positions. Through controlled numerical experiments, we not only validate our theoretical findings but also reproduce position biases observed in real-world LLMs. Our framework offers a principled foundation for understanding positional biases in transformers, shedding light on the complex interplay of attention mechanism components and guiding more informed architectural design.
Authors: Mateusz Piotrowski, Paul M. Riechers, Daniel Filan, Adam S. Shai
Abstract: What computational structures emerge in transformers trained on next-token prediction? In this work, we provide evidence that transformers implement constrained Bayesian belief updating -- a parallelized version of partial Bayesian inference shaped by architectural constraints. To do this, we integrate the model-agnostic theory of optimal prediction with mechanistic interpretability to analyze transformers trained on a tractable family of hidden Markov models that generate rich geometric patterns in neural activations. We find that attention heads carry out an algorithm with a natural interpretation in the probability simplex, and create representations with distinctive geometric structure. We show how both the algorithmic behavior and the underlying geometry of these representations can be theoretically predicted in detail -- including the attention pattern, OV-vectors, and embedding vectors -- by modifying the equations for optimal future token predictions to account for the architectural constraints of attention. Our approach provides a principled lens on how gradient descent resolves the tension between optimal prediction and architectural design.
Authors: Shiju Zhao, Junhao Hu, Rongxiao Huang, Jiaqi Zheng, Guihai Chen
Abstract: The context caching technique is employed to accelerate the Multimodal Large Language Model (MLLM) inference by prevailing serving platforms currently. However, this approach merely reuses the Key-Value (KV) cache of the initial sequence of prompt, resulting in full KV cache recomputation even if the prefix differs slightly. This becomes particularly inefficient in the context of interleaved text and images, as well as multimodal retrieval-augmented generation. This paper proposes position-independent caching as a more effective approach for multimodal information management. We have designed and implemented a caching system, named MPIC, to address both system-level and algorithm-level challenges. MPIC stores the KV cache on local or remote disks when receiving multimodal data, and calculates and loads the KV cache in parallel during inference. To mitigate accuracy degradation, we have incorporated integrated reuse and recompute mechanisms within the system. The experimental results demonstrate that MPIC can achieve up to 54% reduction in response time compared to existing context caching systems, while maintaining negligible or no accuracy loss.
Authors: David S. Hayden, Mao Ye, Timur Garipov, Gregory P. Meyer, Carl Vondrick, Zhao Chen, Yuning Chai, Eric Wolff, Siddhartha S. Srinivasa
Abstract: It is difficult to anticipate the myriad challenges that a predictive model will encounter once deployed. Common practice entails a reactive, cyclical approach: model deployment, data mining, and retraining. We instead develop a proactive longtail discovery process by imagining additional data during training. In particular, we develop general model-based longtail signals, including a differentiable, single forward pass formulation of epistemic uncertainty that does not impact model parameters or predictive performance but can flag rare or hard inputs. We leverage these signals as guidance to generate additional training data from a latent diffusion model in a process we call Longtail Guidance (LTG). Crucially, we can perform LTG without retraining the diffusion model or the predictive model, and we do not need to expose the predictive model to intermediate diffusion states. Data generated by LTG exhibit semantically meaningful variation, yield significant generalization improvements on image classification benchmarks, and can be analyzed to proactively discover, explain, and address conceptual gaps in a predictive model.
Authors: Wenbo Sun, Rihan Hai
Abstract: The machine learning (ML) training over disparate data sources traditionally involves materialization, which can impose substantial time and space overhead due to data movement and replication. Factorized learning, which leverages direct computation on disparate sources through linear algebra (LA) rewriting, has emerged as a viable alternative to improve computational efficiency. However, the adaptation of factorized learning to leverage the full capabilities of modern LA-friendly hardware like GPUs has been limited, often requiring manual intervention for algorithm compatibility. This paper introduces Ilargi, a novel factorized learning framework that utilizes matrix-represented data integration (DI) metadata to facilitate automatic factorization across CPU and GPU environments without the need for costly relational joins. Ilargi incorporates an ML-based cost estimator to intelligently selects between factorization and materialization based on data properties, algorithm complexity, hardware environments, and their interactions. This strategy ensures up to 8.9x speedups on GPUs and achieves over 20% acceleration in batch ML training workloads, thereby enhancing the practicability of ML training across diverse data integration scenarios and hardware platforms. To our knowledge, this work is the very first effort in GPU-compatible factorized learning.
Authors: Tao Zhang, Jia-Shu Pan, Ruiqi Feng, Tailin Wu
Abstract: We introduce Test-time Scalable MCTS-enhanced Diffusion Model (T-SCEND), a novel framework that significantly improves diffusion model's reasoning capabilities with better energy-based training and scaling up test-time computation. We first show that na\"ively scaling up inference budget for diffusion models yields marginal gain. To address this, the training of T-SCEND consists of a novel linear-regression negative contrastive learning objective to improve the performance-energy consistency of the energy landscape, and a KL regularization to reduce adversarial sampling. During inference, T-SCEND integrates the denoising process with a novel hybrid Monte Carlo Tree Search (hMCTS), which sequentially performs best-of-N random search and MCTS as denoising proceeds. On challenging reasoning tasks of Maze and Sudoku, we demonstrate the effectiveness of T-SCEND's training objective and scalable inference method. In particular, trained with Maze sizes of up to $6\times6$, our T-SCEND solves $88\%$ of Maze problems with much larger sizes of $15\times15$, while standard diffusion completely fails.Code to reproduce the experiments can be found at https://github.com/AI4Science-WestlakeU/t_scend.
Authors: Bin Xie, Gady Agam
Abstract: Diffusion models suffer from the huge consumption of time and resources to train. For example, diffusion models need hundreds of GPUs to train for several weeks for a high-resolution generative task to meet the requirements of an extremely large number of iterations and a large batch size. Training diffusion models become a millionaire's game. With limited resources that only fit a small batch size, training a diffusion model always fails. In this paper, we investigate the key reasons behind the difficulties of training diffusion models with limited resources. Through numerous experiments and demonstrations, we identified a major factor: the significant variation in the training losses across different timesteps, which can easily disrupt the progress made in previous iterations. Moreover, different prediction types of $x_0$ exhibit varying effectiveness depending on the task and timestep. We hypothesize that using a mixed-prediction approach to identify the most accurate $x_0$ prediction type could potentially serve as a breakthrough in addressing this issue. In this paper, we outline several challenges and insights, with the hope of inspiring further research aimed at tackling the limitations of training diffusion models with constrained resources, particularly for high-resolution tasks.
Authors: Oscar Skean, Md Rifat Arefin, Dan Zhao, Niket Patel, Jalal Naghiyev, Yann LeCun, Ravid Shwartz-Ziv
Abstract: From extracting features to generating text, the outputs of large language models (LLMs) typically rely on their final layers, following the conventional wisdom that earlier layers capture only low-level cues. However, our analysis shows that intermediate layers can encode even richer representations, often improving performance on a wide range of downstream tasks. To explain and quantify these hidden-layer properties, we propose a unified framework of representation quality metrics based on information theory, geometry, and invariance to input perturbations. Our framework highlights how each model layer balances information compression and signal preservation, revealing why mid-depth embeddings can exceed the last layer's performance. Through extensive experiments on 32 text-embedding tasks and comparisons across model architectures (transformers, state-space models) and domains (language, vision), we demonstrate that intermediate layers consistently provide stronger features. These findings challenge the standard focus on final-layer embeddings and open new directions for model analysis and optimization, including strategic use of mid-layer representations for more robust and accurate AI systems.
Authors: Haohan Zou, Jie Feng, Hao Zhao, Yuanyuan Shi
Abstract: Despite advances in learning-based methods, finding valid Lyapunov functions for nonlinear dynamical systems remains challenging. Current neural network approaches face two main issues: challenges in scalable verification and limited interpretability. To address these, we propose an end-to-end framework using transformers to construct analytical Lyapunov functions (local), which simplifies formal verification, enhances interpretability, and provides valuable insights for control engineers. Our framework consists of a transformer-based trainer that generates candidate Lyapunov functions and a falsifier that verifies candidate expressions and refines the model via risk-seeking policy gradient. Unlike Alfarano et al. (2024), which utilizes pre-training and seeks global Lyapunov functions for low-dimensional systems, our model is trained from scratch via reinforcement learning (RL) and succeeds in finding local Lyapunov functions for high-dimensional and non-polynomial systems. Given the analytical nature of the candidates, we employ efficient optimization methods for falsification during training and formal verification tools for the final verification. We demonstrate the efficiency of our approach on a range of nonlinear dynamical systems with up to ten dimensions and show that it can discover Lyapunov functions not previously identified in the control literature.
Authors: Hanlin Wu, Yuxuan Song, Jingjing Gong, Ziyao Cao, Yawen Ouyang, Jianbing Zhang, Hao Zhou, Wei-Ying Ma, Jingjing Liu
Abstract: Generative modeling of crystal data distribution is an important yet challenging task due to the unique periodic physical symmetry of crystals. Diffusion-based methods have shown early promise in modeling crystal distribution. More recently, Bayesian Flow Networks were introduced to aggregate noisy latent variables, resulting in a variance-reduced parameter space that has been shown to be advantageous for modeling Euclidean data distributions with structural constraints (Song et al., 2023). Inspired by this, we seek to unlock its potential for modeling variables located in non-Euclidean manifolds e.g. those within crystal structures, by overcoming challenging theoretical issues. We introduce CrysBFN, a novel crystal generation method by proposing a periodic Bayesian flow, which essentially differs from the original Gaussian-based BFN by exhibiting non-monotonic entropy dynamics. To successfully realize the concept of periodic Bayesian flow, CrysBFN integrates a new entropy conditioning mechanism and empirically demonstrates its significance compared to time-conditioning. Extensive experiments over both crystal ab initio generation and crystal structure prediction tasks demonstrate the superiority of CrysBFN, which consistently achieves new state-of-the-art on all benchmarks. Surprisingly, we found that CrysBFN enjoys a significant improvement in sampling efficiency, e.g., ~100x speedup 10 v.s. 2000 steps network forwards) compared with previous diffusion-based methods on MP-20 dataset. Code is available at https://github.com/wu-han-lin/CrysBFN.
Authors: Yijia Zhao, Qing Zhou
Abstract: The causal bandit problem aims to sequentially learn the intervention that maximizes the expectation of a reward variable within a system governed by a causal graph. Most existing approaches assume prior knowledge of the graph structure, or impose unrealistically restrictive conditions on the graph. In this paper, we assume a Gaussian linear directed acyclic graph (DAG) over arms and the reward variable, and study the causal bandit problem when the graph structure is unknown. We identify backdoor adjustment sets for each arm using sequentially generated experimental and observational data during the decision process, which allows us to estimate causal effects and construct upper confidence bounds. By integrating estimates from both data sources, we develop a novel bandit algorithm, based on modified upper confidence bounds, to sequentially determine the optimal intervention. We establish both case-dependent and case-independent upper bounds on the cumulative regret for our algorithm, which improve upon the bounds of the standard multi-armed bandit algorithms. Our empirical study demonstrates its advantage over existing methods with respect to cumulative regret and computation time.
Authors: Yuji Tone, Masatoshi Hanai, Mitsuaki Kawamura, Kenjiro Taura, Toyotaro Suzumura
Abstract: The discovery of new materials using crystal structure prediction (CSP) based on generative machine learning models has become a significant research topic in recent years. In this paper, we study invariance and continuity in the generative machine learning for CSP. We propose a new model, called ContinuouSP, which effectively handles symmetry and periodicity in crystals. We clearly formulate the invariance and the continuity, and construct a model based on the energy-based model. Our preliminary evaluation demonstrates the effectiveness of this model with the CSP task.
Authors: Georgios Margaritis, Periklis Petridis, Dimitris J. Bertsimas
Abstract: Recent advancements in machine learning (ML), natural language processing (NLP), and foundational models have shown promise for real-life applications in critical, albeit compute-constrainted fields like healthcare. In such areas, combining foundational models with supervised ML offers potential for automating tasks like diagnosis and treatment planning, but the limited availability of onsite computational resources pose significant challenges before applying these technologies effectively: Current approaches either yield subpar results when using pretrained models without task-specific adaptation, or require substantial computational resources for fine-tuning, which is often a barrier to entry in such environments. This renders them inaccessible in applications where performance and quality standards are high, but computational resources are scarce. To bridge the gap between best-in-class performance and accessibility, we propose a novel method for adapting foundational, multimodal embeddings to downstream tasks, without the need of expensive fine-tuning processes. Our method leverages frozen embeddings from Large Language Models (LLMs) and Vision Models, and uses contrastive learning to train a small, task-specific nonlinear projection that can be used in the downstream task, without having to fine-tune the original foundational models. We show that this efficient procedure leads to significant performance improvements across various downstream tasks, and perhaps more importantly with minimal computational overhead, offering a practical solution for the use of advanced, foundational ML models in resource-constrained settings.
Authors: Krishan Agyakari Raja Babu, Supriti Mulay, Om Prabhu, Mohanasankar Sivaprakasam
Abstract: Deep generative models and synthetic medical data have shown significant promise in addressing key challenges in healthcare, such as privacy concerns, data bias, and the scarcity of realistic datasets. While research in this area has grown rapidly and demonstrated substantial theoretical potential, its practical adoption in clinical settings remains limited. Despite the benefits synthetic data offers, questions surrounding its reliability and credibility persist, leading to a lack of trust among clinicians. This position paper argues that fostering trust in synthetic medical data is crucial for its clinical adoption. It aims to spark a discussion on the viability of synthetic medical data in clinical practice, particularly in the context of current advancements in AI. We present empirical evidence from brain tumor segmentation to demonstrate that the quality, diversity, and proportion of synthetic data directly impact trust in clinical AI models. Our findings provide insights to improve the deployment and acceptance of synthetic data-driven AI systems in real-world clinical workflows.
Authors: Zhiyong Wang, Jiahang Sun, Mingze Kong, Jize Xie, Qinghua Hu, John C. S. Lui, Zhongxiang Dai
Abstract: The contextual multi-armed bandit (MAB) is a widely used framework for problems requiring sequential decision-making under uncertainty, such as recommendation systems. In applications involving a large number of users, the performance of contextual MAB can be significantly improved by facilitating collaboration among multiple users. This has been achieved by the clustering of bandits (CB) methods, which adaptively group the users into different clusters and achieve collaboration by allowing the users in the same cluster to share data. However, classical CB algorithms typically rely on numerical reward feedback, which may not be practical in certain real-world applications. For instance, in recommendation systems, it is more realistic and reliable to solicit preference feedback between pairs of recommended items rather than absolute rewards. To address this limitation, we introduce the first "clustering of dueling bandit algorithms" to enable collaborative decision-making based on preference feedback. We propose two novel algorithms: (1) Clustering of Linear Dueling Bandits (COLDB) which models the user reward functions as linear functions of the context vectors, and (2) Clustering of Neural Dueling Bandits (CONDB) which uses a neural network to model complex, non-linear user reward functions. Both algorithms are supported by rigorous theoretical analyses, demonstrating that user collaboration leads to improved regret bounds. Extensive empirical evaluations on synthetic and real-world datasets further validate the effectiveness of our methods, establishing their potential in real-world applications involving multiple users with preference-based feedback.
Authors: Alan Oursland
Abstract: Neural networks may naturally favor distance-based representations, where smaller activations indicate closer proximity to learned prototypes. This contrasts with intensity-based approaches, which rely on activation magnitudes. To test this hypothesis, we conducted experiments with six MNIST architectural variants constrained to learn either distance or intensity representations. Our results reveal that the underlying representation affects model performance. We develop a novel geometric framework that explains these findings and introduce OffsetL2, a new architecture based on Mahalanobis distance equations, to further validate this framework. This work highlights the importance of considering distance-based learning in neural network design.
Authors: Yawen Chen, Jiande Sun, Jing Li, Huaxiang Zhang
Abstract: Cognitive diagnosis can infer the students' mastery of specific knowledge concepts based on historical response logs. However, the existing cognitive diagnostic models (CDMs) represent students' proficiency via a unidimensional perspective, which can't assess the students' mastery on each knowledge concept comprehensively. Moreover, the Q-matrix binarizes the relationship between exercises and knowledge concepts, and it can't represent the latent relationship between exercises and knowledge concepts. Especially, when the granularity of knowledge attributes refines increasingly, the Q-matrix becomes incomplete correspondingly and the sparse binary representation (0/1) fails to capture the intricate relationships among knowledge concepts. To address these issues, we propose a Concept-aware Latent and Explicit Knowledge Integration model for cognitive diagnosis (CLEKI-CD). Specifically, a multidimensional vector is constructed according to the students' mastery and exercise difficulty for each knowledge concept from multiple perspectives, which enhances the representation capabilities of the model. Moreover, a latent Q-matrix is generated by our proposed attention-based knowledge aggregation method, and it can uncover the coverage degree of exercises over latent knowledge. The latent Q-matrix can supplement the sparse explicit Q-matrix with the inherent relationships among knowledge concepts, and mitigate the knowledge coverage problem. Furthermore, we employ a combined cognitive diagnosis layer to integrate both latent and explicit knowledge, further enhancing cognitive diagnosis performance. Extensive experiments on real-world datasets demonstrate that CLEKI-CD outperforms the state-of-the-art models. The proposed CLEKI-CD is promising in practical applications in the field of intelligent education, as it exhibits good interpretability with diagnostic results.
Authors: Yuxiao Cheng, Xinxin Song, Ziqian Wang, Qin Zhong, Kunlun He, Jinli Suo
Abstract: Recent advances in deep learning (DL) have prompted the development of high-performing early warning score (EWS) systems, predicting clinical deteriorations such as acute kidney injury, acute myocardial infarction, or circulatory failure. DL models have proven to be powerful tools for various tasks but come with the cost of lacking interpretability and limited generalizability, hindering their clinical applications. To develop a practical EWS system applicable to various outcomes, we propose causally-informed explainable early prediction model, which leverages causal discovery to identify the underlying causal relationships of prediction and thus owns two unique advantages: demonstrating the explicit interpretation of the prediction while exhibiting decent performance when applied to unfamiliar environments. Benefiting from these features, our approach achieves superior accuracy for 6 different critical deteriorations and achieves better generalizability across different patient groups, compared to various baseline algorithms. Besides, we provide explicit causal pathways to serve as references for assistant clinical diagnosis and potential interventions. The proposed approach enhances the practical application of deep learning in various medical scenarios.
Authors: Hoang M. Nguyen, Satya N. Shukla, Qiang Zhang, Hanchao Yu, Sreya D. Roy, Taipeng Tian, Lingjiong Zhu, Yuchen Liu
Abstract: Self-supervised learning has been a powerful approach for learning meaningful representations from unlabeled data across various domains, reducing the reliance on large labeled datasets. Inspired by BERT's success in capturing deep bidirectional contexts in natural language processing, similar frameworks have been adapted to other modalities such as audio, with models like BEATs extending the bidirectional training paradigm to audio signals using vector quantization (VQ). However, these frameworks face challenges, notably their dependence on a single codebook for quantization, which may not capture the complex, multifaceted nature of signals. In addition, inefficiencies in codebook utilization lead to underutilized code vectors. To address these limitations, we introduce BRIDLE (Bidirectional Residual Quantization Interleaved Discrete Learning Encoder), a self-supervised encoder pretraining framework that incorporates residual quantization (RQ) into the bidirectional training process, and is generalized for pretraining with audio, image, and video. Using multiple hierarchical codebooks, RQ enables fine-grained discretization in the latent space, enhancing representation quality. BRIDLE involves an interleaved training procedure between the encoder and tokenizer. We evaluate BRIDLE on audio understanding tasks using classification benchmarks, achieving state-of-the-art results, and demonstrate competitive performance on image classification and video classification tasks, showing consistent improvements over traditional VQ methods in downstream performance.
Authors: Ruth Wan Theng Chew, Quoc Phong Nguyen, Bryan Kian Hsiang Low
Abstract: Bilevel optimization is characterized by a two-level optimization structure, where the upper-level problem is constrained by optimal lower-level solutions, and such structures are prevalent in real-world problems. The constraint by optimal lower-level solutions poses significant challenges, especially in noisy, constrained, and derivative-free settings, as repeating lower-level optimizations is sample inefficient and predicted lower-level solutions may be suboptimal. We present BILevel Bayesian Optimization (BILBO), a novel Bayesian optimization algorithm for general bilevel problems with blackbox functions, which optimizes both upper- and lower-level problems simultaneously, without the repeated lower-level optimization required by existing methods. BILBO samples from confidence-bounds based trusted sets, which bounds the suboptimality on the lower level. Moreover, BILBO selects only one function query per iteration, where the function query selection strategy incorporates the uncertainty of estimated lower-level solutions and includes a conditional reassignment of the query to encourage exploration of the lower-level objective. The performance of BILBO is theoretically guaranteed with a sublinear regret bound for commonly used kernels and is empirically evaluated on several synthetic and real-world problems.
Authors: Koen Minartz, Tim d'Hondt, Leon Hillmann, J\"orn Starru{\ss}, Lutz Brusch, Vlado Menkovski
Abstract: The cellular Potts model (CPM) is a powerful computational method for simulating collective spatiotemporal dynamics of biological cells. To drive the dynamics, CPMs rely on physics-inspired Hamiltonians. However, as first principles remain elusive in biology, these Hamiltonians only approximate the full complexity of real multicellular systems. To address this limitation, we propose NeuralCPM, a more expressive cellular Potts model that can be trained directly on observational data. At the core of NeuralCPM lies the Neural Hamiltonian, a neural network architecture that respects universal symmetries in collective cellular dynamics. Moreover, this approach enables seamless integration of domain knowledge by combining known biological mechanisms and the expressive Neural Hamiltonian into a hybrid model. Our evaluation with synthetic and real-world multicellular systems demonstrates that NeuralCPM is able to model cellular dynamics that cannot be accounted for by traditional analytical Hamiltonians.
Authors: Matias D. Cattaneo, Boris Shigida
Abstract: In modern optimization methods used in deep learning, each update depends on the history of previous iterations, often referred to as memory, and this dependence decays fast as the iterates go further into the past. For example, gradient descent with momentum has exponentially decaying memory through exponentially averaged past gradients. We introduce a general technique for identifying a memoryless algorithm that approximates an optimization algorithm with memory. It is obtained by replacing all past iterates in the update by the current one, and then adding a correction term arising from memory (also a function of the current iterate). This correction term can be interpreted as a perturbation of the loss, and the nature of this perturbation can inform how memory implicitly (anti-)regularizes the optimization dynamics. As an application of our theory, we find that Lion does not have the kind of implicit anti-regularization induced by memory that AdamW does, providing a theory-based explanation for Lion's better generalization performance recently documented.
Authors: David Winant, Sonny Achten, Johan A. K. Suykens
Abstract: Modern clustering approaches often trade interpretability for performance, particularly in deep learning-based methods. We present Generative Kernel Spectral Clustering (GenKSC), a novel model combining kernel spectral clustering with generative modeling to produce both well-defined clusters and interpretable representations. By augmenting weighted variance maximization with reconstruction and clustering losses, our model creates an explorable latent space where cluster characteristics can be visualized through traversals along cluster directions. Results on MNIST and FashionMNIST datasets demonstrate the model's ability to learn meaningful cluster representations.
Authors: Frederik Lizak Johansen, Ulrik Friis-Jensen, Erik Bj{\o}rnager Dam, Kirsten Marie {\O}rnsbjerg Jensen, Roc\'io Mercado, Raghavendra Selvan
Abstract: Novel materials drive progress across applications from energy storage to electronics. Automated characterization of material structures with machine learning methods offers a promising strategy for accelerating this key step in material design. In this work, we introduce an autoregressive language model that performs crystal structure prediction (CSP) from powder diffraction data. The presented model, deCIFer, generates crystal structures in the widely used Crystallographic Information File (CIF) format and can be conditioned on powder X-ray diffraction (PXRD) data. Unlike earlier works that primarily rely on high-level descriptors like composition, deCIFer performs CSP from diffraction data. We train deCIFer on nearly 2.3M unique crystal structures and validate on diverse sets of PXRD patterns for characterizing challenging inorganic crystal systems. Qualitative and quantitative assessments using the residual weighted profile and Wasserstein distance show that deCIFer produces structures that more accurately match the target diffraction data when conditioned, compared to the unconditioned case. Notably, deCIFer can achieve a 94% match rate on unseen data. deCIFer bridges experimental diffraction data with computational CSP, lending itself as a powerful tool for crystal structure characterization and accelerating materials discovery.
Authors: Linus Aronsson, Morteza Haghir Chehreghani
Abstract: Signed networks, where edges are labeled as positive or negative to indicate friendly or antagonistic interactions, offer a natural framework for studying polarization, trust, and conflict in social systems. Detecting meaningful group structures in these networks is crucial for understanding online discourse, political division, and trust dynamics. A key challenge is to identify groups that are cohesive internally yet antagonistic externally, while allowing for neutral or unaligned vertices. In this paper, we address this problem by identifying $k$ polarized communities that are large, dense, and balanced in size. We develop an approach based on Frank-Wolfe optimization, leading to a local search procedure with provable convergence guarantees. Our method is both scalable and efficient, outperforming state-of-the-art baselines in solution quality while remaining competitive in terms of computational efficiency.
Authors: Naghmeh Ghanooni, Barbod Pajoum, Harshit Rawal, Sophie Fellenz, Vo Nguyen Le Duy, Marius Kloft
Abstract: Contrastive learning is a well-established paradigm in representation learning. The standard framework of contrastive learning minimizes the distance between "similar" instances and maximizes the distance between dissimilar ones in the projection space, disregarding the various aspects of similarity that can exist between two samples. Current methods rely on a single projection head, which fails to capture the full complexity of different aspects of a sample, leading to suboptimal performance, especially in scenarios with limited training data. In this paper, we present a novel supervised contrastive learning method in a unified framework called multilevel contrastive learning (MLCL), that can be applied to both multi-label and hierarchical classification tasks. The key strength of the proposed method is the ability to capture similarities between samples across different labels and/or hierarchies using multiple projection heads. Extensive experiments on text and image datasets demonstrate that the proposed approach outperforms state-of-the-art contrastive learning methods
Authors: Peiyan Hu, Xiaowei Qian, Wenhao Deng, Rui Wang, Haodong Feng, Ruiqi Feng, Tao Zhang, Long Wei, Yue Wang, Zhi-Ming Ma, Tailin Wu
Abstract: The application of deep learning for partial differential equation (PDE)-constrained control is gaining increasing attention. However, existing methods rarely consider safety requirements crucial in real-world applications. To address this limitation, we propose Safe Diffusion Models for PDE Control (SafeDiffCon), which introduce the uncertainty quantile as model uncertainty quantification to achieve optimal control under safety constraints through both post-training and inference phases. Firstly, our approach post-trains a pre-trained diffusion model to generate control sequences that better satisfy safety constraints while achieving improved control objectives via a reweighted diffusion loss, which incorporates the uncertainty quantile estimated using conformal prediction. Secondly, during inference, the diffusion model dynamically adjusts both its generation process and parameters through iterative guidance and fine-tuning, conditioned on control targets while simultaneously integrating the estimated uncertainty quantile. We evaluate SafeDiffCon on three control tasks: 1D Burgers' equation, 2D incompressible fluid, and controlled nuclear fusion problem. Results demonstrate that SafeDiffCon is the only method that satisfies all safety constraints, whereas other classical and deep learning baselines fail. Furthermore, while adhering to safety constraints, SafeDiffCon achieves the best control performance.
Authors: Edo Cohen-Karlik, Itamar Zimerman, Liane Galanti, Ido Atad, Amir Globerson, Lior Wolf
Abstract: Recent advances in efficient sequence modeling have introduced selective state-space layers, a key component of the Mamba architecture, which have demonstrated remarkable success in a wide range of NLP and vision tasks. While Mamba's empirical performance has matched or surpassed SoTA transformers on such diverse benchmarks, the theoretical foundations underlying its powerful representational capabilities remain less explored. In this work, we investigate the expressivity of selective state-space layers using multivariate polynomials, and prove that they surpass linear transformers in expressiveness. Consequently, our findings reveal that Mamba offers superior representational power over linear attention-based models for long sequences, while not sacrificing their generalization. Our theoretical insights are validated by a comprehensive set of empirical experiments on various datasets.
Authors: Dexiong Chen, Markus Krimmel, Karsten Borgwardt
Abstract: We introduce AutoGraph, a novel autoregressive framework for generating large attributed graphs using decoder-only transformers. At the core of our approach is a reversible "flattening" process that transforms graphs into random sequences. By sampling and learning from these sequences, AutoGraph enables transformers to model and generate complex graph structures in a manner akin to natural language. In contrast to diffusion models that rely on computationally intensive node features, our approach operates exclusively on these sequences. The sampling complexity and sequence length scale linearly with the number of edges, making AutoGraph highly scalable for generating large sparse graphs. Empirically, AutoGraph achieves state-of-the-art performance across diverse synthetic and molecular graph generation benchmarks, while delivering a 100-fold generation and a 3-fold training speedup compared to leading diffusion models. Additionally, it demonstrates promising transfer capabilities and supports substructure-conditioned generation without additional fine-tuning. By extending language modeling techniques to graph generation, this work paves the way for developing graph foundation models.
Authors: Ji\v{r}\'i N\v{e}me\v{c}ek, Mark Kozdoba, Illia Kryvoviaz, Tom\'a\v{s} Pevn\'y, Jakub Mare\v{c}ek
Abstract: Bias evaluation is fundamental to trustworthy AI, both in terms of checking data quality and in terms of checking the outputs of AI systems. In testing data quality, for example, one may study a distance of a given dataset, viewed as a distribution, to a given ground-truth reference dataset. However, classical metrics, such as the Total Variation and the Wasserstein distances, are known to have high sample complexities and, therefore, may fail to provide meaningful distinction in many practical scenarios. In this paper, we propose a new notion of distance, the Maximum Subgroup Discrepancy (MSD). In this metric, two distributions are close if, roughly, discrepancies are low for all feature subgroups. While the number of subgroups may be exponential, we show that the sample complexity is linear in the number of features, thus making it feasible for practical applications. Moreover, we provide a practical algorithm for the evaluation of the distance, based on Mixed-integer optimization (MIO). We also note that the proposed distance is easily interpretable, thus providing clearer paths to fixing the biases once they have been identified. It also provides guarantees for all subgroups. Finally, we empirically evaluate, compare with other metrics, and demonstrate the above properties of MSD on real-world datasets.
Authors: Javier Rando, Jie Zhang, Nicholas Carlini, Florian Tram\`er
Abstract: In the past decade, considerable research effort has been devoted to securing machine learning (ML) models that operate in adversarial settings. Yet, progress has been slow even for simple "toy" problems (e.g., robustness to small adversarial perturbations) and is often hindered by non-rigorous evaluations. Today, adversarial ML research has shifted towards studying larger, general-purpose language models. In this position paper, we argue that the situation is now even worse: in the era of LLMs, the field of adversarial ML studies problems that are (1) less clearly defined, (2) harder to solve, and (3) even more challenging to evaluate. As a result, we caution that yet another decade of work on adversarial ML may fail to produce meaningful progress.
Authors: Donghe Chen, Yubin Peng, Tengjie Zheng, Han Wang, Chaoran Qu, Lin Cheng
Abstract: High-precision control tasks present substantial challenges for reinforcement learning (RL) algorithms, frequently resulting in suboptimal performance attributed to network approximation inaccuracies and inadequate sample quality.These issues are exacerbated when the task requires the agent to achieve a precise goal state, as is common in robotics and other real-world applications.We introduce Adviser-Actor-Critic (AAC), designed to address the precision control dilemma by combining the precision of feedback control theory with the adaptive learning capability of RL and featuring an Adviser that mentors the actor to refine control actions, thereby enhancing the precision of goal attainment.Finally, through benchmark tests, AAC outperformed standard RL algorithms in precision-critical, goal-conditioned tasks, demonstrating AAC's high precision, reliability, and robustness.Code are available at: https://anonymous.4open.science/r/Adviser-Actor-Critic-8AC5.
URLs: https://anonymous.4open.science/r/Adviser-Actor-Critic-8AC5.
Authors: Albert Alcalde, Giovanni Fantuzzi, Enrique Zuazua
Abstract: We prove that hardmax attention transformers perfectly classify datasets of $N$ labeled sequences in $\mathbb{R}^d$, $d\geq 2$. Specifically, given $N$ sequences with an arbitrary but finite length in $\mathbb{R}^d$, we construct a transformer with $\mathcal{O}(N)$ blocks and $\mathcal{O}(Nd)$ parameters perfectly classifying this dataset. Our construction achieves the best complexity estimate to date, independent of the length of the sequences, by innovatively alternating feed-forward and self-attention layers and by capitalizing on the clustering effect inherent to the latter. Our novel constructive method also uses low-rank parameter matrices within the attention mechanism, a common practice in real-life transformer implementations. Consequently, our analysis holds twofold significance: it substantially advances the mathematical theory of transformers and it rigorously justifies their exceptional real-world performance in sequence classification tasks.
Authors: Keanu Sisouk, Julie Delon, Julien Tierny
Abstract: This paper serves as a user guide to sampling strategies for sliced optimal transport. We provide reminders and additional regularity results on the Sliced Wasserstein distance. We detail the construction methods, generation time complexity, theoretical guarantees, and conditions for each strategy. Additionally, we provide insights into their suitability for sliced optimal transport in theory. Extensive experiments on both simulated and real-world data offer a representative comparison of the strategies, culminating in practical recommendations for their best usage.
Authors: Donghe Chen, Jiaxuan Yue, Tengjie Zheng, Lanxuan Wang, Lin Cheng
Abstract: In real-world regression tasks, datasets frequently exhibit imbalanced distributions, characterized by a scarcity of data in high-complexity regions and an abundance in low-complexity areas. This imbalance presents significant challenges for existing classification methods with clear class boundaries, while highlighting a scarcity of approaches specifically designed for imbalanced regression problems. To better address these issues, we introduce a novel concept of Imbalanced Regression, which takes into account both the complexity of the problem and the density of data points, extending beyond traditional definitions that focus only on data density. Furthermore, we propose Error Distribution Smoothing (EDS) as a solution to tackle imbalanced regression, effectively selecting a representative subset from the dataset to reduce redundancy while maintaining balance and representativeness. Through several experiments, EDS has shown its effectiveness, and the related code and dataset can be accessed at https://anonymous.4open.science/r/Error-Distribution-Smoothing-762F.
URLs: https://anonymous.4open.science/r/Error-Distribution-Smoothing-762F.
Authors: Chengrui Li, Yunmiao Wang, Yule Wang, Weihan Li, Dieter Jaeger, Anqi Wu
Abstract: A fully disentangled variational auto-encoder (VAE) aims to identify disentangled latent components from observations. However, enforcing full independence between all latent components may be too strict for certain datasets. In some cases, multiple factors may be entangled together in a non-separable manner, or a single independent semantic meaning could be represented by multiple latent components within a higher-dimensional manifold. To address such scenarios with greater flexibility, we develop the Partially Disentangled VAE (PDisVAE), which generalizes the total correlation (TC) term in fully disentangled VAEs to a partial correlation (PC) term. This framework can handle group-wise independence and can naturally reduce to either the standard VAE or the fully disentangled VAE. Validation through three synthetic experiments demonstrates the correctness and practicality of PDisVAE. When applied to real-world datasets, PDisVAE discovers valuable information that is difficult to find using fully disentangled VAEs, implying its versatility and effectiveness.
Authors: Daniele Lunghi, Yannick Molinghen, Alkis Simitsis, Tom Lenaerts, Gianluca Bontempi
Abstract: Adversarial attacks pose a significant threat to data-driven systems, and researchers have spent considerable resources studying them. Despite its economic relevance, this trend largely overlooked the issue of credit card fraud detection. To address this gap, we propose a new threat model that demonstrates the limitations of existing attacks and highlights the necessity to investigate new approaches. We then design a new adversarial attack for credit card fraud detection, employing reinforcement learning to bypass classifiers. This attack, called FRAUD-RLA, is designed to maximize the attacker's reward by optimizing the exploration-exploitation tradeoff and working with significantly less required knowledge than competitors. Our experiments, conducted on three different heterogeneous datasets and against two fraud detection systems, indicate that FRAUD-RLA is effective, even considering the severe limitations imposed by our threat model.
Authors: Hanlin Yu, Arto Klami, Aapo Hyv\"arinen, Anna Korba, Omar Chehab
Abstract: Density ratio estimation in high dimensions can be reframed as integrating a certain quantity, the time score, over probability paths which interpolate between the two densities. In practice, the time score has to be estimated based on samples from the two densities. However, existing methods for this problem remain computationally expensive and can yield inaccurate estimates. Inspired by recent advances in generative modeling, we introduce a novel framework for time score estimation, based on a conditioning variable. Choosing the conditioning variable judiciously enables a closed-form objective function. We demonstrate that, compared to previous approaches, our approach results in faster learning of the time score and competitive or better estimation accuracies of the density ratio on challenging tasks. Furthermore, we establish theoretical guarantees on the error of the estimated density ratio.
Authors: Shengda Zhuo, Jiwang Fang, Hongguang Lin, Yin Tang, Min Chen, Changdong Wang, Shuqiang Huang
Abstract: Graph Neural Networks (GNNs) have significant advantages in handling non-Euclidean data and have been widely applied across various areas, thus receiving increasing attention in recent years. The framework of GNN models mainly includes the information propagation phase and the aggregation phase, treating nodes and edges as information entities and propagation channels, respectively. However, most existing GNN models face the challenge of disconnection between node and edge feature information, as these models typically treat the learning of edge and node features as independent tasks. To address this limitation, we aim to develop an edge-empowered graph feature preference learning framework that can capture edge embeddings to assist node embeddings. By leveraging the learned multidimensional edge feature matrix, we construct multi-channel filters to more effectively capture accurate node features, thereby obtaining the non-local structural characteristics and fine-grained high-order node features. Specifically, the inclusion of multidimensional edge information enhances the functionality and flexibility of the GNN model, enabling it to handle complex and diverse graph data more effectively. Additionally, integrating relational representation learning into the message passing framework allows graph nodes to receive more useful information, thereby facilitating node representation learning. Finally, experiments on four real-world heterogeneous graphs demonstrate the effectiveness of theproposed model.
Authors: Wangtao Sun, Haotian Xu, Huanxuan Liao, Xuanqing Yu, Zhongtao Jiang, Shizhu He, Jun Zhao, Kang Liu
Abstract: How to interact with LLMs through \emph{instructions} has been widely studied by researchers. However, previous studies have treated the emergence of instructions and the training of LLMs on task data as separate processes, overlooking the inherent unity between the two. This paper proposes a neural network framework, VaiBot, that integrates VAE and VIB, designed to uniformly model, learn, and infer both deduction and induction tasks under LLMs. Through experiments, we demonstrate that VaiBot performs on par with existing baseline methods in terms of deductive capabilities while significantly surpassing them in inductive capabilities. We also find that VaiBot can scale up using general instruction-following data and exhibits excellent one-shot induction abilities. We finally synergistically integrate the deductive and inductive processes of VaiBot. Through T-SNE dimensionality reduction, we observe that its inductive-deductive process significantly improves the distribution of training parameters, enabling it to outperform baseline methods in inductive reasoning tasks. The code and data for this paper can be found at https://anonymous.4open.science/r/VaiBot-021F.
Authors: Onur Celik, Zechu Li, Denis Blessing, Ge Li, Daniel Palanicek, Jan Peters, Georgia Chalvatzaki, Gerhard Neumann
Abstract: Maximum entropy reinforcement learning (MaxEnt-RL) has become the standard approach to RL due to its beneficial exploration properties. Traditionally, policies are parameterized using Gaussian distributions, which significantly limits their representational capacity. Diffusion-based policies offer a more expressive alternative, yet integrating them into MaxEnt-RL poses challenges--primarily due to the intractability of computing their marginal entropy. To overcome this, we propose Diffusion-Based Maximum Entropy RL (DIME). DIME leverages recent advances in approximate inference with diffusion models to derive a lower bound on the maximum entropy objective. Additionally, we propose a policy iteration scheme that provably converges to the optimal diffusion policy. Our method enables the use of expressive diffusion-based policies while retaining the principled exploration benefits of MaxEnt-RL, significantly outperforming other diffusion-based methods on challenging high-dimensional control benchmarks. It is also competitive with state-of-the-art non-diffusion based RL methods while requiring fewer algorithmic design choices and smaller update-to-data ratios, reducing computational complexity.
Authors: Josua Faller, J\"org Martin
Abstract: Subspace inference for neural networks assumes that a subspace of their parameter space suffices to produce a reliable uncertainty quantification. In this work, we mathematically derive the optimal subspace model to a Bayesian inference scenario based on the Laplace approximation. We demonstrate empirically that, in the optimal case, often a fraction of parameters less than 1% is sufficient to obtain a reliable estimate of the full Laplace approximation. Since the optimal solution is derived, we can evaluate all other subspace models against a baseline. In addition, we give an approximation of our method that is applicable to larger problem settings, in which the optimal solution is not computable, and compare it to existing subspace models from the literature. In general, our approximation scheme outperforms previous work. Furthermore, we present a metric to qualitatively compare different subspace models even if the exact Laplace approximation is unknown.
Authors: Alice Vian, Diego Andre Eifer, Mauricio Anes, Guilherme Ribeiro Garcia, Mariana Recamonde-Mendoza
Abstract: Artificial intelligence (AI) is increasingly being utilized to optimize magnetic resonance imaging (MRI) protocols. Given that image details are critical for diagnostic accuracy, optimizing MRI acquisition protocols is essential for enhancing image quality. While medical physicists are responsible for this optimization, the variability in equipment usage and the wide range of MRI protocols in clinical settings pose significant challenges. This study aims to validate the application of AI in optimizing MRI protocols using dynamic data from clinical practice, specifically DICOM metadata. To achieve this, four MRI spine exam databases were created, with the target attribute being the binary classification of image quality (good or bad). Five AI models were trained to identify trends in acquisition parameters that influence image quality, grounded in MRI theory. These trends were analyzed using SHAP graphs. The models achieved F1 performance ranging from 77% to 93% for datasets containing 292 or more instances, with the observed trends aligning with MRI theory. The models effectively reflected the practical realities of clinical MRI settings, offering a valuable tool for medical physicists in quality control tasks. In conclusion, AI has demonstrated its potential to optimize MRI protocols, supporting medical physicists in improving image quality and enhancing the efficiency of quality control in clinical practice.
Authors: Alexander Kolesov, Manukhov Stepan, Vladimir V. Palyulin, Alexander Korotin
Abstract: We propose Electrostatic Field Matching (EFM), a novel method that is suitable for both generative modeling and distribution transfer tasks. Our approach is inspired by the physics of an electrical capacitor. We place source and target distributions on the capacitor plates and assign them positive and negative charges, respectively. We then learn the electrostatic field of the capacitor using a neural network approximator. To map the distributions to each other, we start at one plate of the capacitor and move the samples along the learned electrostatic field lines until they reach the other plate. We theoretically justify that this approach provably yields the distribution transfer. In practice, we demonstrate the performance of our EFM in toy and image data experiments.
Authors: Corinna Coupette, Jeremy Wayland, Emily Simons, Bastian Rieck
Abstract: Benchmark datasets have proved pivotal to the success of graph learning, and good benchmark datasets are crucial to guide the development of the field. Recent research has highlighted problems with graph-learning datasets and benchmarking practices -- revealing, for example, that methods which ignore the graph structure can outperform graph-based approaches on popular benchmark datasets. Such findings raise two questions: (1) What makes a good graph-learning dataset, and (2) how can we evaluate dataset quality in graph learning? Our work addresses these questions. As the classic evaluation setup uses datasets to evaluate models, it does not apply to dataset evaluation. Hence, we start from first principles. Observing that graph-learning datasets uniquely combine two modes -- the graph structure and the node features -- , we introduce RINGS, a flexible and extensible mode-perturbation framework to assess the quality of graph-learning datasets based on dataset ablations -- i.e., by quantifying differences between the original dataset and its perturbed representations. Within this framework, we propose two measures -- performance separability and mode complementarity -- as evaluation tools, each assessing, from a distinct angle, the capacity of a graph dataset to benchmark the power and efficacy of graph-learning methods. We demonstrate the utility of our framework for graph-learning dataset evaluation in an extensive set of experiments and derive actionable recommendations for improving the evaluation of graph-learning methods. Our work opens new research directions in data-centric graph learning, and it constitutes a first step toward the systematic evaluation of evaluations.
Authors: Alireza Amiri, Xinting Huang, Mark Rofin, Michael Hahn
Abstract: Chain-of-thought reasoning and scratchpads have emerged as critical tools for enhancing the computational capabilities of transformers. While theoretical results show that polynomial-length scratchpads can extend transformers' expressivity from $TC^0$ to $PTIME$, their required length remains poorly understood. Empirical evidence even suggests that transformers need scratchpads even for many problems in $TC^0$, such as Parity or Multiplication, challenging optimistic bounds derived from circuit complexity. In this work, we initiate the study of systematic lower bounds for the number of CoT steps across different algorithmic problems, in the hard-attention regime. We study a variety of algorithmic problems, and provide bounds that are tight up to logarithmic factors. Overall, these results contribute to emerging understanding of the power and limitations of chain-of-thought reasoning.
Authors: Sidak Pal Singh, Hossein Mobahi, Atish Agarwala, Yann Dauphin
Abstract: Curvature regularization techniques like Sharpness Aware Minimization (SAM) have shown great promise in improving generalization on vision tasks. However, we find that SAM performs poorly in domains like natural language processing (NLP), often degrading performance -- even with twice the compute budget. We investigate the discrepancy across domains and find that in the NLP setting, SAM is dominated by regularization of the logit statistics -- instead of improving the geometry of the function itself. We use this observation to develop an alternative algorithm we call Functional-SAM, which regularizes curvature only through modification of the statistics of the overall function implemented by the neural network, and avoids spurious minimization through logit manipulation. Furthermore, we argue that preconditioning the SAM perturbation also prevents spurious minimization, and when combined with Functional-SAM, it gives further improvements. Our proposed algorithms show improved performance over AdamW and SAM baselines when trained for an equal number of steps, in both fixed-length and Chinchilla-style training settings, at various model scales (including billion-parameter scale). On the whole, our work highlights the importance of more precise characterizations of sharpness in broadening the applicability of curvature regularization to large language models (LLMs).
Authors: Jan Schuchardt, Mina Dalirrooyfard, Jed Guzelkabaagac, Anderson Schneider, Yuriy Nevmyvaka, Stephan G\"unnemann
Abstract: Many forms of sensitive data, such as web traffic, mobility data, or hospital occupancy, are inherently sequential. The standard method for training machine learning models while ensuring privacy for units of sensitive information, such as individual hospital visits, is differentially private stochastic gradient descent (DP-SGD). However, we observe in this work that the formal guarantees of DP-SGD are incompatible with timeseries-specific tasks like forecasting, since they rely on the privacy amplification attained by training on small, unstructured batches sampled from an unstructured dataset. In contrast, batches for forecasting are generated by (1) sampling sequentially structured time series from a dataset, (2) sampling contiguous subsequences from these series, and (3) partitioning them into context and ground-truth forecast windows. We theoretically analyze the privacy amplification attained by this structured subsampling to enable the training of forecasting models with sound and tight event- and user-level privacy guarantees. Towards more private models, we additionally prove how data augmentation amplifies privacy in self-supervised training of sequence models. Our empirical evaluation demonstrates that amplification by structured subsampling enables the training of forecasting models with strong formal privacy guarantees.
Authors: Huakun Luo, Haixu Wu, Hang Zhou, Lanxiang Xing, Yichen Di, Jianmin Wang, Mingsheng Long
Abstract: Although deep models have been widely explored in solving partial differential equations (PDEs), previous works are primarily limited to data only with up to tens of thousands of mesh points, far from the million-point scale required by industrial simulations that involve complex geometries. In the spirit of advancing neural PDE solvers to real industrial applications, we present Transolver++, a highly parallel and efficient neural solver that can accurately solve PDEs on million-scale geometries. Building upon previous advancements in solving PDEs by learning physical states via Transolver, Transolver++ is further equipped with an extremely optimized parallelism framework and a local adaptive mechanism to efficiently capture eidetic physical states from massive mesh points, successfully tackling the thorny challenges in computation and physics learning when scaling up input mesh size. Transolver++ increases the single-GPU input capacity to million-scale points for the first time and is capable of continuously scaling input size in linear complexity by increasing GPUs. Experimentally, Transolver++ yields 13% relative promotion across six standard PDE benchmarks and achieves over 20% performance gain in million-scale high-fidelity industrial simulations, whose sizes are 100$\times$ larger than previous benchmarks, covering car and 3D aircraft designs.
Authors: Markus Krimmel, Jenna Wiens, Karsten Borgwardt, Dexiong Chen
Abstract: Graph generative models often face a critical trade-off between learning complex distributions and achieving fast generation speed. We introduce Autoregressive Noisy Filtration Modeling (ANFM), a novel approach that addresses both challenges. ANFM leverages filtration, a concept from topological data analysis, to transform graphs into short sequences of monotonically increasing subgraphs. This formulation extends the sequence families used in previous autoregressive models. To learn from these sequences, we propose a novel autoregressive graph mixer model. Our experiments suggest that exposure bias might represent a substantial hurdle in autoregressive graph generation and we introduce two mitigation strategies to address it: noise augmentation and a reinforcement learning approach. Incorporating these techniques leads to substantial performance gains, making ANFM competitive with state-of-the-art diffusion models across diverse synthetic and real-world datasets. Notably, ANFM produces remarkably short sequences, achieving a 100-fold speedup in generation time compared to diffusion models. This work marks a significant step toward high-throughput graph generation.
Authors: Matthias Wolff, Florian Eilers, Xiaoyi Jiang
Abstract: In this work we propose CKAN, a complex-valued KAN, to join the intrinsic interpretability of KANs and the advantages of Complex-Valued Neural Networks (CVNNs). We show how to transfer a KAN and the necessary associated mechanisms into the complex domain. To confirm that CKAN meets expectations we conduct experiments on symbolic complex-valued function fitting and physically meaningful formulae as well as on a more realistic dataset from knot theory. Our proposed CKAN is more stable and performs on par or better than real-valued KANs while requiring less parameters and a shallower network architecture, making it more explainable.
Authors: Jiaju Kang, Puyu Han, Yang Chun, Xu Wang, Luqi Gong
Abstract: Phase-sensitive optical time-domain reflectometry ({\Phi}-OTDR) is a widely used distributed fiber optic sensing system in engineering. Machine learning algorithms for {\Phi}-OTDR event classification require high volumes and quality of datasets; however, high-quality datasets are currently extremely scarce in the field, leading to a lack of robustness in models, which is manifested by higher false alarm rates in real-world scenarios. One promising approach to address this issue is to augment existing data using generative models combined with a small amount of real-world data. We explored mapping both {\Phi}-OTDR features in a GAN-based generative pipeline and signal features in a Transformer classifier to hyperbolic space to seek more effective model generalization. The results indicate that state-of-the-art models exhibit stronger generalization performance and lower false alarm rates in real-world scenarios when trained on augmented datasets. TransformDAS, in particular, demonstrates the best classification performance, highlighting the benefits of Riemannian manifold mapping in {\Phi}-OTDR data generation and model classification.
Authors: Depen Morwani, Nikhil Vyas, Hanlin Zhang, Sham Kakade
Abstract: Recent advancements in deep learning optimization have introduced new algorithms, such as Schedule-Free optimizers, AdEMAMix, MARS and Lion which modify traditional momentum mechanisms. In a separate line of work, theoretical acceleration of stochastic gradient descent (SGD) in noise-dominated regime has been achieved by decoupling the momentum coefficient from the current gradient's weight. In this paper, we establish explicit connections between these two lines of work. We substantiate our theoretical findings with preliminary experiments on a 150m language modeling task. We find that AdEMAMix, which most closely resembles accelerated versions of stochastic gradient descent, exhibits superior performance. Building on these insights, we introduce a modification to AdEMAMix, termed Simplified-AdEMAMix, which maintains the same performance as AdEMAMix across both large and small batch-size settings while eliminating the need for two different momentum terms. The code for Simplified-AdEMAMix is available on the repository: https://github.com/DepenM/Simplified-AdEMAMix/.
Authors: Mohammadmehdi Ataei, Hyunmin Cheong, Adrian Butscher
Abstract: Deep neural networks are increasingly employed in fields such as climate modeling, robotics, and industrial control, where strict output constraints must be upheld. Although prior methods like the POLICE algorithm can enforce affine constraints in a single convex region by adjusting network parameters, they struggle with multiple disjoint regions, often leading to conflicts or unintended affine extensions. We present mPOLICE, a new method that extends POLICE to handle constraints imposed on multiple regions. mPOLICE assigns a distinct activation pattern to each constrained region, preserving exact affine behavior locally while avoiding overreach into other parts of the input domain. We formulate a layer-wise optimization problem that adjusts both the weights and biases to assign unique activation patterns to each convex region, ensuring that constraints are met without conflicts, while maintaining the continuity and smoothness of the learned function. Our experiments show the enforcement of multi-region constraints for multiple scenarios, including regression and classification, function approximation, and non-convex regions through approximation. Notably, mPOLICE adds zero inference overhead and minimal training overhead.
Authors: Phil Ostheimer, Mayank Nagda, Marius Kloft, Sophie Fellenz
Abstract: Sparse data is ubiquitous, appearing in numerous domains, from economics and recommender systems to astronomy and biomedical sciences. However, efficiently and realistically generating sparse data remains a significant challenge. We introduce Sparse Data Diffusion (SDD), a novel method for generating sparse data. SDD extends continuous state-space diffusion models by explicitly modeling sparsity through the introduction of Sparsity Bits. Empirical validation on image data from various domains-including two scientific applications, physics and biology-demonstrates that SDD achieves high fidelity in representing data sparsity while preserving the quality of the generated data.
Authors: Satvik Golechha, Maheep Chaudhary, Joan Velja, Alessandro Abate, Nandi Schoots
Abstract: An approach to improve neural network interpretability is via clusterability, i.e., splitting a model into disjoint clusters that can be studied independently. We define a measure for clusterability and show that pre-trained models form highly enmeshed clusters via spectral graph clustering. We thus train models to be more modular using a ``clusterability loss'' function that encourages the formation of non-interacting clusters. Using automated interpretability techniques, we show that our method can help train models that are more modular and learn different, disjoint, and smaller circuits. We investigate CNNs trained on MNIST and CIFAR, small transformers trained on modular addition, and language models. Our approach provides a promising direction for training neural networks that learn simpler functions and are easier to interpret.
Authors: Xiyuan Wang, Muhan Zhang
Abstract: Recent advances in Graph Neural Networks (GNNs) have explored the potential of random noise as an input feature to enhance expressivity across diverse tasks. However, naively incorporating noise can degrade performance, while architectures tailored to exploit noise for specific tasks excel yet lack broad applicability. This paper tackles these issues by laying down a theoretical framework that elucidates the increased sample complexity when introducing random noise into GNNs without careful design. We further propose Equivariant Noise GNN (ENGNN), a novel architecture that harnesses the symmetrical properties of noise to mitigate sample complexity and bolster generalization. Our experiments demonstrate that using noise equivariantly significantly enhances performance on node-level, link-level, subgraph, and graph-level tasks and achieves comparable performance to models designed for specific tasks, thereby offering a general method to boost expressivity across various graph tasks.
Authors: Fabian J. Roth, Dominik K. Klein, Maximilian Kannapinn, Jan Peters, Oliver Weeger
Abstract: In recent years, nonlinear dynamic system identification using artificial neural networks has garnered attention due to its manifold potential applications in virtually all branches of science and engineering. However, purely data-driven approaches often struggle with extrapolation and may yield physically implausible forecasts. Furthermore, the learned dynamics can exhibit instabilities, making it difficult to apply such models safely and robustly. This article proposes stable port-Hamiltonian neural networks, a machine learning architecture that incorporates the physical biases of energy conservation or dissipation while guaranteeing global Lyapunov stability of the learned dynamics. Evaluations with illustrative examples and real-world measurement data demonstrate the model's ability to generalize from sparse data, outperforming purely data-driven approaches and avoiding instability issues. In addition, the model's potential for data-driven surrogate modeling is highlighted in application to multi-physics simulation data.
Authors: Valentin De Bortoli, Alexandre Galashov, J. Swaroop Guntupalli, Guangyao Zhou, Kevin Murphy, Arthur Gretton, Arnaud Doucet
Abstract: Diffusion models generate high-quality synthetic data. They operate by defining a continuous-time forward process which gradually adds Gaussian noise to data until fully corrupted. The corresponding reverse process progressively "denoises" a Gaussian sample into a sample from the data distribution. However, generating high-quality outputs requires many discretization steps to obtain a faithful approximation of the reverse process. This is expensive and has motivated the development of many acceleration methods. We propose to accomplish sample generation by learning the posterior {\em distribution} of clean data samples given their noisy versions, instead of only the mean of this distribution. This allows us to sample from the probability transitions of the reverse process on a coarse time scale, significantly accelerating inference with minimal degradation of the quality of the output. This is accomplished by replacing the standard regression loss used to estimate conditional means with a scoring rule. We validate our method on image and robot trajectory generation, where we consistently outperform standard diffusion models at few discretization steps.
Authors: Xiyuan Wang, Yewei Liu, Lexi Pang, Siwei Chen, Muhan Zhang
Abstract: Diffusion models have gained popularity in graph generation tasks; however, the extent of their expressivity concerning the graph distributions they can learn is not fully understood. Unlike models in other domains, popular backbones for graph diffusion models, such as Graph Transformers, do not possess universal expressivity to accurately model the distribution scores of complex graph data. Our work addresses this limitation by focusing on the frequency of specific substructures as a key characteristic of target graph distributions. When evaluating existing models using this metric, we find that they fail to maintain the distribution of substructure counts observed in the training set when generating new graphs. To address this issue, we establish a theoretical connection between the expressivity of Graph Neural Networks (GNNs) and the overall performance of graph diffusion models, demonstrating that more expressive GNN backbones can better capture complex distribution patterns. By integrating advanced GNNs into the backbone architecture, we achieve significant improvements in substructure generation.
Authors: Yize Wu, Ke Gao, Yanjun Wu
Abstract: Speculative decoding is an effective and lossless method for Large Language Model (LLM) inference acceleration. It employs a smaller model to generate a draft token sequence, which is then verified by the original base model. In multi-GPU systems, inference latency can be further reduced through tensor parallelism (TP), while the optimal TP size of the draft model is typically smaller than that of the base model, leading to GPU idling during the drafting stage. To solve this problem, we propose EasySpec, a layer-parallel speculation strategy that optimizes the efficiency of multi-GPU utilization.EasySpec breaks the sequential execution order of layers in the drafting model, enabling multi-layer parallelization across devices, albeit with some induced approximation errors. After each drafting-and-verification iteration, the draft model's key-value (KV) cache is calibrated in a single forward pass, preventing long-term error accumulation at minimal additional latency. We evaluated EasySpec on several mainstream open-source LLMs, using smaller versions of models from the same series as drafters. The results demonstrate that EasySpec can achieve a peak speedup of 4.17x compared to vanilla decoding, while preserving the original distribution of the base LLMs. Specifically, the drafting stage can be accelerated by up to 1.62x with a maximum accuracy drop of only 7%, requiring no training or fine-tuning on the draft models.
Authors: Dylan Sam, Ayan Chakrabarti, Afshin Rostamizadeh, Srikumar Ramalingam, Gui Citovsky, Sanjiv Kumar
Abstract: Similarity between training examples is used to curate pretraining datasets for language models by many methods -- for diversification and to select examples similar to high-quality data. However, similarity is typically measured with off-the-shelf embedding models that are generic or trained for tasks such as retrieval. This paper introduces a framework to analyze the suitability of embedding models specifically for data curation in the language model pretraining setting. We quantify the correlation between similarity in the embedding space to similarity in pretraining loss between different training examples, and how diversifying in the embedding space affects pretraining quality. We analyze a variety of embedding models in our framework, with experiments using the Pile dataset for pretraining a 1.7B parameter decoder-only language model. We find that the embedding models we consider are all useful for pretraining data curation. Moreover, a simple approach of averaging per-token embeddings proves to be surprisingly competitive with more sophisticated embedding models -- likely because the latter are not designed specifically for pretraining data curation. Indeed, we believe our analysis and evaluation framework can serve as a foundation for the design of embedding models that specifically reason about similarity in pretraining datasets.
Authors: Chris Kolb, Tobias Weber, Bernd Bischl, David R\"ugamer
Abstract: Sparse regularization techniques are well-established in machine learning, yet their application in neural networks remains challenging due to the non-differentiability of penalties like the $L_1$ norm, which is incompatible with stochastic gradient descent. A promising alternative is shallow weight factorization, where weights are decomposed into two factors, allowing for smooth optimization of $L_1$-penalized neural networks by adding differentiable $L_2$ regularization to the factors. In this work, we introduce deep weight factorization, extending previous shallow approaches to more than two factors. We theoretically establish equivalence of our deep factorization with non-convex sparse regularization and analyze its impact on training dynamics and optimization. Due to the limitations posed by standard training practices, we propose a tailored initialization scheme and identify important learning rate requirements necessary for training factorized networks. We demonstrate the effectiveness of our deep weight factorization through experiments on various architectures and datasets, consistently outperforming its shallow counterpart and widely used pruning methods.
Authors: Etienne Meunier, David Kamm, Guillaume Gachon, Redouane Lguensat, Julie Deshayes
Abstract: Ocean General Circulation Models require extensive computational resources to reach equilibrium states, while deep learning emulators, despite offering fast predictions, lack the physical interpretability and long-term stability necessary for climate scientists to understand climate sensitivity (to greenhouse gas emissions) and mechanisms of abrupt % variability such as tipping points. We propose to take the best from both worlds by leveraging deep generative models to produce physically consistent oceanic states that can serve as initial conditions for climate projections. We assess the viability of this hybrid approach through both physical metrics and numerical experiments, and highlight the benefits of enforcing physical constraints during generation. Although we train here on ocean variables from idealized numerical simulations, we claim that this hybrid approach, combining the computational efficiency of deep learning with the physical accuracy of numerical models, can effectively reduce the computational burden of running climate models to equilibrium, and reduce uncertainties in climate projections by minimizing drifts in baseline simulations.
Authors: Marco Bertolini, Tuan Le, Djork-Arn\'e Clevert
Abstract: We extend Euclidean score-based diffusion processes to generative modeling on Lie groups. Through the formalism of Generalized Score Matching, our approach yields a Langevin dynamics which decomposes as a direct sum of Lie algebra representations, enabling generative processes on Lie groups while operating in Euclidean space. Unlike equivariant models, which restrict the space of learnable functions by quotienting out group orbits, our method can model any target distribution on any (non-Abelian) Lie group. Standard score matching emerges as a special case of our framework when the Lie group is the translation group. We prove that our generalized generative processes arise as solutions to a new class of paired stochastic differential equations (SDEs), introduced here for the first time. We validate our approach through experiments on diverse data types, demonstrating its effectiveness in real-world applications such as SO(3)-guided molecular conformer generation and modeling ligand-specific global SE(3) transformations for molecular docking, showing improvement in comparison to Riemannian diffusion on the group itself. We show that an appropriate choice of Lie group enhances learning efficiency by reducing the effective dimensionality of the trajectory space and enables the modeling of transitions between complex data distributions. Additionally, we demonstrate the universality of our approach by deriving how it extends to flow matching.
Authors: Alessio Russo, Aldo Pacchiano
Abstract: We study the policy evaluation problem in an online multi-reward multi-policy discounted setting, where multiple reward functions must be evaluated simultaneously for different policies. We adopt an $(\epsilon,\delta)$-PAC perspective to achieve $\epsilon$-accurate estimates with high confidence across finite or convex sets of rewards, a setting that has not been investigated in the literature. Building on prior work on Multi-Reward Best Policy Identification, we adapt the MR-NaS exploration scheme to jointly minimize sample complexity for evaluating different policies across different reward sets. Our approach leverages an instance-specific lower bound revealing how the sample complexity scales with a measure of value deviation, guiding the design of an efficient exploration policy. Although computing this bound entails a hard non-convex optimization, we propose an efficient convex approximation that holds for both finite and convex reward sets. Experiments in tabular domains demonstrate the effectiveness of this adaptive exploration scheme.
Authors: Sarah Mercer, Samuel Spillard, Daniel P. Martin
Abstract: In late January 2025, DeepSeek released their new reasoning model (DeepSeek R1); which was developed at a fraction of the cost yet remains competitive with OpenAI's models, despite the US's GPU export ban. This report discusses the model, and what its release means for the field of Generative AI more widely. We briefly discuss other models released from China in recent weeks, their similarities; innovative use of Mixture of Experts (MoE), Reinforcement Learning (RL) and clever engineering appear to be key factors in the capabilities of these models. This think piece has been written to a tight time-scale, providing broad coverage of the topic, and serves as introductory material for those looking to understand the model's technical advancements, as well as it's place in the ecosystem. Several further areas of research are identified.
Authors: Si-Yang Liu, Han-Jia Ye
Abstract: TabPFN has emerged as a promising in-context learning model for tabular data, capable of directly predicting the labels of test samples given labeled training examples. It has demonstrated competitive performance, particularly on small-scale classification tasks. However, despite its effectiveness, TabPFN still requires further refinement in several areas, including handling high-dimensional features, aligning with downstream datasets, and scaling to larger datasets. In this paper, we revisit existing variants of TabPFN and observe that most approaches focus either on reducing bias or variance, often neglecting the need to address the other side, while also increasing inference overhead. To fill this gap, we propose Beta (Bagging and Encoder-based Fine-tuning for TabPFN Adaptation), a novel and effective method designed to minimize both bias and variance. To reduce bias, we introduce a lightweight encoder to better align downstream tasks with the pre-trained TabPFN. By increasing the number of encoders in a lightweight manner, Beta mitigate variance, thereby further improving the model's performance. Additionally, bootstrapped sampling is employed to further reduce the impact of data perturbations on the model, all while maintaining computational efficiency during inference. Our approach enhances TabPFN's ability to handle high-dimensional data and scale to larger datasets. Experimental results on over 200 benchmark classification datasets demonstrate that Beta either outperforms or matches state-of-the-art methods.
Authors: Blake Bordelon, Cengiz Pehlevan
Abstract: We theoretically characterize gradient descent dynamics in deep linear networks trained at large width from random initialization and on large quantities of random data. Our theory captures the ``wider is better" effect of mean-field/maximum-update parameterized networks as well as hyperparameter transfer effects, which can be contrasted with the neural-tangent parameterization where optimal learning rates shift with model width. We provide asymptotic descriptions of both non-residual and residual neural networks, the latter of which enables an infinite depth limit when branches are scaled as $1/\sqrt{\text{depth}}$. We also compare training with one-pass stochastic gradient descent to the dynamics when training data are repeated at each iteration. Lastly, we show that this model recovers the accelerated power law training dynamics for power law structured data in the rich regime observed in recent works.
Authors: Han Zhou, Xingchen Wan, Ruoxi Sun, Hamid Palangi, Shariq Iqbal, Ivan Vuli\'c, Anna Korhonen, Sercan \"O. Ar{\i}k
Abstract: Large language models, employed as multiple agents that interact and collaborate with each other, have excelled at solving complex tasks. The agents are programmed with prompts that declare their functionality, along with the topologies that orchestrate interactions across agents. Designing prompts and topologies for multi-agent systems (MAS) is inherently complex. To automate the entire design process, we first conduct an in-depth analysis of the design space aiming to understand the factors behind building effective MAS. We reveal that prompts together with topologies play critical roles in enabling more effective MAS design. Based on the insights, we propose Multi-Agent System Search (MASS), a MAS optimization framework that efficiently exploits the complex MAS design space by interleaving its optimization stages, from local to global, from prompts to topologies, over three stages: 1) block-level (local) prompt optimization; 2) workflow topology optimization; 3) workflow-level (global) prompt optimization, where each stage is conditioned on the iteratively optimized prompts/topologies from former stages. We show that MASS-optimized multi-agent systems outperform a spectrum of existing alternatives by a substantial margin. Based on the MASS-found systems, we finally propose design principles behind building effective multi-agent systems.
Authors: Seohong Park, Qiyang Li, Sergey Levine
Abstract: We present flow Q-learning (FQL), a simple and performant offline reinforcement learning (RL) method that leverages an expressive flow-matching policy to model arbitrarily complex action distributions in data. Training a flow policy with RL is a tricky problem, due to the iterative nature of the action generation process. We address this challenge by training an expressive one-step policy with RL, rather than directly guiding an iterative flow policy to maximize values. This way, we can completely avoid unstable recursive backpropagation, eliminate costly iterative action generation at test time, yet still mostly maintain expressivity. We experimentally show that FQL leads to strong performance across 73 challenging state- and pixel-based OGBench and D4RL tasks in offline RL and offline-to-online RL. Project page: https://seohong.me/projects/fql/
Authors: Abhinav Kumar, Jaechul Roh, Ali Naseh, Marzena Karpinska, Mohit Iyyer, Amir Houmansadr, Eugene Bagdasarian
Abstract: We increase overhead for applications that rely on reasoning LLMs-we force models to spend an amplified number of reasoning tokens, i.e., "overthink", to respond to the user query while providing contextually correct answers. The adversary performs an OVERTHINK attack by injecting decoy reasoning problems into the public content that is used by the reasoning LLM (e.g., for RAG applications) during inference time. Due to the nature of our decoy problems (e.g., a Markov Decision Process), modified texts do not violate safety guardrails. We evaluated our attack across closed-(OpenAI o1, o1-mini, o3-mini) and open-(DeepSeek R1) weights reasoning models on the FreshQA and SQuAD datasets. Our results show up to 46x slowdown and high transferability of the attack across models. To protect applications, we discuss and implement defenses leveraging LLM-based and system design approaches. Finally, we discuss societal, financial, and energy impacts of OVERTHINK attack which could amplify the costs for third party applications operating reasoning models.
Authors: Zhiyuan Wu, Changkyu Choi, Xiangcheng Cao, Volkan Cevher, Ali Ramezani-Kebrya
Abstract: We address the challenge of minimizing true risk in multi-node distributed learning. These systems are frequently exposed to both inter-node and intra-node label shifts, which present a critical obstacle to effectively optimizing model performance while ensuring that data remains confined to each node. To tackle this, we propose the Versatile Robust Label Shift (VRLS) method, which enhances the maximum likelihood estimation of the test-to-train label density ratio. VRLS incorporates Shannon entropy-based regularization and adjusts the density ratio during training to better handle label shifts at the test time. In multi-node learning environments, VRLS further extends its capabilities by learning and adapting density ratios across nodes, effectively mitigating label shifts and improving overall model performance. Experiments conducted on MNIST, Fashion MNIST, and CIFAR-10 demonstrate the effectiveness of VRLS, outperforming baselines by up to 20% in imbalanced settings. These results highlight the significant improvements VRLS offers in addressing label shifts. Our theoretical analysis further supports this by establishing high-probability bounds on estimation errors.
Authors: Leonardo Defilippis, Yatin Dandi, Pierre Mergny, Florent Krzakala, Bruno Loureiro
Abstract: We consider the problem of how many samples from a Gaussian multi-index model are required to weakly reconstruct the relevant index subspace. Despite its increasing popularity as a testbed for investigating the computational complexity of neural networks, results beyond the single-index setting remain elusive. In this work, we introduce spectral algorithms based on the linearization of a message passing scheme tailored to this problem. Our main contribution is to show that the proposed methods achieve the optimal reconstruction threshold. Leveraging a high-dimensional characterization of the algorithms, we show that above the critical threshold the leading eigenvector correlates with the relevant index subspace, a phenomenon reminiscent of the Baik-Ben Arous-Peche (BBP) transition in spiked models arising in random matrix theory. Supported by numerical experiments and a rigorous theoretical framework, our work bridges critical gaps in the computational limits of weak learnability in multi-index model.
Authors: Haonan Zhu, Andre R. Goncalves, Camilo Valdes, Hiranmayi Ranganathan, Boya Zhang, Jose Manuel Mart\'i, Car Reen Kok, Monica K. Borucki, Nisha J. Mulakken, James B. Thissen, Crystal Jaing, Alfred Hero, Nicholas A. Be
Abstract: This paper proposes a hierarchical Bayesian multitask learning model that is applicable to the general multi-task binary classification learning problem where the model assumes a shared sparsity structure across different tasks. We derive a computationally efficient inference algorithm based on variational inference to approximate the posterior distribution. We demonstrate the potential of the new approach on various synthetic datasets and for predicting human health status based on microbiome profile. Our analysis incorporates data pooled from multiple microbiome studies, along with a comprehensive comparison with other benchmark methods. Results in synthetic datasets show that the proposed approach has superior support recovery property when the underlying regression coefficients share a common sparsity structure across different tasks. Our experiments on microbiome classification demonstrate the utility of the method in extracting informative taxa while providing well-calibrated predictions with uncertainty quantification and achieving competitive performance in terms of prediction metrics. Notably, despite the heterogeneity of the pooled datasets (e.g., different experimental objectives, laboratory setups, sequencing equipment, patient demographics), our method delivers robust results.
Authors: Shayan Kiyani, George Pappas, Aaron Roth, Hamed Hassani
Abstract: A fundamental question in data-driven decision making is how to quantify the uncertainty of predictions in ways that can usefully inform downstream action. This interface between prediction uncertainty and decision-making is especially important in risk-sensitive domains, such as medicine. In this paper, we develop decision-theoretic foundations that connect uncertainty quantification using prediction sets with risk-averse decision-making. Specifically, we answer three fundamental questions: (1) What is the correct notion of uncertainty quantification for risk-averse decision makers? We prove that prediction sets are optimal for decision makers who wish to optimize their value at risk. (2) What is the optimal policy that a risk averse decision maker should use to map prediction sets to actions? We show that a simple max-min decision policy is optimal for risk-averse decision makers. Finally, (3) How can we derive prediction sets that are optimal for such decision makers? We provide an exact characterization in the population regime and a distribution free finite-sample construction. Answering these questions naturally leads to an algorithm, Risk-Averse Calibration (RAC), which follows a provably optimal design for deriving action policies from predictions. RAC is designed to be both practical-capable of leveraging the quality of predictions in a black-box manner to enhance downstream utility-and safe-adhering to a user-defined risk threshold and optimizing the corresponding risk quantile of the user's downstream utility. Finally, we experimentally demonstrate the significant advantages of RAC in applications such as medical diagnosis and recommendation systems. Specifically, we show that RAC achieves a substantially improved trade-off between safety and utility, offering higher utility compared to existing methods while maintaining the safety guarantee.
Authors: Connor Schenck, Isaac Reid, Mithun George Jacob, Alex Bewley, Joshua Ainslie, David Rendleman, Deepali Jain, Mohit Sharma, Avinava Dubey, Ayzaan Wahid, Sumeet Singh, Rene Wagner, Tianli Ding, Chuyuan Fu, Arunkumar Byravan, Jake Varley, Alexey Gritsenko, Matthias Minderer, Dmitry Kalashnikov, Jonathan Tompson, Vikas Sindhwani, Krzysztof Choromanski
Abstract: We introduce STRING: Separable Translationally Invariant Position Encodings. STRING extends Rotary Position Encodings, a recently proposed and widely used algorithm in large language models, via a unifying theoretical framework. Importantly, STRING still provides exact translation invariance, including token coordinates of arbitrary dimensionality, whilst maintaining a low computational footprint. These properties are especially important in robotics, where efficient 3D token representation is key. We integrate STRING into Vision Transformers with RGB(-D) inputs (color plus optional depth), showing substantial gains, e.g. in open-vocabulary object detection and for robotics controllers. We complement our experiments with a rigorous mathematical analysis, proving the universality of our methods.
Authors: Tianyang Xie, Yong Ge
Abstract: Survival analysis, a vital tool for predicting the time to event, has been used in many domains such as healthcare, criminal justice, and finance. Like classification tasks, survival analysis can exhibit bias against disadvantaged groups, often due to biases inherent in data or algorithms. Several studies in both the IS and CS communities have attempted to address fairness in survival analysis. However, existing methods often overlook the importance of prediction fairness at pre-defined evaluation time points, which is crucial in real-world applications where decision making often hinges on specific time frames. To address this critical research gap, we introduce a new fairness concept: equalized odds (EO) in survival analysis, which emphasizes prediction fairness at pre-defined time points. To achieve the EO fairness in survival analysis, we propose a Conditional Mutual Information Augmentation (CMIA) approach, which features a novel fairness regularization term based on conditional mutual information and an innovative censored data augmentation technique. Our CMIA approach can effectively balance prediction accuracy and fairness, and it is applicable to various survival models. We evaluate the CMIA approach against several state-of-the-art methods within three different application domains, and the results demonstrate that CMIA consistently reduces prediction disparity while maintaining good accuracy and significantly outperforms the other competing methods across multiple datasets and survival models (e.g., linear COX, deep AFT).
Authors: Philipp Hoellmer, Thomas Egg, Maya M. Martirossyan, Eric Fuemmeler, Amit Gupta, Zeren Shui, Pawan Prakash, Adrian Roitberg, Mingjie Liu, George Karypis, Mark Transtrum, Richard G. Hennig, Ellad B. Tadmor, Stefano Martiniani
Abstract: The discovery of new materials is essential for enabling technological advancements. Computational approaches for predicting novel materials must effectively learn the manifold of stable crystal structures within an infinite design space. We introduce Open Materials Generation (OMG), a unifying framework for the generative design and discovery of inorganic crystalline materials. OMG employs stochastic interpolants (SI) to bridge an arbitrary base distribution to the target distribution of inorganic crystals via a broad class of tunable stochastic processes, encompassing both diffusion models and flow matching as special cases. In this work, we adapt the SI framework by integrating an equivariant graph representation of crystal structures and extending it to account for periodic boundary conditions in unit cell representations. Additionally, we couple the SI flow over spatial coordinates and lattice vectors with discrete flow matching for atomic species. We benchmark OMG's performance on two tasks: Crystal Structure Prediction (CSP) for specified compositions, and 'de novo' generation (DNG) aimed at discovering stable, novel, and unique structures. In our ground-up implementation of OMG, we refine and extend both CSP and DNG metrics compared to previous works. OMG establishes a new state-of-the-art in generative modeling for materials discovery, outperforming purely flow-based and diffusion-based implementations. These results underscore the importance of designing flexible deep learning frameworks to accelerate progress in materials science.
Authors: Zongyu Lin, Yao Tang, Xingcheng Yao, Da Yin, Ziniu Hu, Yizhou Sun, Kai-Wei Chang
Abstract: Language agents have become a promising solution to complex interactive tasks. One of the key ingredients to the success of language agents is the reward model on the trajectory of the agentic workflow, which provides valuable guidance during training or inference. However, due to the lack of annotations of intermediate interactions, most existing works use an outcome reward model to optimize policies across entire trajectories. This may lead to sub-optimal policies and hinder the overall performance. To address this, we propose QLASS (Q-guided Language Agent Stepwise Search), to automatically generate annotations by estimating Q-values in a stepwise manner for open language agents. By introducing a reasoning tree and performing process reward modeling, QLASS provides effective intermediate guidance for each step. With the stepwise guidance, we propose a Q-guided generation strategy to enable language agents to better adapt to long-term value, resulting in significant performance improvement during model inference on complex interactive agent tasks. Notably, even with almost half the annotated data, QLASS retains strong performance, demonstrating its efficiency in handling limited supervision. We also empirically demonstrate that QLASS can lead to more effective decision making through qualitative analysis. We will release our code and data.
Authors: Abhinav Pratap, Amit Pathak
Abstract: This paper examines how social media algorithms and filter bubbles contribute to the fragmentation of online discourse, fostering ideological divides and undermining shared understanding. Drawing on Michael Sandels philosophical emphasis on community and shared values, the study explores how digital platforms amplify discrimination discourse including sexism, racism, xenophobia, ableism, homophobia, and religious intolerance during periods of heightened societal tension. By analyzing the dynamics of digital communities, the research highlights mechanisms driving the emergence and evolution of discourse fragments in response to real world events. The findings reveal how social media structures exacerbate polarization, restrict cross group dialogue, and erode the collective reasoning essential for a just society. This study situates philosophical perspectives within a computational analysis of social media interactions, offering a nuanced understanding of the challenges posed by fragmented discourse in the digital age.
Authors: Abhinav Pratap, Sushant Kumar, Suchinton Chakravarty
Abstract: This study addresses the need for accurate and efficient object detection in assistive technologies for visually impaired individuals. We evaluate four real-time object detection algorithms YOLO, SSD, Faster R-CNN, and Mask R-CNN within the context of indoor navigation assistance. Using the Indoor Objects Detection dataset, we analyze detection accuracy, processing speed, and adaptability to indoor environments. Our findings highlight the trade-offs between precision and efficiency, offering insights into selecting optimal algorithms for realtime assistive navigation. This research advances adaptive machine learning applications, enhancing indoor navigation solutions for the visually impaired and promoting accessibility.
Authors: Afsara Benazir, Felix Xiaozhu Lin
Abstract: Robust speech recognition systems rely on cloud service providers for inference. It needs to ensure that an untrustworthy provider cannot deduce the sensitive content in speech. Sanitization can be done on speech content keeping in mind that it has to avoid compromising transcription accuracy. Realizing the under utilized capabilities of tiny speech foundation models (FMs), for the first time, we propose a novel use: enhancing speech privacy on resource-constrained devices. We introduce XYZ, an edge/cloud privacy preserving speech inference engine that can filter sensitive entities without compromising transcript accuracy. We utilize a timestamp based on-device masking approach that utilizes a token to entity prediction model to filter sensitive entities. Our choice of mask strategically conceals parts of the input and hides sensitive data. The masked input is sent to a trusted cloud service or to a local hub to generate the masked output. The effectiveness of XYZ hinges on how well the entity time segments are masked. Our recovery is a confidence score based approach that chooses the best prediction between cloud and on-device model. We implement XYZ on a 64 bit Raspberry Pi 4B. Experiments show that our solution leads to robust speech recognition without forsaking privacy. XYZ with < 100 MB memory, achieves state-of-the-art (SOTA) speech transcription performance while filtering about 83% of private entities directly on-device. XYZ is 16x smaller in memory and 17x more compute efficient than prior privacy preserving speech frameworks and has a relative reduction in word error rate (WER) by 38.8-77.5% when compared to existing offline transcription services.
Authors: Jiale Fu, Yuchu Jiang, Junkai Chen, Jiaming Fan, Xin Geng, Xu Yang
Abstract: Ensemble methods enhance Large Language Models (LLMs) by combining multiple models but suffer from high computational costs. In this paper, we introduce Speculative Ensemble, a novel framework that accelerates LLM ensembles without sacrificing performance, inspired by Speculative Decoding-where a small proposal model generates tokens sequentially, and a larger target model verifies them in parallel. Our approach builds on two key insights: (1) the verification distribution can be the ensemble distribution of both the proposal and target models, and (2) alternating each model as the proposer and verifier can further enhance efficiency. We generalize this method to ensembles with n models and theoretically prove that SE is never slower than a standard ensemble, typically achieving faster speed. Extensive experiments demonstrate speed improvements of 1.11x-2.23x over standard ensemble techniques without compromising generation quality. Our code is available at https://github.com/Kamichanw/Speculative-Ensemble/
Authors: Jingming Xia, Guanqun Cao, Guang Ma, Yiben Luo, Qinzhao Li, John Oyekan
Abstract: Monocular depth estimation involves predicting depth from a single RGB image and plays a crucial role in applications such as autonomous driving, robotic navigation, 3D reconstruction, etc. Recent advancements in learning-based methods have significantly improved depth estimation performance. Generative models, particularly Stable Diffusion, have shown remarkable potential in recovering fine details and reconstructing missing regions through large-scale training on diverse datasets. However, models like CLIP, which rely on textual embeddings, face limitations in complex outdoor environments where rich context information is needed. These limitations reduce their effectiveness in such challenging scenarios. Here, we propose a novel image-based semantic embedding that extracts contextual information directly from visual features, significantly improving depth prediction in complex environments. Evaluated on the KITTI and Waymo datasets, our method achieves performance comparable to state-of-the-art models while addressing the shortcomings of CLIP embeddings in handling outdoor scenes. By leveraging visual semantics directly, our method demonstrates enhanced robustness and adaptability in depth estimation tasks, showcasing its potential for application to other visual perception tasks.
Authors: Shupeng Ning, Hanqing Zhu, Chenghao Feng, Jiaqi Gu, David Z. Pan, Ray T. Chen
Abstract: Recent advancements in artificial intelligence (AI) and deep neural networks (DNNs) have revolutionized numerous fields, enabling complex tasks by extracting intricate features from large datasets. However, the exponential growth in computational demands has outstripped the capabilities of traditional electrical hardware accelerators. Optical computing offers a promising alternative due to its inherent advantages of parallelism, high computational speed, and low power consumption. Yet, current photonic integrated circuits (PICs) designed for general matrix multiplication (GEMM) are constrained by large footprints, high costs of electro-optical (E-O) interfaces, and high control complexity, limiting their scalability. To overcome these challenges, we introduce a block-circulant photonic tensor core (CirPTC) for a structure-compressed optical neural network (StrC-ONN) architecture. By applying a structured compression strategy to weight matrices, StrC-ONN significantly reduces model parameters and hardware requirements while preserving the universal representability of networks and maintaining comparable expressivity. Additionally, we propose a hardware-aware training framework to compensate for on-chip nonidealities to improve model robustness and accuracy. We experimentally demonstrate image processing and classification tasks, achieving up to a 74.91% reduction in trainable parameters while maintaining competitive accuracies. Performance analysis expects a computational density of 5.84 tera operations per second (TOPS) per mm^2 and a power efficiency of 47.94 TOPS/W, marking a 6.87-times improvement achieved through the hardware-software co-design approach. By reducing both hardware requirements and control complexity across multiple dimensions, this work explores a new pathway to push the limits of optical computing in the pursuit of high efficiency and scalability.
Authors: Manqing Liu, Andrew L. Beam
Abstract: We present Doubly Robust Monte Carlo Tree Search (DR-MCTS), a novel algorithm that integrates Doubly Robust (DR) off-policy estimation into Monte Carlo Tree Search (MCTS) to enhance sample efficiency and decision quality in complex environments. Our approach introduces a hybrid estimator that combines MCTS rollouts with DR estimation, offering theoretical guarantees of unbiasedness and variance reduction under specified conditions. Empirical evaluations in Tic-Tac-Toe and the partially observable VirtualHome environment demonstrate DR-MCTS's superior performance over standard MCTS. In Tic-Tac-Toe, DR-MCTS achieves an 88% win rate compared to a 10% win rate for standard MCTS. In compound VirtualHome tasks, DR-MCTS attains a 20.7% success rate versus 10.3% for standard MCTS. Our scaling analysis reveals that DR-MCTS exhibits better sample efficiency, notably outperforming standard MCTS with larger language models while using a smaller model. These results underscore DR-MCTS's potential for efficient decision-making in complex, real-world scenarios where sample efficiency is paramount.
Authors: Francesco Pezone
Abstract: As digital technologies advance, communication networks face challenges in handling the vast data generated by intelligent devices. Autonomous vehicles, smart sensors, and IoT systems necessitate new paradigms. This thesis addresses these challenges by integrating semantic communication and generative models for optimized image compression and edge network resource allocation. Unlike bit-centric systems, semantic communication prioritizes transmitting meaningful data specifically selected to convey the meaning rather than obtain a faithful representation of the original data. The communication infrastructure can benefit to significant improvements in bandwidth efficiency and latency reduction. Central to this work is the design of semantic-preserving image compression using Generative Adversarial Networks and Denoising Diffusion Probabilistic Models. These models compress images by encoding only semantically relevant features, allowing for high-quality reconstruction with minimal transmission. Additionally, a Goal-Oriented edge network optimization framework is introduced, leveraging the Information Bottleneck principle and stochastic optimization to dynamically allocate resources and enhance efficiency. By integrating semantic communication into edge networks, this approach balances computational efficiency and communication effectiveness, making it suitable for real-time applications. The thesis compares semantic-aware models with conventional image compression techniques using classical and semantic evaluation metrics. Results demonstrate the potential of combining generative AI and semantic communication to create more efficient semantic-goal-oriented communication networks that meet the demands of modern data-driven applications.
Authors: Bo Pang, Tingrui Qiao, Caroline Walker, Chris Cunningham, Yun Sing Koh
Abstract: Large Language Models (LLMs) have significantly advanced natural language processing applications, yet their widespread use raises concerns regarding inherent biases that may reduce utility or harm for particular social groups. Despite the advancement in addressing LLM bias, existing research has two major limitations. First, existing LLM bias evaluation focuses on the U.S. cultural context, making it challenging to reveal stereotypical biases of LLMs toward other cultures, leading to unfair development and use of LLMs. Second, current bias evaluation often assumes models are familiar with the target social groups. When LLMs encounter words beyond their knowledge boundaries that are unfamiliar in their training data, they produce irrelevant results in the local context due to hallucinations and overconfidence, which are not necessarily indicative of inherent bias. This research addresses these limitations with a Local Integrated Bias Recognition and Assessment Framework (LIBRA) for measuring bias using datasets sourced from local corpora without crowdsourcing. Implementing this framework, we develop a dataset comprising over 360,000 test cases in the New Zealand context. Furthermore, we propose the Enhanced Idealized CAT Score (EiCAT), integrating the iCAT score with a beyond knowledge boundary score (bbs) and a distribution divergence-based bias measurement to tackle the challenge of LLMs encountering words beyond knowledge boundaries. Our results show that the BERT family, GPT-2, and Llama-3 models seldom understand local words in different contexts. While Llama-3 exhibits larger bias, it responds better to different cultural contexts. The code and dataset are available at: https://github.com/ipangbo/LIBRA.
Authors: Juno Kim, Denny Wu, Jason Lee, Taiji Suzuki
Abstract: A key paradigm to improve the reasoning capabilities of large language models (LLMs) is to allocate more inference-time compute to search against a verifier or reward model. This process can then be utilized to refine the pretrained model or distill its reasoning patterns into more efficient models. In this paper, we study inference-time compute by viewing chain-of-thought (CoT) generation as a metastable Markov process: easy reasoning steps (e.g., algebraic manipulations) form densely connected clusters, while hard reasoning steps (e.g., applying a relevant theorem) create sparse, low-probability edges between clusters, leading to phase transitions at longer timescales. Under this framework, we prove that implementing a search protocol that rewards sparse edges improves CoT by decreasing the expected number of steps to reach different clusters. In contrast, we establish a limit on reasoning capability when the model is restricted to local information of the pretrained graph. We also show that the information gained by search can be utilized to obtain a better reasoning model: (1) the pretrained model can be directly finetuned to favor sparse edges via policy gradient methods, and moreover (2) a compressed metastable representation of the reasoning dynamics can be distilled into a smaller, more efficient model.
Authors: Alan Zhu, Parth Asawa, Jared Quincy Davis, Lingjiao Chen, Ion Stoica, Joseph E. Gonzalez, Matei Zaharia
Abstract: As the demand for high-quality data in model training grows, researchers and developers are increasingly generating synthetic data to tune and train LLMs. A common assumption about synthetic data is that sampling from instruct-tuned models is sufficient; however, these models struggle to produce diverse outputs-a key requirement for generalization. Despite various prompting methods, in this work we show that achieving meaningful diversity from instruct-tuned models remains challenging. In contrast, we find base models without post-training exhibit greater diversity, but are less capable at instruction following and hence of lower quality. Leveraging this insight, we propose Base-Refine (BARE), a synthetic data generation method that combines the diversity of base models with the quality of instruct-tuned models through a two-stage process. With minimal few-shot examples and curation, BARE generates diverse and high-quality datasets, improving downstream task performance. We show that fine-tuning with as few as 1,000 BARE-generated samples can reach performance comparable to the best similarly sized models on LiveCodeBench tasks. Furthermore, fine-tuning with BARE-generated data achieves a 101% improvement over instruct-only data on GSM8K and a 18.4% improvement over SOTA methods on RAFT.
Authors: Christopher J. Anders, Kim A. Nicoli, Bingting Wu, Naima Elosegui, Samuele Pedrielli, Lena Funcke, Karl Jansen, Stefan K\"uhn, Shinichi Nakajima
Abstract: The objective to be minimized in the variational quantum eigensolver (VQE) has a restricted form, which allows a specialized sequential minimal optimization (SMO) that requires only a few observations in each iteration. However, the SMO iteration is still costly due to the observation noise -- one observation at a point typically requires averaging over hundreds to thousands of repeated quantum measurement shots for achieving a reasonable noise level. In this paper, we propose an adaptive cost control method, named subspace in confident region (SubsCoRe), for SMO. SubsCoRe uses the Gaussian process (GP) surrogate, and requires it to have low uncertainty over the subspace being updated, so that optimization in each iteration is performed with guaranteed accuracy. The adaptive cost control is performed by first setting the required accuracy according to the progress of the optimization, and then choosing the minimum number of measurement shots and their distribution such that the required accuracy is satisfied. We demonstrate that SubsCoRe significantly improves the efficiency of SMO, and outperforms the state-of-the-art methods.
Authors: Alexei Figueroa, Justus Westerhoff, Atefi Golzar, Dennis Fast, Benjamin Winter, Felix Alexader Gers, Alexander L\"oser, Wolfang Nejdl
Abstract: Biologically inspired neural networks offer alternative avenues to model data distributions. FlyVec is a recent example that draws inspiration from the fruit fly's olfactory circuit to tackle the task of learning word embeddings. Surprisingly, this model performs competitively even against deep learning approaches specifically designed to encode text, and it does so with the highest degree of computational efficiency. We pose the question of whether this performance can be improved further. For this, we introduce Comply. By incorporating positional information through complex weights, we enable a single-layer neural network to learn sequence representations. Our experiments show that Comply not only supersedes FlyVec but also performs on par with significantly larger state-of-the-art models. We achieve this without additional parameters. Comply yields sparse contextual representations of sentences that can be interpreted explicitly from the neuron weights.
Authors: Christopher Simic, Korbinian Riedhammer, Tobias Bocklet
Abstract: We present an approach to Audio-Visual Speech Recognition that builds on a pre-trained Whisper model. To infuse visual information into this audio-only model, we extend it with an AV fusion module and LoRa adapters, one of the most up-to-date adapter approaches. One advantage of adapter-based approaches, is that only a relatively small number of parameters are trained, while the basic model remains unchanged. Common AVSR approaches train single models to handle several noise categories and noise levels simultaneously. Taking advantage of the lightweight nature of adapter approaches, we train noise-scenario-specific adapter-sets, each covering individual noise-categories or a specific noise-level range. The most suitable adapter-set is selected by previously classifying the noise-scenario. This enables our models to achieve an optimum coverage across different noise-categories and noise-levels, while training only a minimum number of parameters. Compared to a full fine-tuning approach with SOTA performance our models achieve almost comparable results over the majority of the tested noise-categories and noise-levels, with up to 88.5% less trainable parameters. Our approach can be extended by further noise-specific adapter-sets to cover additional noise scenarios. It is also possible to utilize the underlying powerful ASR model when no visual information is available, as it remains unchanged.
Authors: Floris Holstege, Mackenzie Jorgensen, Kirtan Padh, Jurriaan Parie, Joel Persson, Krsto Prorokovic, Lukas Snoek
Abstract: Algorithms are increasingly used to automate or aid human decisions, yet recent research shows that these algorithms may exhibit bias across legally protected demographic groups. However, data on these groups may be unavailable to organizations or external auditors due to privacy legislation. This paper studies bias detection using an unsupervised clustering tool when data on demographic groups are unavailable. We collaborate with the Dutch Executive Agency for Education to audit an algorithm that was used to assign risk scores to college students at the national level in the Netherlands between 2012-2023. Our audit covers more than 250,000 students from the whole country. The unsupervised clustering tool highlights known disparities between students with a non-European migration background and Dutch origin. Our contributions are three-fold: (1) we assess bias in a real-world, large-scale and high-stakes decision-making process by a governmental organization; (2) we use simulation studies to highlight potential pitfalls of using the unsupervised clustering tool to detect true bias when demographic group data are unavailable and provide recommendations for valid inferences; (3) we provide the unsupervised clustering tool in an open-source library. Our work serves as a starting point for a deliberative assessment by human experts to evaluate potential discrimination in algorithmic-supported decision-making processes.
Authors: Nupur Kumari, Xi Yin, Jun-Yan Zhu, Ishan Misra, Samaneh Azadi
Abstract: Customization of text-to-image models enables users to insert custom concepts and generate the concepts in unseen settings. Existing methods either rely on costly test-time optimization or train encoders on single-image training datasets without multi-image supervision, leading to worse image quality. We propose a simple approach that addresses both limitations. We first leverage existing text-to-image models and 3D datasets to create a high-quality Synthetic Customization Dataset (SynCD) consisting of multiple images of the same object in different lighting, backgrounds, and poses. We then propose a new encoder architecture based on shared attention mechanisms that better incorporate fine-grained visual details from input images. Finally, we propose a new inference technique that mitigates overexposure issues during inference by normalizing the text and image guidance vectors. Through extensive experiments, we show that our model, trained on the synthetic dataset with the proposed encoder and inference algorithm, outperforms existing tuning-free methods on standard customization benchmarks.
Authors: Nina Corvelo Benz, Stratis Tsirtsis, Eleni Straitouri, Ivi Chatzi, Ander Artola Velasco, Suhas Thejaswi, Manuel Gomez-Rodriguez
Abstract: State of the art large language models rely on randomization to respond to a prompt. As an immediate consequence, a model may respond differently to the same prompt if asked multiple times. In this work, we argue that the evaluation and ranking of large language models should control for the randomization underpinning their functioning. Our starting point is the development of a causal model for coupled autoregressive generation, which allows different large language models to sample responses with the same source of randomness. Building upon our causal model, we first show that, on evaluations based on benchmark datasets, coupled autoregressive generation leads to the same conclusions as vanilla autoregressive generation but using provably fewer samples. However, we further show that, on evaluations based on (human) pairwise comparisons, coupled and vanilla autoregressive generation can surprisingly lead to different rankings when comparing more than two models, even with an infinite amount of samples. This suggests that the apparent advantage of a model over others in existing evaluation protocols may not be genuine but rather confounded by the randomness inherent to the generation process. To illustrate and complement our theoretical results, we conduct experiments with several large language models from the Llama family. We find that, across multiple knowledge areas from the popular MMLU benchmark dataset, coupled autoregressive generation requires up to 40% fewer samples to reach the same conclusions as vanilla autoregressive generation. Further, using data from the LMSYS Chatbot Arena platform, we find that the win-rates derived from pairwise comparisons by a strong large language model to prompts differ under coupled and vanilla autoregressive generation.
Authors: Haocheng Xi, Shuo Yang, Yilong Zhao, Chenfeng Xu, Muyang Li, Xiuyu Li, Yujun Lin, Han Cai, Jintao Zhang, Dacheng Li, Jianfei Chen, Ion Stoica, Kurt Keutzer, Song Han
Abstract: Diffusion Transformers (DiTs) dominate video generation but their high computational cost severely limits real-world applicability, usually requiring tens of minutes to generate a few seconds of video even on high-performance GPUs. This inefficiency primarily arises from the quadratic computational complexity of 3D Full Attention with respect to the context length. In this paper, we propose a training-free framework termed Sparse VideoGen (SVG) that leverages the inherent sparsity in 3D Full Attention to boost inference efficiency. We reveal that the attention heads can be dynamically classified into two groups depending on distinct sparse patterns: (1) Spatial Head, where only spatially-related tokens within each frame dominate the attention output, and (2) Temporal Head, where only temporally-related tokens across different frames dominate. Based on this insight, SVG proposes an online profiling strategy to capture the dynamic sparse patterns and predicts the type of attention head. Combined with a novel hardware-efficient tensor layout transformation and customized kernel implementations, SVG achieves up to 2.28x and 2.33x end-to-end speedup on CogVideoX-v1.5 and HunyuanVideo, respectively, while preserving generation quality.
Authors: Hongju Park, Shuyang Bai, Zhenyao Ye, Hwiyoung Lee, Tianzhou Ma, Shuo Chen
Abstract: Canonical correlation analysis (CCA) is a widely used technique for estimating associations between two sets of multi-dimensional variables. Recent advancements in CCA methods have expanded their application to decipher the interactions of multiomics datasets, imaging-omics datasets, and more. However, conventional CCA methods are limited in their ability to incorporate structured patterns in the cross-correlation matrix, potentially leading to suboptimal estimations. To address this limitation, we propose the graph Canonical Correlation Analysis (gCCA) approach, which calculates canonical correlations based on the graph structure of the cross-correlation matrix between the two sets of variables. We develop computationally efficient algorithms for gCCA, and provide theoretical results for finite sample analysis of best subset selection and canonical correlation estimation by introducing concentration inequalities and stopping time rule based on martingale theories. Extensive simulations demonstrate that gCCA outperforms competing CCA methods. Additionally, we apply gCCA to a multiomics dataset of DNA methylation and RNA-seq transcriptomics, identifying both positively and negatively regulated gene expression pathways by DNA methylation pathways.
Authors: Haruka Kiyohara, Fan Yao, Sarah Dean
Abstract: In two-sided platforms (e.g., video streaming or e-commerce), viewers and providers engage in interactive dynamics, where an increased provider population results in higher viewer utility and the increase of viewer population results in higher provider utility. Despite the importance of such "population effects" on long-term platform health, recommendation policies do not generally take the participation dynamics into account. This paper thus studies the dynamics and policy design on two-sided platforms under the population effects for the first time. Our control- and game-theoretic findings warn against the use of myopic-greedy policy and shed light on the importance of provider-side considerations (i.e., effectively distributing exposure among provider groups) to improve social welfare via population growth. We also present a simple algorithm to optimize long-term objectives by considering the population effects, and demonstrate its effectiveness in synthetic and real-data experiments.
Authors: Aidan Curtis, Eric Li, Michael Noseworthy, Nishad Gothoskar, Sachin Chitta, Hui Li, Leslie Pack Kaelbling, Nicole Carey
Abstract: Domain randomization in reinforcement learning is an established technique for increasing the robustness of control policies trained in simulation. By randomizing environment properties during training, the learned policy can become robust to uncertainties along the randomized dimensions. While the environment distribution is typically specified by hand, in this paper we investigate automatically discovering a sampling distribution via entropy-regularized reward maximization of a normalizing-flow-based neural sampling distribution. We show that this architecture is more flexible and provides greater robustness than existing approaches that learn simpler, parameterized sampling distributions, as demonstrated in six simulated and one real-world robotics domain. Lastly, we explore how these learned sampling distributions, combined with a privileged value function, can be used for out-of-distribution detection in an uncertainty-aware multi-step manipulation planner.
Authors: Diyana Muhammed, Gollam Rabby, S\"oren Auer
Abstract: Detecting hallucinations in Large Language Models (LLMs) remains a critical challenge for their reliable deployment in real-world applications. To address this, we introduce SelfCheckAgent, a novel framework integrating three different agents: the Symbolic Agent, the Specialized Detection Agent, and the Contextual Consistency Agent. These agents provide a robust multi-dimensional approach to hallucination detection. Notable results include the Contextual Consistency Agent leveraging Llama 3.1 with Chain-of-Thought (CoT) to achieve outstanding performance on the WikiBio dataset, with NonFactual hallucination detection scoring 93.64%, Factual 70.26%, and Ranking 78.48% respectively. On the AIME dataset, GPT-4o with CoT excels in NonFactual detection with 94.89% but reveals trade-offs in Factual with 30.58% and Ranking with 30.68%, underscoring the complexity of hallucination detection in the complex mathematical domains. The framework also incorporates a triangulation strategy, which increases the strengths of the SelfCheckAgent, yielding significant improvements in real-world hallucination identification. The comparative analysis demonstrates SelfCheckAgent's applicability across diverse domains, positioning it as a crucial advancement for trustworthy LLMs. These findings highlight the potentiality of consistency-driven methodologies in detecting hallucinations in LLMs.
Authors: Dazhou Yu, Genpei Zhang, Liang Zhao
Abstract: Ubiquitous geometric objects can be precisely and efficiently represented as polyhedra. The transformation of a polyhedron into a vector, known as polyhedra representation learning, is crucial for manipulating these shapes with mathematical and statistical tools for tasks like classification, clustering, and generation. Recent years have witnessed significant strides in this domain, yet most efforts focus on the vertex sequence of a polyhedron, neglecting the complex surface modeling crucial in real-world polyhedral objects. This study proposes \textbf{PolyhedronNet}, a general framework tailored for learning representations of 3D polyhedral objects. We propose the concept of the surface-attributed graph to seamlessly model the vertices, edges, faces, and their geometric interrelationships within a polyhedron. To effectively learn the representation of the entire surface-attributed graph, we first propose to break it down into local rigid representations to effectively learn each local region's relative positions against the remaining regions without geometric information loss. Subsequently, we propose PolyhedronGNN to hierarchically aggregate the local rigid representation via intra-face and inter-face geometric message passing modules, to obtain a global representation that minimizes information loss while maintaining rotation and translation invariance. Our experimental evaluations on four distinct datasets, encompassing both classification and retrieval tasks, substantiate PolyhedronNet's efficacy in capturing comprehensive and informative representations of 3D polyhedral objects. Code and data are available at {https://github.com/dyu62/3D_polyhedron}.
Authors: Kavitha Viswanathan, Shashwat Pathak, Piyush Bharambe, Harsh Choudhary, Amit Sethi
Abstract: Transformer-based video super-resolution (VSR) models have set new benchmarks in recent years, but their substantial computational demands make most of them unsuitable for deployment on resource-constrained devices. Achieving a balance between model complexity and output quality remains a formidable challenge in VSR. Although lightweight models have been introduced to address this issue, they often struggle to deliver state-of-the-art performance. We propose a novel lightweight, parameter-efficient deep residual deformable convolution network for VSR. Unlike prior methods, our model enhances feature utilization through residual connections and employs deformable convolution for precise frame alignment, addressing motion dynamics effectively. Furthermore, we introduce a single memory tensor to capture information accrued from the past frames and improve motion estimation across frames. This design enables an efficient balance between computational cost and reconstruction quality. With just 2.3 million parameters, our model achieves state-of-the-art SSIM of 0.9175 on the REDS4 dataset, surpassing existing lightweight and many heavy models in both accuracy and resource efficiency. Architectural insights from our model pave the way for real-time VSR on streaming data.
Authors: Yilin Wu, Ran Tian, Gokul Swamy, Andrea Bajcsy
Abstract: While generative robot policies have demonstrated significant potential in learning complex, multimodal behaviors from demonstrations, they still exhibit diverse failures at deployment-time. Policy steering offers an elegant solution to reducing the chance of failure by using an external verifier to select from low-level actions proposed by an imperfect generative policy. Here, one might hope to use a Vision Language Model (VLM) as a verifier, leveraging its open-world reasoning capabilities. However, off-the-shelf VLMs struggle to understand the consequences of low-level robot actions as they are represented fundamentally differently than the text and images the VLM was trained on. In response, we propose FOREWARN, a novel framework to unlock the potential of VLMs as open-vocabulary verifiers for runtime policy steering. Our key idea is to decouple the VLM's burden of predicting action outcomes (foresight) from evaluation (forethought). For foresight, we leverage a latent world model to imagine future latent states given diverse low-level action plans. For forethought, we align the VLM with these predicted latent states to reason about the consequences of actions in its native representation--natural language--and effectively filter proposed plans. We validate our framework across diverse robotic manipulation tasks, demonstrating its ability to bridge representational gaps and provide robust, generalizable policy steering.
Authors: Marco Paul E. Apolinario, Kaushik Roy, Charlotte Frenkel
Abstract: The demand for low-power inference and training of deep neural networks (DNNs) on edge devices has intensified the need for algorithms that are both scalable and energy-efficient. While spiking neural networks (SNNs) allow for efficient inference by processing complex spatio-temporal dynamics in an event-driven fashion, training them on resource-constrained devices remains challenging due to the high computational and memory demands of conventional error backpropagation (BP)-based approaches. In this work, we draw inspiration from biological mechanisms such as eligibility traces, spike-timing-dependent plasticity, and neural activity synchronization to introduce TESS, a temporally and spatially local learning rule for training SNNs. Our approach addresses both temporal and spatial credit assignments by relying solely on locally available signals within each neuron, thereby allowing computational and memory overheads to scale linearly with the number of neurons, independently of the number of time steps. Despite relying on local mechanisms, we demonstrate performance comparable to the backpropagation through time (BPTT) algorithm, within $\sim1.4$ accuracy points on challenging computer vision scenarios relevant at the edge, such as the IBM DVS Gesture dataset, CIFAR10-DVS, and temporal versions of CIFAR10, and CIFAR100. Being able to produce comparable performance to BPTT while keeping low time and memory complexity, TESS enables efficient and scalable on-device learning at the edge.
Authors: Mohammed Kharma, Soohyeon Choi, Mohammed AlKhanafseh, David Mohaisen
Abstract: Artificial Intelligence (AI)-driven code generation tools are increasingly used throughout the software development lifecycle to accelerate coding tasks. However, the security of AI-generated code using Large Language Models (LLMs) remains underexplored, with studies revealing various risks and weaknesses. This paper analyzes the security of code generated by LLMs across different programming languages. We introduce a dataset of 200 tasks grouped into six categories to evaluate the performance of LLMs in generating secure and maintainable code. Our research shows that while LLMs can automate code creation, their security effectiveness varies by language. Many models fail to utilize modern security features in recent compiler and toolkit updates, such as Java 17. Moreover, outdated methods are still commonly used, particularly in C++. This highlights the need for advancing LLMs to enhance security and quality while incorporating emerging best practices in programming languages.
Authors: Reza Sadeghian, Niloofar Hooshyaripour, Chris Joslin, WonSook Lee
Abstract: Accurate and robust 3D object detection is essential for autonomous driving, where fusing data from sensors like LiDAR and camera enhances detection accuracy. However, sensor malfunctions such as corruption or disconnection can degrade performance, and existing fusion models often struggle to maintain reliability when one modality fails. To address this, we propose ReliFusion, a novel LiDAR-camera fusion framework operating in the bird's-eye view (BEV) space. ReliFusion integrates three key components: the Spatio-Temporal Feature Aggregation (STFA) module, which captures dependencies across frames to stabilize predictions over time; the Reliability module, which assigns confidence scores to quantify the dependability of each modality under challenging conditions; and the Confidence-Weighted Mutual Cross-Attention (CW-MCA) module, which dynamically balances information from LiDAR and camera modalities based on these confidence scores. Experiments on the nuScenes dataset show that ReliFusion significantly outperforms state-of-the-art methods, achieving superior robustness and accuracy in scenarios with limited LiDAR fields of view and severe sensor malfunctions.
Authors: Zhimin Zhao
Abstract: Foundation models (FMs), particularly large language models (LLMs), have shown significant promise in various software engineering (SE) tasks, including code generation, debugging, and requirement refinement. Despite these advances, existing evaluation frameworks are insufficient for assessing model performance in iterative, context-rich workflows characteristic of SE activities. To address this limitation, we introduce SE Arena, an interactive platform designed to evaluate SE-focused chatbots. SE Arena provides a transparent, open-source leaderboard, supports multi-round conversational workflows, and enables end-to-end model comparisons. Moreover, SE Arena incorporates a new feature called RepoChat, which automatically injects repository-related context (e.g., issues, commits, pull requests) into the conversation, further aligning evaluations with real-world development processes. This paper outlines the design and capabilities of SE Arena, emphasizing its potential to advance the evaluation and practical application of FMs in software engineering.
Authors: Peter Chen, Bryan Chang, Olivia Annette Creasey, Julie Beth Sneddon, Yining Liu
Abstract: 3D cellular image segmentation methods are commonly divided into non-2D-based and 2D-based approaches, the latter reconstructing 3D shapes from the segmentation results of 2D layers. However, errors in 2D results often propagate, leading to oversegmentations in the final 3D results. To tackle this issue, we introduce an interpretable geometric framework that addresses the oversegmentations by correcting the 2D segmentation results based on geometric information from adjacent layers. Leveraging both geometric (layer-to-layer, 2D) and topological (3D shape) features, we use binary classification to determine whether neighboring cells should be stitched. We develop a pre-trained classifier on public plant cell datasets and validate its performance on animal cell datasets, confirming its effectiveness in correcting oversegmentations under the transfer learning setting. Furthermore, we demonstrate that our framework can be extended to correcting oversegmentation on non-2D-based methods. A clear pipeline is provided for end-users to build the pre-trained model to any labeled dataset.
Authors: Goodarz Mehr, Azim Eskandarian
Abstract: Bird's-eye view (BEV) perception for autonomous driving has garnered significant attention in recent years, in part because BEV representation facilitates the fusion of multi-sensor data. This enables a variety of perception tasks including BEV segmentation, a concise view of the environment that can be used to plan a vehicle's trajectory. However, this representation is not fully supported by existing datasets, and creation of new datasets can be a time-consuming endeavor. To address this problem, in this paper we introduce SimBEV, an extensively configurable and scalable randomized synthetic data generation tool that incorporates information from multiple sources to capture accurate BEV ground truth data, supports a comprehensive array of sensors, and enables a variety of perception tasks including BEV segmentation and 3D object detection. We use SimBEV to create the SimBEV dataset, a large collection of annotated perception data from diverse driving scenarios.
Authors: Andrew Van Horn, Lauryn Smith, Mahamad Mahmoud, Michael McMaster, Clara Pinchbeck, Ina Martin, Andrew Lininger, Anthony Ingrisano, Adam Lowe, Carlos Bayod, Elizabeth Bolman, Kenneth Singer, Michael Hinczewski
Abstract: The history of art has seen significant shifts in the manner in which artworks are created, making understanding of creative processes a central question in technical art history. In the Renaissance and Early Modern period, paintings were largely produced by master painters directing workshops of apprentices who often contributed to projects. The masters varied significantly in artistic and managerial styles, meaning different combinations of artists and implements might be seen both between masters and within workshops or even individual canvases. Information on how different workshops were managed and the processes by which artworks were created remains elusive. Machine learning methods have potential to unearth new information about artists' creative processes by extending the analysis of brushwork to a microscopic scale. Analysis of workshop paintings, however, presents a challenge in that documentation of the artists and materials involved is sparse, meaning external examples are not available to train networks to recognize their contributions. Here we present a novel machine learning approach we call pairwise assignment training for classifying heterogeneity (PATCH) that is capable of identifying individual artistic practice regimes with no external training data, or "ground truth." The method achieves unsupervised results by supervised means, and outperforms both simple statistical procedures and unsupervised machine learning methods. We apply this method to two historical paintings by the Spanish Renaissance master, El Greco: The Baptism of Christ and Christ on the Cross with Landscape, and our findings regarding the former potentially challenge previous work that has assigned the painting to workshop members. Further, the results of our analyses create a measure of heterogeneity of artistic practice that can be used to characterize artworks across time and space.
Authors: Shu-yuan Wang, Hikaru Sasaki, Takamitsu Matsubara
Abstract: Learning control policies for real-world robotic tasks often involve challenges such as multimodality, local discontinuities, and the need for computational efficiency. These challenges arise from the complexity of robotic environments, where multiple solutions may coexist. To address these issues, we propose Composite Gaussian Processes Flows (CGP-Flows), a novel semi-parametric model for robotic policy. CGP-Flows integrate Overlapping Mixtures of Gaussian Processes (OMGPs) with the Continuous Normalizing Flows (CNFs), enabling them to model complex policies addressing multimodality and local discontinuities. This hybrid approach retains the computational efficiency of OMGPs while incorporating the flexibility of CNFs. Experiments conducted in both simulated and real-world robotic tasks demonstrate that CGP-flows significantly improve performance in modeling control policies. In a simulation task, we confirmed that CGP-Flows had a higher success rate compared to the baseline method, and the success rate of GCP-Flow was significantly different from the success rate of other baselines in chi-square tests.
Authors: Tim-Lukas Habich, Aran Mohammad, Simon F. G. Ehlers, Martin Bensch, Thomas Seel, Moritz Schappler
Abstract: Soft robots can revolutionize several applications with high demands on dexterity and safety. When operating these systems, real-time estimation and control require fast and accurate models. However, prediction with first-principles (FP) models is slow, and learned black-box models have poor generalizability. Physics-informed machine learning offers excellent advantages here, but it is currently limited to simple, often simulated systems without considering changes after training. We propose physics-informed neural networks (PINNs) for articulated soft robots (ASRs) with a focus on data efficiency. The amount of expensive real-world training data is reduced to a minimum - one dataset in one system domain. Two hours of data in different domains are used for a comparison against two gold-standard approaches: In contrast to a recurrent neural network, the PINN provides a high generalizability. The prediction speed of an accurate FP model is improved with the PINN by up to a factor of 466 at slightly reduced accuracy. This enables nonlinear model predictive control (MPC) of the pneumatic ASR. In nine dynamic MPC experiments, an average joint-tracking error of 1.3{\deg} is achieved.
Authors: Zachary Cooper-Baldock, Stephen Turnock, Karl Sammut
Abstract: Autonomous Underwater Vehicles (AUVs) encounter significant energy, control and navigation challenges in complex underwater environments, particularly during close-proximity operations, such as launch and recovery (LAR), where fluid interactions and wake effects present additional navigational and energy challenges. Traditional path planning methods fail to incorporate these detailed wake structures, resulting in increased energy consumption, reduced control stability, and heightened safety risks. This paper presents a novel wake-informed, 3D path planning approach that fully integrates localized wake effects and global currents into the planning algorithm. Two variants of the A* algorithm - a current-informed planner and a wake-informed planner - are created to assess its validity and two neural network models are then trained to approximate these planners for real-time applications. Both the A* planners and NN models are evaluated using important metrics such as energy expenditure, path length, and encounters with high-velocity and turbulent regions. The results demonstrate a wake-informed A* planner consistently achieves the lowest energy expenditure and minimizes encounters with high-velocity regions, reducing energy consumption by up to 11.3%. The neural network models are observed to offer computational speedup of 6 orders of magnitude, but exhibit 4.51 - 19.79% higher energy expenditures and 9.81 - 24.38% less optimal paths. These findings underscore the importance of incorporating detailed wake structures into traditional path planning algorithms and the benefits of neural network approximations to enhance energy efficiency and operational safety for AUVs in complex 3D domains.
Authors: Lancelot F. James, Juho Lee, Abhinav Pandey
Abstract: In this work, we present a comprehensive Bayesian posterior analysis of what we term Poisson Hierarchical Indian Buffet Processes, designed for complex random sparse count species sampling models that allow for the sharing of information across and within groups. This analysis covers a potentially infinite number of species and unknown parameters, which, within a Bayesian machine learning context, we are able to learn from as more information is sampled. To achieve our refined results, we employ a range of methodologies drawn from Bayesian latent feature models, random occupancy models, and excursion theory. Despite this complexity, our goal is to make our findings accessible to practitioners, including those who may not be familiar with these areas. To facilitate understanding, we adopt a pseudo-expository style that emphasizes clarity and practical utility. We aim to express our findings in a language that resonates with experts in microbiome and ecological studies, addressing gaps in modeling capabilities while acknowledging that we are not experts ourselves in these fields. This approach encourages the use of our models as basic components of more sophisticated frameworks employed by domain experts, embodying the spirit of the seminal work on the Dirichlet Process. Ultimately, our refined posterior analysis not only yields tractable computational procedures but also enables practical statistical implementation and provides a clear mapping to relevant quantities in microbiome analysis.
Authors: Avery Ma, Yangchen Pan, Amir-massoud Farahmand
Abstract: Many-shot jailbreaking circumvents the safety alignment of large language models by exploiting their ability to process long input sequences. To achieve this, the malicious target prompt is prefixed with hundreds of fabricated conversational turns between the user and the model. These fabricated exchanges are randomly sampled from a pool of malicious questions and responses, making it appear as though the model has already complied with harmful instructions. In this paper, we present PANDAS: a hybrid technique that improves many-shot jailbreaking by modifying these fabricated dialogues with positive affirmations, negative demonstrations, and an optimized adaptive sampling method tailored to the target prompt's topic. Extensive experiments on AdvBench and HarmBench, using state-of-the-art LLMs, demonstrate that PANDAS significantly outperforms baseline methods in long-context scenarios. Through an attention analysis, we provide insights on how long-context vulnerabilities are exploited and show how PANDAS further improves upon many-shot jailbreaking.
Authors: Zelai Xu, Chao Yu, Ruize Zhang, Huining Yuan, Xiangmin Yi, Shilong Ji, Chuqi Wang, Wenhao Tang, Yu Wang
Abstract: Multi-agent reinforcement learning (MARL) has made significant progress, largely fueled by the development of specialized testbeds that enable systematic evaluation of algorithms in controlled yet challenging scenarios. However, existing testbeds often focus on purely virtual simulations or limited robot morphologies such as robotic arms, quadrupeds, and humanoids, leaving high-mobility platforms with real-world physical constraints like drones underexplored. To bridge this gap, we present VolleyBots, a new MARL testbed where multiple drones cooperate and compete in the sport of volleyball under physical dynamics. VolleyBots features a turn-based interaction model under volleyball rules, a hierarchical decision-making process that combines motion control and strategic play, and a high-fidelity simulation for seamless sim-to-real transfer. We provide a comprehensive suite of tasks ranging from single-drone drills to multi-drone cooperative and competitive tasks, accompanied by baseline evaluations of representative MARL and game-theoretic algorithms. Results in simulation show that while existing algorithms handle simple tasks effectively, they encounter difficulty in complex tasks that require both low-level control and high-level strategy. We further demonstrate zero-shot deployment of a simulation-learned policy to real-world drones, highlighting VolleyBots' potential to propel MARL research involving agile robotic platforms. The project page is at https://sites.google.com/view/volleybots/home.
Authors: Kiana Asgari, Andrea Montanari, Basil Saeed
Abstract: We consider a general model for high-dimensional empirical risk minimization whereby the data $\mathbf{x}_i$ are $d$-dimensional isotropic Gaussian vectors, the model is parametrized by $\mathbf{\Theta}\in\mathbb{R}^{d\times k}$, and the loss depends on the data via the projection $\mathbf{\Theta}^\mathsf{T}\mathbf{x}_i$. This setting covers as special cases classical statistics methods (e.g. multinomial regression and other generalized linear models), but also two-layer fully connected neural networks with $k$ hidden neurons. We use the Kac-Rice formula from Gaussian process theory to derive a bound on the expected number of local minima of this empirical risk, under the proportional asymptotics in which $n,d\to\infty$, with $n\asymp d$. Via Markov's inequality, this bound allows to determine the positions of these minimizers (with exponential deviation bounds) and hence derive sharp asymptotics on the estimation and prediction error. In this paper, we apply our characterization to convex losses, where high-dimensional asymptotics were not (in general) rigorously established for $k\ge 2$. We show that our approach is tight and allows to prove previously conjectured results. In addition, we characterize the spectrum of the Hessian at the minimizer. A companion paper applies our general result to non-convex examples.
Authors: Shashank Sharma, Janina Hoffmann, Vinay Namboodiri
Abstract: In this paper, we address the challenge of long-horizon visual planning tasks using Hierarchical Reinforcement Learning (HRL). Our key contribution is a Discrete Hierarchical Planning (DHP) method, an alternative to traditional distance-based approaches. We provide theoretical foundations for the method and demonstrate its effectiveness through extensive empirical evaluations. Our agent recursively predicts subgoals in the context of a long-term goal and receives discrete rewards for constructing plans as compositions of abstract actions. The method introduces a novel advantage estimation strategy for tree trajectories, which inherently encourages shorter plans and enables generalization beyond the maximum tree depth. The learned policy function allows the agent to plan efficiently, requiring only $\log N$ computational steps, making re-planning highly efficient. The agent, based on a soft-actor critic (SAC) framework, is trained using on-policy imagination data. Additionally, we propose a novel exploration strategy that enables the agent to generate relevant training examples for the planning modules. We evaluate our method on long-horizon visual planning tasks in a 25-room environment, where it significantly outperforms previous benchmarks at success rate and average episode length. Furthermore, an ablation study highlights the individual contributions of key modules to the overall performance.
Authors: Chengwei Xia, Chaoxi Niu, Kun Zhan
Abstract: Multiview feature learning aims to learn discriminative features by integrating the distinct information in each view. However, most existing methods still face significant challenges in learning view-consistency features, which are crucial for effective multiview learning. Motivated by the theories of CCA and contrastive learning in multiview feature learning, we propose the hierarchical consensus network (HCN) in this paper. The HCN derives three consensus indices for capturing the hierarchical consensus across views, which are classifying consensus, coding consensus, and global consensus, respectively. Specifically, classifying consensus reinforces class-level correspondence between views from a CCA perspective, while coding consensus closely resembles contrastive learning and reflects contrastive comparison of individual instances. Global consensus aims to extract consensus information from two perspectives simultaneously. By enforcing the hierarchical consensus, the information within each view is better integrated to obtain more comprehensive and discriminative features. The extensive experimental results obtained on four multiview datasets demonstrate that the proposed method significantly outperforms several state-of-the-art methods.
Authors: Haolin Wang, Yafei Ou, Prasoon Ambalathankandy, Gen Ota, Pengyu Dai, Masayuki Ikebe, Kenji Suzuki, Tamotsu Kamishima
Abstract: Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by joint inflammation and progressive structural damage. Joint space width (JSW) is a critical indicator in conventional radiography for evaluating disease progression, which has become a prominent research topic in computer-aided diagnostic (CAD) systems. However, deep learning-based radiological CAD systems for JSW analysis face significant challenges in data quality, including data imbalance, limited variety, and annotation difficulties. This work introduced a challenging image synthesis scenario and proposed Layer Separation Networks (LSN) to accurately separate the soft tissue layer, the upper bone layer, and the lower bone layer in conventional radiographs of finger joints. Using these layers, the adjustable JSW images can be synthesized to address data quality challenges and achieve ground truth (GT) generation. Experimental results demonstrated that LSN-based synthetic images closely resemble real radiographs, and significantly enhanced the performance in downstream tasks. The code and dataset will be available.
Authors: Wenhao Zheng, Yixiao Chen, Weitong Zhang, Souvik Kundu, Yun Li, Zhengzhong Liu, Eric P. Xing, Hongyi Wang, Huaxiu Yao
Abstract: Large language models have achieved remarkable success in various tasks but suffer from high computational costs during inference, limiting their deployment in resource-constrained applications. To address this issue, we propose a novel CITER (\textbf{C}ollaborative \textbf{I}nference with \textbf{T}oken-l\textbf{E}vel \textbf{R}outing) framework that enables efficient collaboration between small and large language models (SLMs & LLMs) through a token-level routing strategy. Specifically, CITER routes non-critical tokens to an SLM for efficiency and routes critical tokens to an LLM for generalization quality. We formulate router training as a policy optimization, where the router receives rewards based on both the quality of predictions and the inference costs of generation. This allows the router to learn to predict token-level routing scores and make routing decisions based on both the current token and the future impact of its decisions. To further accelerate the reward evaluation process, we introduce a shortcut which significantly reduces the costs of the reward estimation and improving the practicality of our approach. Extensive experiments on five benchmark datasets demonstrate that CITER reduces the inference costs while preserving high-quality generation, offering a promising solution for real-time and resource-constrained applications.
Authors: Prathamesh Pradeep Khole, Zahra Kais Petiwala, Shri Prathaa Magesh, Ehsan Mirafzali, Utkarsh Gupta, Jing-Rebecca Li, Andrada Ianus, Razvan Marinescu
Abstract: We propose ReMiDi, a novel method for inferring neuronal microstructure as arbitrary 3D meshes using a differentiable diffusion Magnetic Resonance Imaging (dMRI) simulator. We first implemented in PyTorch a differentiable dMRI simulator that simulates the forward diffusion process using a finite-element method on an input 3D microstructure mesh. To achieve significantly faster simulations, we solve the differential equation semi-analytically using a matrix formalism approach. Given a reference dMRI signal $S_{ref}$, we use the differentiable simulator to iteratively update the input mesh such that it matches $S_{ref}$ using gradient-based learning. Since directly optimizing the 3D coordinates of the vertices is challenging, particularly due to ill-posedness of the inverse problem, we instead optimize a lower-dimensional latent space representation of the mesh. The mesh is first encoded into spectral coefficients, which are further encoded into a latent $\textbf{z}$ using an auto-encoder, and are then decoded back into the true mesh. We present an end-to-end differentiable pipeline that simulates signals that can be tuned to match a reference signal by iteratively updating the latent representation $\textbf{z}$. We demonstrate the ability to reconstruct microstructures of arbitrary shapes represented by finite-element meshes, with a focus on axonal geometries found in the brain white matter, including bending, fanning and beading fibers. Our source code will be made available online.
Authors: Arnav Grover
Abstract: In response to Task II of the FinRL Challenge at ACM ICAIF 2024, this study proposes a novel prompt framework for fine-tuning large language models (LLM) with Reinforcement Learning from Market Feedback (RLMF). Our framework incorporates market-specific features and short-term price dynamics to generate more precise trading signals. Traditional LLMs, while competent in sentiment analysis, lack contextual alignment for financial market applications. To bridge this gap, we fine-tune the LLaMA-3.2-3B-Instruct model using a custom RLMF prompt design that integrates historical market data and reward-based feedback. Our evaluation shows that this RLMF-tuned framework outperforms baseline methods in signal consistency and achieving tighter trading outcomes; awarded as winner of Task II. You can find the code for this project on GitHub.
Authors: Masanari Kimura, Howard Bondell
Abstract: Fr\'echet regression extends classical regression methods to non-Euclidean metric spaces, enabling the analysis of data relationships on complex structures such as manifolds and graphs. This work establishes a rigorous theoretical analysis for Fr\'echet regression through the lens of comparison geometry which leads to important considerations for its use in practice. The analysis provides key results on the existence, uniqueness, and stability of the Fr\'echet mean, along with statistical guarantees for nonparametric regression, including exponential concentration bounds and convergence rates. Additionally, insights into angle stability reveal the interplay between curvature of the manifold and the behavior of the regression estimator in these non-Euclidean contexts. Empirical experiments validate the theoretical findings, demonstrating the effectiveness of proposed hyperbolic mappings, particularly for data with heteroscedasticity, and highlighting the practical usefulness of these results.
Authors: Kaja Gruntkowska, Hanmin Li, Aadi Rane, Peter Richt\'arik
Abstract: Non-smooth and non-convex global optimization poses significant challenges across various applications, where standard gradient-based methods often struggle. We propose the Ball-Proximal Point Method, Broximal Point Method, or Ball Point Method (BPM) for short - a novel algorithmic framework inspired by the classical Proximal Point Method (PPM) (Rockafellar, 1976), which, as we show, sheds new light on several foundational optimization paradigms and phenomena, including non-convex and non-smooth optimization, acceleration, smoothing, adaptive stepsize selection, and trust-region methods. At the core of BPM lies the ball-proximal ("broximal") operator, which arises from the classical proximal operator by replacing the quadratic distance penalty by a ball constraint. Surprisingly, and in sharp contrast with the sublinear rate of PPM in the nonsmooth convex regime, we prove that BPM converges linearly and in a finite number of steps in the same regime. Furthermore, by introducing the concept of ball-convexity, we prove that BPM retains the same global convergence guarantees under weaker assumptions, making it a powerful tool for a broader class of potentially non-convex optimization problems. Just like PPM plays the role of a conceptual method inspiring the development of practically efficient algorithms and algorithmic elements, e.g., gradient descent, adaptive step sizes, acceleration (Ahn & Sra, 2020), and "W" in AdamW (Zhuang et al., 2022), we believe that BPM should be understood in the same manner: as a blueprint and inspiration for further development.
Authors: Pengxiao Lin, Zhongwang Zhang, Zhi-Qin John Xu
Abstract: Since the inception of Large Language Models (LLMs), the quest to efficiently train them for superior reasoning capabilities has been a pivotal challenge. The dominant training paradigm for LLMs is based on next token prediction (NTP). Alternative methodologies, called Critical Token Prediction (CTP), focused exclusively on specific critical tokens (such as the answer in Q\&A dataset), aiming to reduce the overfitting of extraneous information and noise. Contrary to initial assumptions, our research reveals that despite NTP's exposure to noise during training, it surpasses CTP in reasoning ability. We attribute this counterintuitive outcome to the regularizing influence of noise on the training dynamics. Our empirical analysis shows that NTP-trained models exhibit enhanced generalization and robustness across various benchmark reasoning datasets, demonstrating greater resilience to perturbations and achieving flatter loss minima. These findings illuminate that NTP is instrumental in fostering reasoning abilities during pretraining, whereas CTP is more effective for finetuning, thereby enriching our comprehension of optimal training strategies in LLM development.
Authors: Ziyang Ye, Triet Huynh Minh Le, M. Ali Babar
Abstract: Security misconfigurations in Container Orchestrators (COs) can pose serious threats to software systems. While Static Analysis Tools (SATs) can effectively detect these security vulnerabilities, the industry currently lacks automated solutions capable of fixing these misconfigurations. The emergence of Large Language Models (LLMs), with their proven capabilities in code understanding and generation, presents an opportunity to address this limitation. This study introduces LLMSecConfig, an innovative framework that bridges this gap by combining SATs with LLMs. Our approach leverages advanced prompting techniques and Retrieval-Augmented Generation (RAG) to automatically repair security misconfigurations while preserving operational functionality. Evaluation of 1,000 real-world Kubernetes configurations achieved a 94\% success rate while maintaining a low rate of introducing new misconfigurations. Our work makes a promising step towards automated container security management, reducing the manual effort required for configuration maintenance.
Authors: Shuo Wang, Bokui Wang, Zhixiang Shen, Boyan Deng, Zhao Kang
Abstract: Recent advances in CV and NLP have inspired researchers to develop general-purpose graph foundation models through pre-training across diverse domains. However, a fundamental challenge arises from the substantial differences in graph topologies across domains. Additionally, real-world graphs are often sparse and prone to noisy connections and adversarial attacks. To address these issues, we propose the Multi-Domain Graph Foundation Model (MDGFM), a unified framework that aligns and leverages cross-domain topological information to facilitate robust knowledge transfer. MDGFM bridges different domains by adaptively balancing features and topology while refining original graphs to eliminate noise and align topological structures. To further enhance knowledge transfer, we introduce an efficient prompt-tuning approach. By aligning topologies, MDGFM not only improves multi-domain pre-training but also enables robust knowledge transfer to unseen domains. Theoretical analyses provide guarantees of MDGFM's effectiveness and domain generalization capabilities. Extensive experiments on both homophilic and heterophilic graph datasets validate the robustness and efficacy of our method.
Authors: Yaodong Yang, Guangyong Chen, Hongyao Tang, Furui Liu, Danruo Deng, Pheng Ann Heng
Abstract: Overestimation in single-agent reinforcement learning has been extensively studied. In contrast, overestimation in the multiagent setting has received comparatively little attention although it increases with the number of agents and leads to severe learning instability. Previous works concentrate on reducing overestimation in the estimation process of target Q-value. They ignore the follow-up optimization process of online Q-network, thus making it hard to fully address the complex multiagent overestimation problem. To solve this challenge, in this study, we first establish an iterative estimation-optimization analysis framework for multiagent value-mixing Q-learning. Our analysis reveals that multiagent overestimation not only comes from the computation of target Q-value but also accumulates in the online Q-network's optimization. Motivated by it, we propose the Dual Ensembled Multiagent Q-Learning with Hypernet Regularizer algorithm to tackle multiagent overestimation from two aspects. First, we extend the random ensemble technique into the estimation of target individual and global Q-values to derive a lower update target. Second, we propose a novel hypernet regularizer on hypernetwork weights and biases to constrain the optimization of online global Q-network to prevent overestimation accumulation. Extensive experiments in MPE and SMAC show that the proposed method successfully addresses overestimation across various tasks.
Authors: Mokshagna Sai Teja Karanam, Krithika Iyer, Sarang Joshi, Shireen Elhabian
Abstract: Spatial transformations that capture population-level morphological statistics are critical for medical image analysis. Commonly used smoothness regularizers for image registration fail to integrate population statistics, leading to anatomically inconsistent transformations. Inverse consistency regularizers promote geometric consistency but lack population morphometrics integration. Regularizers that constrain deformation to low-dimensional manifold methods address this. However, they prioritize reconstruction over interpretability and neglect diffeomorphic properties, such as group composition and inverse consistency. We introduce MORPH-LER, a Log-Euclidean regularization framework for population-aware unsupervised image registration. MORPH-LER learns population morphometrics from spatial transformations to guide and regularize registration networks, ensuring anatomically plausible deformations. It features a bottleneck autoencoder that computes the principal logarithm of deformation fields via iterative square-root predictions. It creates a linearized latent space that respects diffeomorphic properties and enforces inverse consistency. By integrating a registration network with a diffeomorphic autoencoder, MORPH-LER produces smooth, meaningful deformation fields. The framework offers two main contributions: (1) a data-driven regularization strategy that incorporates population-level anatomical statistics to enhance transformation validity and (2) a linearized latent space that enables compact and interpretable deformation fields for efficient population morphometrics analysis. We validate MORPH-LER across two families of deep learning-based registration networks, demonstrating its ability to produce anatomically accurate, computationally efficient, and statistically meaningful transformations on the OASIS-1 brain imaging dataset.
Authors: Nikhil Bhendawade, Mahyar Najibi, Devang Naik, Irina Belousova
Abstract: Residual transformations enhance the representational depth and expressive power of large language models (LLMs). However, applying static residual transformations across all tokens in auto-regressive generation leads to a suboptimal trade-off between inference efficiency and generation fidelity. Existing methods, including Early Exiting, Skip Decoding, and Mixture-of-Depth address this by modulating the residual transformation based on token-level complexity. Nevertheless, these approaches predominantly consider the distance traversed by tokens through the model layers, neglecting the underlying velocity of residual evolution. We introduce Mixture of Multi-rate Residuals (M2R2), a framework that dynamically modulates residual velocity to improve early alignment, enhancing inference efficiency. Evaluations on reasoning oriented tasks such as Koala, Self-Instruct, WizardLM, and MT-Bench show M2R2 surpasses state-of-the-art distance-based strategies, balancing generation quality and speedup. In self-speculative decoding setup, M2R2 achieves up to 2.8x speedups on MT-Bench, outperforming methods like 2-model speculative decoding, Medusa, LookAhead Decoding, and DEED. In Mixture-of-Experts (MoE) architectures, integrating early residual alignment with ahead-of-time expert loading into high-bandwidth memory (HBM) accelerates decoding, reduces expert-switching bottlenecks, and achieves a 2.9x speedup, making it highly effective in resource-constrained environments.
Authors: Minwoo Kim, Geunsik Bae, Jinwoo Lee, Woojae Shin, Changseung Kim, Myong-Yol Choi, Heejung Shin, Hyondong Oh
Abstract: This paper introduces a learning-based visual planner for agile drone flight in cluttered environments. The proposed planner generates collision-free waypoints in milliseconds, enabling drones to perform agile maneuvers in complex environments without building separate perception, mapping, and planning modules. Learning-based methods, such as behavior cloning (BC) and reinforcement learning (RL), demonstrate promising performance in visual navigation but still face inherent limitations. BC is susceptible to compounding errors due to limited expert imitation, while RL struggles with reward function design and sample inefficiency. To address these limitations, this paper proposes an inverse reinforcement learning (IRL)-based framework for high-speed visual navigation. By leveraging IRL, it is possible to reduce the number of interactions with simulation environments and improve capability to deal with high-dimensional spaces while preserving the robustness of RL policies. A motion primitive-based path planning algorithm collects an expert dataset with privileged map data from diverse environments, ensuring comprehensive scenario coverage. By leveraging both the acquired expert and learner dataset gathered from the agent's interactions with the simulation environments, a robust reward function and policy are learned across diverse states. While the proposed method is trained in a simulation environment only, it can be directly applied to real-world scenarios without additional training or tuning. The performance of the proposed method is validated in both simulation and real-world environments, including forests and various structures. The trained policy achieves an average speed of 7 m/s and a maximum speed of 8.8 m/s in real flight experiments. To the best of our knowledge, this is the first work to successfully apply an IRL framework for high-speed visual navigation of drones.
Authors: Raghav Arora, Shivam Singh, Karthik Swaminathan, Ahana Datta, Snehasis Banerjee, Brojeshwar Bhowmick, Krishna Murthy Jatavallabhula, Mohan Sridharan, Madhava Krishna
Abstract: Assistive agents performing household tasks such as making the bed or cooking breakfast often compute and execute actions that accomplish one task at a time. However, efficiency can be improved by anticipating upcoming tasks and computing an action sequence that jointly achieves these tasks. State-of-the-art methods for task anticipation use data-driven deep networks and Large Language Models (LLMs), but they do so at the level of high-level tasks and/or require many training examples. Our framework leverages the generic knowledge of LLMs through a small number of prompts to perform high-level task anticipation, using the anticipated tasks as goals in a classical planning system to compute a sequence of finer-granularity actions that jointly achieve these goals. We ground and evaluate our framework's abilities in realistic scenarios in the VirtualHome environment and demonstrate a 31% reduction in execution time compared with a system that does not consider upcoming tasks.
Authors: Shivam Singh, Karthik Swaminathan, Nabanita Dash, Ramandeep Singh, Snehasis Banerjee, Mohan Sridharan, Madhava Krishna
Abstract: Embodied agents assisting humans are often asked to complete a new task in a new scenario. An agent preparing a particular dish in the kitchen based on a known recipe may be asked to prepare a new dish or to perform cleaning tasks in the storeroom. There may not be sufficient resources, e.g., time or labeled examples, to train the agent for these new situations. Large Language Models (LLMs) trained on considerable knowledge across many domains are able to predict a sequence of abstract actions for such new tasks and scenarios, although it may not be possible for the agent to execute this action sequence due to task-, agent-, or domain-specific constraints. Our framework addresses these challenges by leveraging the generic predictions provided by LLM and the prior domain-specific knowledge encoded in a Knowledge Graph (KG), enabling an agent to quickly adapt to new tasks and scenarios. The robot also solicits and uses human input as needed to refine its existing knowledge. Based on experimental evaluation over cooking and cleaning tasks in simulation domains, we demonstrate that the interplay between LLM, KG, and human input leads to substantial performance gains compared with just using the LLM output.
Authors: Ruisi Zhang, Neusha Javidnia, Nojan Sheybani, Farinaz Koushanfar
Abstract: This paper introduces RoSe, the first-of-its-kind ML/Crypto codesign watermarking framework that regulates LLM-generated code to avoid intellectual property rights violations and inappropriate misuse in software development. High-quality watermarks adhering to the detectability-fidelity-robustness tri-objective are limited due to codes' low-entropy nature. Watermark verification, however, often needs to reveal the signature and requires re-encoding new ones for code reuse, which potentially compromising the system's usability. To overcome these challenges, RoSe obtains high-quality watermarks by training the watermark insertion and extraction modules end-to-end to ensure (i) unaltered watermarked code functionality and (ii) enhanced detectability and robustness leveraging pre-trained CodeT5 as the insertion backbone to enlarge the code syntactic and variable rename transformation search space. In the deployment, RoSe uses zero-knowledge proofs for secure verification without revealing the underlying signatures. Extensive evaluations demonstrated RoSe achieves high detection accuracy while preserving the code functionality. RoSe is also robust against attacks and provides efficient secure watermark verification.
Authors: Poojan Shah, Shashwat Agrawal, Ragesh Jaiswal
Abstract: The $k$-$\mathtt{means}$++ seeding algorithm (Arthur & Vassilvitskii, 2007) is widely used in practice for the $k$-means clustering problem where the goal is to cluster a dataset $\mathcal{X} \subset \mathbb{R} ^d$ into $k$ clusters. The popularity of this algorithm is due to its simplicity and provable guarantee of being $O(\log k)$ competitive with the optimal solution in expectation. However, its running time is $O(|\mathcal{X}|kd)$, making it expensive for large datasets. In this work, we present a simple and effective rejection sampling based approach for speeding up $k$-$\mathtt{means}$++. Our first method runs in time $\tilde{O}(\mathtt{nnz} (\mathcal{X}) + \beta k^2d)$ while still being $O(\log k )$ competitive in expectation. Here, $\beta$ is a parameter which is the ratio of the variance of the dataset to the optimal $k$-$\mathtt{means}$ cost in expectation and $\tilde{O}$ hides logarithmic factors in $k$ and $|\mathcal{X}|$. Our second method presents a new trade-off between computational cost and solution quality. It incurs an additional scale-invariant factor of $ k^{-\Omega( m/\beta)} \operatorname{Var} (\mathcal{X})$ in addition to the $O(\log k)$ guarantee of $k$-$\mathtt{means}$++ improving upon a result of (Bachem et al, 2016a) who get an additional factor of $m^{-1}\operatorname{Var}(\mathcal{X})$ while still running in time $\tilde{O}(\mathtt{nnz}(\mathcal{X}) + mk^2d)$. We perform extensive empirical evaluations to validate our theoretical results and to show the effectiveness of our approach on real datasets.
Authors: Sanket Shinde, Raviraj Joshi
Abstract: While topic modeling in English has become a prevalent and well-explored area, venturing into topic modeling for Indic languages remains relatively rare. The limited availability of resources, diverse linguistic structures, and unique challenges posed by Indic languages contribute to the scarcity of research and applications in this domain. Despite the growing interest in natural language processing and machine learning, there exists a noticeable gap in the comprehensive exploration of topic modeling methodologies tailored specifically for languages such as Hindi, Marathi, Tamil, and others. In this paper, we examine several topic modeling approaches applied to the Marathi language. Specifically, we compare various BERT and non-BERT approaches, including multilingual and monolingual BERT models, using topic coherence and topic diversity as evaluation metrics. Our analysis provides insights into the performance of these approaches for Marathi language topic modeling. The key finding of the paper is that BERTopic, when combined with BERT models trained on Indic languages, outperforms LDA in terms of topic modeling performance.
Authors: Rudolf Reiter, Jasper Hoffmann, Dirk Reinhardt, Florian Messerer, Katrin Baumg\"artner, Shamburaj Sawant, Joschka Boedecker, Moritz Diehl, Sebastien Gros
Abstract: The fields of MPC and RL consider two successful control techniques for Markov decision processes. Both approaches are derived from similar fundamental principles, and both are widely used in practical applications, including robotics, process control, energy systems, and autonomous driving. Despite their similarities, MPC and RL follow distinct paradigms that emerged from diverse communities and different requirements. Various technical discrepancies, particularly the role of an environment model as part of the algorithm, lead to methodologies with nearly complementary advantages. Due to their orthogonal benefits, research interest in combination methods has recently increased significantly, leading to a large and growing set of complex ideas leveraging MPC and RL. This work illuminates the differences, similarities, and fundamentals that allow for different combination algorithms and categorizes existing work accordingly. Particularly, we focus on the versatile actor-critic RL approach as a basis for our categorization and examine how the online optimization approach of MPC can be used to improve the overall closed-loop performance of a policy.
Authors: Youngsung Kim
Abstract: Neural networks, as currently designed, fall short of achieving true logical intelligence. Modern AI models rely on standard neural computation-inner-product-based transformations and nonlinear activations-to approximate patterns from data. While effective for inductive learning, this architecture lacks the structural guarantees necessary for deductive inference and logical consistency. As a result, deep networks struggle with rule-based reasoning, structured generalization, and interpretability without extensive post-hoc modifications. This position paper argues that standard neural layers must be fundamentally rethought to integrate logical reasoning. We advocate for Logical Neural Units (LNUs)-modular components that embed differentiable approximations of logical operations (e.g., AND, OR, NOT) directly within neural architectures. We critique existing neurosymbolic approaches, highlight the limitations of standard neural computation for logical inference, and present LNUs as a necessary paradigm shift in AI. Finally, we outline a roadmap for implementation, discussing theoretical foundations, architectural integration, and key challenges for future research.
Authors: Amaury Gouverneur, Borja Rodriguez G\'alvez, Tobias Oechtering, Mikael Skoglund
Abstract: This paper studies the Bayesian regret of the Thompson Sampling algorithm for bandit problems, building on the information-theoretic framework introduced by Russo and Van Roy (2015). Specifically, it extends the rate-distortion analysis of Dong and Van Roy (2018), which provides near-optimal bounds for linear bandits. A limitation of these results is the assumption of a finite action space. We address this by extending the analysis to settings with infinite and continuous action spaces. Additionally, we specialize our results to bandit problems with expected rewards that are Lipschitz continuous with respect to the action space, deriving a regret bound that explicitly accounts for the complexity of the action space.
Authors: Ruiqi Feng, Tailin Wu, Chenglei Yu, Wenhao Deng, Peiyan Hu
Abstract: Flow matching has shown state-of-the-art performance in various generative tasks, ranging from image generation to decision-making, where guided generation is pivotal. However, the guidance of flow matching is more general than and thus substantially different from that of its predecessor, diffusion models. Therefore, the challenge in guidance for general flow matching remains largely underexplored. In this paper, we propose the first framework of general guidance for flow matching. From this framework, we derive a family of guidance techniques that can be applied to general flow matching. These include a new training-free asymptotically exact guidance, novel training losses for training-based guidance, and two classes of approximate guidance that cover classical gradient guidance methods as special cases. We theoretically investigate these different methods to give a practical guideline for choosing suitable methods in different scenarios. Experiments on synthetic datasets, image inverse problems, and offline reinforcement learning demonstrate the effectiveness of our proposed guidance methods and verify the correctness of our flow matching guidance framework. Code to reproduce the experiments can be found at https://github.com/AI4Science-WestlakeU/flow_guidance.
URLs: https://github.com/AI4Science-WestlakeU/flow_guidance.
Authors: Thien Q. Tran, Akifumi Wachi, Rei Sato, Takumi Tanabe, Youhei Akimoto
Abstract: Safety alignment is an essential research topic for real-world AI applications. Despite the multifaceted nature of safety and trustworthiness in AI, current safety alignment methods often focus on a comprehensive notion of safety. By carefully assessing models from the existing safety-alignment methods, we found that, while they generally improved overall safety performance, they failed to ensure safety in specific categories. Our study first identified the difficulty of eliminating such vulnerabilities without sacrificing the model's helpfulness. We observed that, while smaller KL penalty parameters, increased training iterations, and dataset cleansing can enhance safety, they do not necessarily improve the trade-off between safety and helpfulness. We discovered that safety alignment could even induce undesired effects and result in a model that prefers generating negative tokens leading to rejective responses, regardless of the input context. To address this, we introduced a learning-free method, Token-level Safety-Debiased Inference (TSDI), to estimate and correct this bias during the generation process using randomly constructed prompts. Our experiments demonstrated that our method could enhance the model's helpfulness while maintaining safety, thus improving the trade-off Pareto-front.
Authors: Siyu Xu, Yunke Wang, Chenghao Xia, Dihao Zhu, Tao Huang, Chang Xu
Abstract: Vision-Language-Action (VLA) model can process instructions and visual perception to directly generate actions as output in an end-to-end fashion due to its strong multi-modal reasoning capabilities. While the performance of VLA models is promising, their computational cost can be substantial. This raises challenge for applying them on robotics tasks, which requires real-time decision-making to respond quickly to environmental changes. Since robotic control involves sequential decision-making, the visual input often exhibits minimal variation between successive steps. A natural idea is to reuse the computational results of unchanged visual tokens from the last step. Motivated by this idea, we propose VLA-Cache, an efficient vision-language-action model. VLA-Cache incorporates a token-selection mechanism that compares the visual input at each step with the input from the previous step, adaptively identifying visual tokens with minimal changes. The computational results for these unchanged tokens are then reused in subsequent steps via KV-cache, thereby significantly improving the efficiency of the VLA-Cache model. Experimental results on both simulation (e.g., LIBERO benchmark and SIMPLER) and real-world robot valid VLA-Cache can achieve practical acceleration with minimal sacrifice in success rate.
Authors: Felix Hofst\"atter, Teun van der Weij, Jayden Teoh, Henning Bartsch, Francis Rhys Ward
Abstract: Capability evaluations are required to understand and regulate AI systems that may be deployed or further developed. Therefore, it is important that evaluations provide an accurate estimation of an AI system's capabilities. However, in numerous cases, previously latent capabilities have been elicited from models, sometimes long after initial release. Accordingly, substantial efforts have been made to develop methods for eliciting latent capabilities from models. In this paper, we evaluate the effectiveness of capability elicitation techniques by intentionally training model organisms -- language models with hidden capabilities that are revealed by a password. We introduce a novel method for training model organisms, based on circuit breaking, which is more robust to elicitation techniques than standard password-locked models. We focus on elicitation techniques based on prompting and activation steering, and compare these to fine-tuning methods. Prompting techniques can elicit the actual capability of both password-locked and circuit-broken model organisms in an MCQA setting, while steering fails to do so. For a code-generation task, only fine-tuning can elicit the hidden capabilities of our novel model organism. Additionally, our results suggest that combining techniques improves elicitation. Still, if possible, fine-tuning should be the method of choice to improve the trustworthiness of capability evaluations.
Authors: Maxence Faldor, Robert Tjarko Lange, Antoine Cully
Abstract: Quality-Diversity has emerged as a powerful family of evolutionary algorithms that generate diverse populations of high-performing solutions by implementing local competition principles inspired by biological evolution. While these algorithms successfully foster diversity and innovation, their specific mechanisms rely on heuristics, such as grid-based competition in MAP-Elites or nearest-neighbor competition in unstructured archives. In this work, we propose a fundamentally different approach: using meta-learning to automatically discover novel Quality-Diversity algorithms. By parameterizing the competition rules using attention-based neural architectures, we evolve new algorithms that capture complex relationships between individuals in the descriptor space. Our discovered algorithms demonstrate competitive or superior performance compared to established Quality-Diversity baselines while exhibiting strong generalization to higher dimensions, larger populations, and out-of-distribution domains like robot control. Notably, even when optimized solely for fitness, these algorithms naturally maintain diverse populations, suggesting meta-learning rediscovers that diversity is fundamental to effective optimization.
Authors: Feng Wang, Hong Qiu, Yingying Huang, Xiaozhe Gu, Renfang Wang, Bo Yang
Abstract: Magnetotelluric (MT) forward modeling is fundamental for improving the accuracy and efficiency of MT inversion. Neural operators (NOs) have been effectively used for rapid MT forward modeling, demonstrating their promising performance in solving the MT forward modeling-related partial differential equations (PDEs). Particularly, they can obtain the electromagnetic field at arbitrary locations and frequencies. In these NOs, the projection layers have been dominated by multi-layer perceptrons (MLPs), which may potentially reduce the accuracy of solution due to they usually suffer from the disadvantages of MLPs, such as lack of interpretability, overfitting, and so on. Therefore, to improve the accuracy of MT forward modeling with NOs and explore the potential alternatives to MLPs, we propose a novel neural operator by extending the Fourier neural operator (FNO) with Kolmogorov-Arnold network (EFKAN). Within the EFKAN framework, the FNO serves as the branch network to calculate the apparent resistivity and phase from the resistivity model in the frequency domain. Meanwhile, the KAN acts as the trunk network to project the resistivity and phase, determined by the FNO, to the desired locations and frequencies. Experimental results demonstrate that the proposed method not only achieves higher accuracy in obtaining apparent resistivity and phase compared to the NO equipped with MLPs at the desired frequencies and locations but also outperforms traditional numerical methods in terms of computational speed.
Authors: Felix Drinkall, Janet B. Pierrehumbert, Stefan Zohren
Abstract: Large language models (LLMs) have shown remarkable success in language modelling due to scaling laws found in model size and the hidden dimension of the model's text representation. Yet, we demonstrate that compressed representations of text can yield better performance in LLM-based regression tasks. In this paper, we compare the relative performance of embedding compression in three different signal-to-noise contexts: financial return prediction, writing quality assessment and review scoring. Our results show that compressing embeddings, in a minimally supervised manner using an autoencoder's hidden representation, can mitigate overfitting and improve performance on noisy tasks, such as financial return prediction; but that compression reduces performance on tasks that have high causal dependencies between the input and target data. Our results suggest that the success of interpretable compressed representations such as sentiment may be due to a regularising effect.
Authors: Giovanni Birolo, Ivan Rossi, Flavio Sartori, Cesare Rollo, Tiziana Sanavia, Piero Fariselli
Abstract: Survival analysis, a foundational tool for modeling time-to-event data, has seen growing integration with machine learning (ML) approaches to handle the complexities of censored data and time-varying risks. Despite these advances, leveraging state-of-the-art survival models remains a challenge due to the fragmented nature of existing implementations, which lack standardized interfaces and require extensive preprocessing. We introduce SurvHive, a Python-based framework designed to unify survival analysis methods within a coherent and extensible interface modeled on scikit-learn. SurvHive integrates classical statistical models with cutting-edge deep learning approaches, including transformer-based architectures and parametric survival models. Using a consistent API, SurvHive simplifies model training, evaluation, and optimization, significantly reducing the barrier to entry for ML practitioners exploring survival analysis. The package includes enhanced support for hyper-parameter tuning, time-dependent risk evaluation metrics, and cross-validation strategies tailored to censored data. With its extensibility and focus on usability, SurvHive provides a bridge between survival analysis and the broader ML community, facilitating advancements in time-to-event modeling across domains. The SurvHive code and documentation are available freely at https://github.com/compbiomed-unito/survhive.
Authors: Satyajeet Sahoo, Jhareswar Maiti
Abstract: Cosine similarity is a popular distance measure that measures the similarity between two vectors in the inner product space. It is widely used in many data classification algorithms like K-Nearest Neighbors, Clustering etc. This study demonstrates limitations of application of cosine similarity. Particularly, this study demonstrates that traditional cosine similarity metric is valid only in the Euclidean space, whereas the original data resides in a random variable space. When there is variance and correlation in the data, then cosine distance is not a completely accurate measure of similarity. While new similarity and distance metrics have been developed to make up for the limitations of cosine similarity, these metrics are used as substitutes to cosine distance, and do not make modifications to cosine distance to overcome its limitations. Subsequently, we propose a modified cosine similarity metric, where cosine distance is adjusted by variance-covariance of the data. Application of variance-adjusted cosine distance gives better similarity performance compared to traditional cosine distance. KNN modelling on the Wisconsin Breast Cancer Dataset is performed using both traditional and modified cosine similarity measures and compared. The modified formula shows 100% test accuracy on the data.
Authors: Zhenglai Li, Yuqi Shi, Xiao He, Chang Tang
Abstract: Multi-view clustering (MvC) utilizes information from multiple views to uncover the underlying structures of data. Despite significant advancements in MvC, mitigating the impact of missing samples in specific views on the integration of knowledge from different views remains a critical challenge. This paper proposes a novel Mask-informed Deep Contrastive Incomplete Multi-view Clustering (Mask-IMvC) method, which elegantly identifies a view-common representation for clustering. Specifically, we introduce a mask-informed fusion network that aggregates incomplete multi-view information while considering the observation status of samples across various views as a mask, thereby reducing the adverse effects of missing values. Additionally, we design a prior knowledge-assisted contrastive learning loss that boosts the representation capability of the aggregated view-common representation by injecting neighborhood information of samples from different views. Finally, extensive experiments are conducted to demonstrate the superiority of the proposed Mask-IMvC method over state-of-the-art approaches across multiple MvC datasets, both in complete and incomplete scenarios.
Authors: Bangzhen Liu, Chenxi Zheng, Xuemiao Xu, Cheng Xu, Huaidong Zhang, Shengfeng He
Abstract: The vulnerability of 3D point cloud analysis to unpredictable rotations poses an open yet challenging problem: orientation-aware 3D domain generalization. Cross-domain robustness and adaptability of 3D representations are crucial but not easily achieved through rotation augmentation. Motivated by the inherent advantages of intricate orientations in enhancing generalizability, we propose an innovative rotation-adaptive domain generalization framework for 3D point cloud analysis. Our approach aims to alleviate orientational shifts by leveraging intricate samples in an iterative learning process. Specifically, we identify the most challenging rotation for each point cloud and construct an intricate orientation set by optimizing intricate orientations. Subsequently, we employ an orientation-aware contrastive learning framework that incorporates an orientation consistency loss and a margin separation loss, enabling effective learning of categorically discriminative and generalizable features with rotation consistency. Extensive experiments and ablations conducted on 3D cross-domain benchmarks firmly establish the state-of-the-art performance of our proposed approach in the context of orientation-aware 3D domain generalization.
Authors: Yuxuan Zhang, Luciano Sebastian Martinez-Rau, Quynh Nguyen Phuong Vu, Bengt Oelmann, Sebastian Bader
Abstract: Structural Health Monitoring (SHM) ensures the safety and longevity of infrastructure by enabling timely damage detection. Vision-based crack detection, combined with UAVs, addresses the limitations of traditional sensor-based SHM methods but requires the deployment of efficient deep learning models on resource-constrained devices. This study evaluates two lightweight convolutional neural network models, MobileNetV1x0.25 and MobileNetV2x0.5, across TensorFlow, PyTorch, and Open Neural Network Exchange platforms using three quantization techniques: dynamic quantization, post-training quantization (PTQ), and quantization-aware training (QAT). Results show that QAT consistently achieves near-floating-point accuracy, such as an F1-score of 0.8376 for MBNV2x0.5 with Torch-QAT, while maintaining efficient resource usage. PTQ significantly reduces memory and energy consumption but suffers from accuracy loss, particularly in TensorFlow. Dynamic quantization preserves accuracy but faces deployment challenges on PyTorch. By leveraging QAT, this work enables real-time, low-power crack detection on UAVs, enhancing safety, scalability, and cost-efficiency in SHM applications, while providing insights into balancing accuracy and efficiency across different platforms for autonomous inspections.
Authors: Hyeonho Noh, Byonghyo Shim, Hyun Jong Yang
Abstract: Deep learning (DL) has made notable progress in addressing complex radio access network control challenges that conventional analytic methods have struggled to solve. However, DL has shown limitations in solving constrained NP-hard problems often encountered in network optimization, such as those involving quality of service (QoS) or discrete variables like user indices. Current solutions rely on domain-specific architectures or heuristic techniques, and a general DL approach for constrained optimization remains undeveloped. Moreover, even minor changes in communication objectives demand time-consuming retraining, limiting their adaptability to dynamic environments where task objectives, constraints, environmental factors, and communication scenarios frequently change. To address these challenges, we propose a large language model for resource allocation optimizer (LLM-RAO), a novel approach that harnesses the capabilities of LLMs to address the complex resource allocation problem while adhering to QoS constraints. By employing a prompt-based tuning strategy to flexibly convey ever-changing task descriptions and requirements to the LLM, LLM-RAO demonstrates robust performance and seamless adaptability in dynamic environments without requiring extensive retraining. Simulation results reveal that LLM-RAO achieves up to a 40% performance enhancement compared to conventional DL methods and up to an $80$\% improvement over analytical approaches. Moreover, in scenarios with fluctuating communication objectives, LLM-RAO attains up to 2.9 times the performance of traditional DL-based networks.
Authors: Fotis I. Giasemis, Vladimir Lon\v{c}ar, Bertrand Granado, Vladimir Vava Gligorov
Abstract: In high-energy physics, the increasing luminosity and detector granularity at the Large Hadron Collider are driving the need for more efficient data processing solutions. Machine Learning has emerged as a promising tool for reconstructing charged particle tracks, due to its potentially linear computational scaling with detector hits. The recent implementation of a graph neural network-based track reconstruction pipeline in the first level trigger of the LHCb experiment on GPUs serves as a platform for comparative studies between computational architectures in the context of high-energy physics. This paper presents a novel comparison of the throughput of ML model inference between FPGAs and GPUs, focusing on the first step of the track reconstruction pipeline$\unicode{x2013}$an implementation of a multilayer perceptron. Using HLS4ML for FPGA deployment, we benchmark its performance against the GPU implementation and demonstrate the potential of FPGAs for high-throughput, low-latency inference without the need for an expertise in FPGA development and while consuming significantly less power.
Authors: Galen Reeves, Henry D. Pfister
Abstract: This paper provides an elementary, self-contained analysis of diffusion-based sampling methods for generative modeling. In contrast to existing approaches that rely on continuous-time processes and then discretize, our treatment works directly with discrete-time stochastic processes and yields precise non-asymptotic convergence guarantees under broad assumptions. The key insight is to couple the sampling process of interest with an idealized comparison process that has an explicit Gaussian-convolution structure. We then leverage simple identities from information theory, including the I-MMSE relationship, to bound the discrepancy (in terms of the Kullback-Leibler divergence) between these two discrete-time processes. In particular, we show that, if the diffusion step sizes are chosen sufficiently small and one can approximate certain conditional mean estimators well, then the sampling distribution is provably close to the target distribution. Our results also provide a transparent view on how to accelerate convergence by introducing additional randomness in each step to match higher order moments in the comparison process.
Authors: Nils Ingelhag, Jesper Munkeby, Michael C. Welle, Marco Moletta, Danica Kragic
Abstract: We present a Real-Time Operator Takeover (RTOT) paradigm enabling operators to seamlessly take control of a live visuomotor diffusion policy, guiding the system back into desirable states or reinforcing specific demonstrations. We presents new insights in using the Mahalonobis distance to automaicaly identify undesirable states. Once the operator has intervened and redirected the system, the control is seamlessly returned to the policy, which resumes generating actions until further intervention is required. We demonstrate that incorporating the targeted takeover demonstrations significantly improves policy performance compared to training solely with an equivalent number of, but longer, initial demonstrations. We provide an in-depth analysis of using the Mahalanobis distance to detect out-of-distribution states, illustrating its utility for identifying critical failure points during execution. Supporting materials, including videos of initial and takeover demonstrations and all rice-scooping experiments, are available on the project website: https://operator-takeover.github.io/
Authors: Lavanya Ratnabala, Aleksey Fedoseev, Robinroy Peter, Dzmitry Tsetserukou
Abstract: This paper addresses the challenge of decentralized task allocation within heterogeneous multi-agent systems operating under communication constraints. We introduce a novel framework that integrates graph neural networks (GNNs) with a centralized training and decentralized execution (CTDE) paradigm, further enhanced by a tailored Proximal Policy Optimization (PPO) algorithm for multi-agent deep reinforcement learning (MARL). Our approach enables unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs) to dynamically allocate tasks efficiently without necessitating central coordination in a 3D grid environment. The framework minimizes total travel time while simultaneously avoiding conflicts in task assignments. For the cost calculation and routing, we employ reservation-based A* and R* path planners. Experimental results revealed that our method achieves a high 92.5% conflict-free success rate, with only a 7.49% performance gap compared to the centralized Hungarian method, while outperforming the heuristic decentralized baseline based on greedy approach. Additionally, the framework exhibits scalability with up to 20 agents with allocation processing of 2.8 s and robustness in responding to dynamically generated tasks, underscoring its potential for real-world applications in complex multi-agent scenarios.
Authors: Siyu Wang, Xiaocong Chen, Lina Yao
Abstract: In offline reinforcement learning-based recommender systems (RLRS), learning effective state representations is crucial for capturing user preferences that directly impact long-term rewards. However, raw state representations often contain high-dimensional, noisy information and components that are not causally relevant to the reward. Additionally, missing transitions in offline data make it challenging to accurately identify features that are most relevant to user satisfaction. To address these challenges, we propose Policy-Guided Causal Representation (PGCR), a novel two-stage framework for causal feature selection and state representation learning in offline RLRS. In the first stage, we learn a causal feature selection policy that generates modified states by isolating and retaining only the causally relevant components (CRCs) while altering irrelevant components. This policy is guided by a reward function based on the Wasserstein distance, which measures the causal effect of state components on the reward and encourages the preservation of CRCs that directly influence user interests. In the second stage, we train an encoder to learn compact state representations by minimizing the mean squared error (MSE) loss between the latent representations of the original and modified states, ensuring that the representations focus on CRCs. We provide a theoretical analysis proving the identifiability of causal effects from interventions, validating the ability of PGCR to isolate critical state components for decision-making. Extensive experiments demonstrate that PGCR significantly improves recommendation performance, confirming its effectiveness for offline RL-based recommender systems.
Authors: Nikita Tsoy, Ivan Kirev, Negin Rahimiyazdi, Nikola Konstantinov
Abstract: Performativity, the phenomenon where outcomes are influenced by predictions, is particularly prevalent in social contexts where individuals strategically respond to a deployed model. In order to preserve the high accuracy of machine learning models under distribution shifts caused by performativity, Perdomo et al. (2020) introduced the concept of performative risk minimization (PRM). While this framework ensures model accuracy, it overlooks the impact of the PRM on the underlying distributions and the predictions of the model. In this paper, we initiate the analysis of the impact of PRM, by studying performativity for a sequential performative risk minimization problem with binary random variables and linear performative shifts. We formulate two natural measures of impact. In the case of full information, where the distribution dynamics are known, we derive explicit formulas for the PRM solution and our impact measures. In the case of partial information, we provide performative-aware statistical estimators, as well as simulations. Our analysis contrasts PRM to alternatives that do not model data shift and indicates that PRM can have amplified side effects compared to such methods.
Authors: Donglin Zhan, Leonardo F. Toso, James Anderson
Abstract: We study task selection to enhance sample efficiency in model-agnostic meta-reinforcement learning (MAML-RL). Traditional meta-RL typically assumes that all available tasks are equally important, which can lead to task redundancy when they share significant similarities. To address this, we propose a coreset-based task selection approach that selects a weighted subset of tasks based on how diverse they are in gradient space, prioritizing the most informative and diverse tasks. Such task selection reduces the number of samples needed to find an $\epsilon$-close stationary solution by a factor of O(1/$\epsilon$). Consequently, it guarantees a faster adaptation to unseen tasks while focusing training on the most relevant tasks. As a case study, we incorporate task selection to MAML-LQR (Toso et al., 2024b), and prove a sample complexity reduction proportional to O(log(1/$\epsilon$)) when the task specific cost also satisfy gradient dominance. Our theoretical guarantees underscore task selection as a key component for scalable and sample-efficient meta-RL. We numerically validate this trend across multiple RL benchmark problems, illustrating the benefits of task selection beyond the LQR baseline.
Authors: Jean Panaioti Jordanou, Eduardo Camponogara, Eduardo Gildin
Abstract: Linear Parameter Varying (LPV) Systems are a well-established class of nonlinear systems with a rich theory for stability analysis, control, and analytical response finding, among other aspects. Although there are works on data-driven identification of such systems, the literature is quite scarce in terms of works that tackle the identification of LPV models for large-scale systems. Since large-scale systems are ubiquitous in practice, this work develops a methodology for the local and global identification of large-scale LPV systems based on nonintrusive reduced-order modeling. The developed method is coined as DMD-LPV for being inspired in the Dynamic Mode Decomposition (DMD). To validate the proposed identification method, we identify a system described by a discretized linear diffusion equation, with the diffusion gain defined by a polynomial over a parameter. The experiments show that the proposed method can easily identify a reduced-order LPV model of a given large-scale system without the need to perform identification in the full-order dimension, and with almost no performance decay over performing a reduction, given that the model structure is well-established.
Authors: Wenzhe Yin, Zehao Xiao, Jiayi Shen, Yunlu Chen, Cees G. M. Snoek, Jan-Jakob Sonke, Efstratios Gavves
Abstract: This paper addresses the challenge of Neural Field (NeF) generalization, where models must efficiently adapt to new signals given only a few observations. To tackle this, we propose Geometric Neural Process Fields (G-NPF), a probabilistic framework for neural radiance fields that explicitly captures uncertainty. We formulate NeF generalization as a probabilistic problem, enabling direct inference of NeF function distributions from limited context observations. To incorporate structural inductive biases, we introduce a set of geometric bases that encode spatial structure and facilitate the inference of NeF function distributions. Building on these bases, we design a hierarchical latent variable model, allowing G-NPF to integrate structural information across multiple spatial levels and effectively parameterize INR functions. This hierarchical approach improves generalization to novel scenes and unseen signals. Experiments on novel-view synthesis for 3D scenes, as well as 2D image and 1D signal regression, demonstrate the effectiveness of our method in capturing uncertainty and leveraging structural information for improved generalization.
Authors: Stefano Cortinovis, Fran\c{c}ois Caron
Abstract: Prediction-powered inference (PPI) enables valid statistical inference by combining experimental data with machine learning predictions. When a sufficient number of high-quality predictions is available, PPI results in more accurate estimates and tighter confidence intervals than traditional methods. In this paper, we propose to inform the PPI framework with prior knowledge on the quality of the predictions. The resulting method, which we call frequentist, assisted by Bayes, PPI (FAB-PPI), improves over PPI when the observed prediction quality is likely under the prior, while maintaining its frequentist guarantees. Furthermore, when using heavy-tailed priors, FAB-PPI adaptively reverts to standard PPI in low prior probability regions. We demonstrate the benefits of FAB-PPI in real and synthetic examples.
Authors: Yaroslav Balytskyi, Inna Hubenko, Alina Balytska, Christopher V. Kelly
Abstract: Accurate identification of druggable pockets is essential for structure-based drug design. However, most pocket-identification algorithms prioritize their geometric properties over downstream docking performance. To address this limitation, we developed RAPID-Net, a pocket-finding algorithm for seamless integration with docking workflows. When guiding AutoDock Vina, RAPID-Net outperforms DiffBindFR on the PoseBusters benchmark and enables blind docking on large proteins that AlphaFold 3 cannot process as a whole. Furthermore, RAPID-Net surpasses PUResNet and Kalasanty in docking accuracy and pocket-ligand intersection rates across diverse datasets, including PoseBusters, Astex Diverse Set, BU48, and Coach420. When accuracy is evaluated as ``at least one correct pose in the ensemble'', RAPID-Net outperforms AlphaFold 3 on the PoseBusters benchmark, suggesting that our approach can be further improved with a suitable pose reweighting tool offering a cost-effective and competitive alternative to AlphaFold 3 for docking. Finally, using several therapeutically relevant examples, we demonstrate the ability of RAPID-Net to identify remote functional sites, highlighting its potential to facilitate the development of innovative therapeutics.
Authors: Federico Zocco, Joan Garc\'ia, Wassim M. Haddad
Abstract: The alteration of the climate in various areas of the world is of increasing concern since climate stability is a necessary condition for human survival as well as every living organism. The main reason of climate change is the greenhouse effect caused by the accumulation of carbon dioxide in the atmosphere. In this paper, we design a networked system underpinned by compartmental dynamical thermodynamics to circulate the atmospheric carbon dioxide. Specifically, in the carbon dioxide emitter compartment, we develop an initial-condition-dependent finite-time stabilizing controller that guarantees stability within a desired time leveraging the system property of affinity in the control. Then, to compensate for carbon emissions we show that a cultivation of microalgae with a volume 625 times bigger than the one of the carbon emitter is required. To increase the carbon uptake of the microalgae, we implement the nonaffine-in-the-control microalgae dynamical equations as an environment of a state-of-the-art library for reinforcement learning (RL), namely, Stable-Baselines3, and then, through the library, we test the performance of eight RL algorithms for training a controller that maximizes the microalgae absorption of carbon through the light intensity. All the eight controllers increased the carbon absorption of the cultivation during a training of 200,000 time steps with a maximum episode length of 200 time steps and with no termination conditions. This work is a first step towards approaching net zero as a classical and learning-based network control problem. The source code is publicly available.
Authors: Yangyang Li, Yuhua Xu, Wen Li, Guoxin Li, Zhibing Feng, Songyi Liu, Jiatao Du, Xinran Li
Abstract: This paper addresses the challenge of anti-jamming in moving reactive jamming scenarios. The moving reactive jammer initiates high-power tracking jamming upon detecting any transmission activity, and when unable to detect a signal, resorts to indiscriminate jamming. This presents dual imperatives: maintaining hiding to avoid the jammer's detection and simultaneously evading indiscriminate jamming. Spread spectrum techniques effectively reduce transmitting power to elude detection but fall short in countering indiscriminate jamming. Conversely, changing communication frequencies can help evade indiscriminate jamming but makes the transmission vulnerable to tracking jamming without spread spectrum techniques to remain hidden. Current methodologies struggle with the complexity of simultaneously optimizing these two requirements due to the expansive joint action spaces and the dynamics of moving reactive jammers. To address these challenges, we propose a parallelized deep reinforcement learning (DRL) strategy. The approach includes a parallelized network architecture designed to decompose the action space. A parallel exploration-exploitation selection mechanism replaces the $\varepsilon $-greedy mechanism, accelerating convergence. Simulations demonstrate a nearly 90\% increase in normalized throughput.
Authors: Tzu-Tao Chang, Shivaram Venkataraman
Abstract: Cross-attention is commonly adopted in multimodal large language models (MLLMs) for integrating visual information into the language backbone. However, in applications with large visual inputs, such as video understanding, processing a large number of visual tokens in cross-attention layers leads to high memory demands and often necessitates distributed computation across multiple GPUs. Existing distributed attention mechanisms face significant communication overheads, making cross-attention layers a critical bottleneck for efficient training and inference of MLLMs. To address this, we propose LV-XAttn, a distributed, exact cross-attention mechanism with minimal communication overhead. We observe that in applications involving large visual inputs the size of the query block is typically much smaller than that of the key-value blocks. Thus, in LV-XAttn we keep the large key-value blocks locally on each GPU and exchange smaller query blocks across GPUs. We also introduce an efficient activation recomputation technique enabling support for longer visual context. We theoretically analyze the communication benefits of LV-XAttn and show that it can achieve speedups for a wide range of models. Our evaluations with mPLUG-Owl3 and OpenFlamingo models find that LV-XAttn achieves up to 5.58$\times$ end-to-end speedup compared to existing approaches.
Authors: Reemt Hinrichs, J\"orn Ostermann
Abstract: Cochlear implants (CIs) are surgically implanted hearing devices, which allow to restore a sense of hearing in people suffering from profound hearing loss. Wireless streaming of audio from external devices to CI signal processors has become common place. Specialized compression based on the stimulation patterns of a CI by deep recurrent autoencoders can decrease the power consumption in such a wireless streaming application through bit-rate reduction at zero latency. While previous research achieved considerable bit-rate reductions, model sizes were ignored, which can be of crucial importance in hearing-aids due to their limited computational resources. This work investigates maximizing objective speech intelligibility of the coded stimulation patterns of deep recurrent autoencoders while minimizing model size. For this purpose, a pruning-aware loss is proposed, which captures the impact of pruning during training. This training with a pruning-aware loss is compared to conventional magnitude-informed pruning and is found to yield considerable improvements in objective intelligibility, especially at higher pruning rates. After fine-tuning, little to no degradation of objective intelligibility is observed up to a pruning rate of about 55\,\%. The proposed pruning-aware loss yields substantial gains in objective speech intelligibility scores after pruning compared to the magnitude-informed baseline for pruning rates above 45\,\%.
Authors: R\'obert Busa-Fekete, Julian Zimmert, Andr\'as Gy\"orgy, Linhai Qiu, Tzu-Wei Sung, Hao Shen, Hyomin Choi, Sharmila Subramaniam, Li Xiao
Abstract: Web refresh crawling is the problem of keeping a cache of web pages fresh, that is, having the most recent copy available when a page is requested, given a limited bandwidth available to the crawler. Under the assumption that the change and request events, resp., to each web page follow independent Poisson processes, the optimal scheduling policy was derived by Azar et al. 2018. In this paper, we study an extension of this problem where side information indicating content changes, such as various types of web pings, for example, signals from sitemaps, content delivery networks, etc., is available. Incorporating such side information into the crawling policy is challenging, because (i) the signals can be noisy with false positive events and with missing change events; and (ii) the crawler should achieve a fair performance over web pages regardless of the quality of the side information, which might differ from web page to web page. We propose a scalable crawling algorithm which (i) uses the noisy side information in an optimal way under mild assumptions; (ii) can be deployed without heavy centralized computation; (iii) is able to crawl web pages at a constant total rate without spikes in the total bandwidth usage over any time interval, and automatically adapt to the new optimal solution when the total bandwidth changes without centralized computation. Experiments clearly demonstrate the versatility of our approach.
Authors: Chendi Qian, Christopher Morris
Abstract: Recently, message-passing graph neural networks (MPNNs) have shown potential for solving combinatorial and continuous optimization problems due to their ability to capture variable-constraint interactions. While existing approaches leverage MPNNs to approximate solutions or warm-start traditional solvers, they often lack guarantees for feasibility, particularly in convex optimization settings. Here, we propose an iterative MPNN framework to solve convex optimization problems with provable feasibility guarantees. First, we demonstrate that MPNNs can provably simulate standard interior-point methods for solving quadratic problems with linear constraints, covering relevant problems such as SVMs. Secondly, to ensure feasibility, we introduce a variant that starts from a feasible point and iteratively restricts the search within the feasible region. Experimental results show that our approach outperforms existing neural baselines in solution quality and feasibility, generalizes well to unseen problem sizes, and, in some cases, achieves faster solution times than state-of-the-art solvers such as Gurobi.
Authors: Ting-Ju Wei, Tung-Huan Su, Chuin-Shan Chen
Abstract: Multiscale simulations are indispensable for connecting microstructural features to the macroscopic behavior of polycrystalline materials, but their high computational demands limit their practicality. Deep material networks (DMNs) have been proposed as efficient surrogate models, yet they fall short of capturing texture evolution. To address this limitation, we propose the orientation-aware interaction-based deep material network (ODMN), which incorporates an orientation-aware mechanism and an interaction mechanism grounded in the Hill-Mandel principle. The orientation-aware mechanism learns the crystallographic textures, while the interaction mechanism captures stress-equilibrium directions among representative volume element (RVE) subregions, offering insight into internal microstructural mechanics. Notably, ODMN requires only linear elastic data for training yet generalizes effectively to complex nonlinear and anisotropic responses. Our results show that ODMN accurately predicts both mechanical responses and texture evolution under complex plastic deformation, thus expanding the applicability of DMNs to polycrystalline materials. By balancing computational efficiency with predictive fidelity, ODMN provides a robust framework for multiscale simulations of polycrystalline materials.
Authors: George Whittle, Juliusz Ziomek, Jacob Rawling, Michael A Osborne
Abstract: While Bayesian inference provides a principled framework for reasoning under uncertainty, its widespread adoption is limited by the intractability of exact posterior computation, necessitating the use of approximate inference. However, existing methods are often computationally expensive, or demand costly retraining when priors change, limiting their utility, particularly in sequential inference problems such as real-time sensor fusion. To address these challenges, we introduce the Distribution Transformer -- a novel architecture that can learn arbitrary distribution-to-distribution mappings. Our method can be trained to map a prior to the corresponding posterior, conditioned on some dataset -- thus performing approximate Bayesian inference. Our novel architecture represents a prior distribution as a (universally-approximating) Gaussian Mixture Model (GMM), and transforms it into a GMM representation of the posterior. The components of the GMM attend to each other via self-attention, and to the datapoints via cross-attention. We demonstrate that Distribution Transformers both maintain flexibility to vary the prior, and significantly reduces computation times-from minutes to milliseconds-while achieving log-likelihood performance on par with or superior to existing approximate inference methods across tasks such as sequential inference, quantum system parameter inference, and Gaussian Process predictive posterior inference with hyperpriors.
Authors: Valentina Vadori, Antonella Peruffo, Jean-Marie Gra\"ic, Livio Finos, Enrico Grisan
Abstract: Recent advancements in foundation models have transformed computer vision, driving significant performance improvements across diverse domains, including digital histopathology. However, the advantages of domain-specific histopathology foundation models over general-purpose models for specialized tasks such as cell analysis remain underexplored. This study investigates the representation learning gap between these two categories by analyzing multi-level patch embeddings applied to cell instance segmentation and classification. We implement an encoder-decoder architecture with a consistent decoder and various encoders. These include convolutional, vision transformer (ViT), and hybrid encoders pre-trained on ImageNet-22K or LVD-142M, representing general-purpose foundation models. These are compared against ViT encoders from the recently released UNI, Virchow2, and Prov-GigaPath foundation models, trained on patches extracted from hundreds of thousands of histopathology whole-slide images. The decoder integrates patch embeddings from different encoder depths via skip connections to generate semantic and distance maps. These maps are then post-processed to create instance segmentation masks where each label corresponds to an individual cell and to perform cell-type classification. All encoders remain frozen during training to assess their pre-trained feature extraction capabilities. Using the PanNuke and CoNIC histopathology datasets, and the newly introduced Nissl-stained CytoDArk0 dataset for brain cytoarchitecture studies, we evaluate instance-level detection, segmentation accuracy, and cell-type classification. This study provides insights into the comparative strengths and limitations of general-purpose vs. histopathology foundation models, offering guidance for model selection in cell-focused histopathology and brain cytoarchitecture analysis workflows.
Authors: Grigory Bartosh, Dmitry Vetrov, Christian A. Naesseth
Abstract: The Latent Stochastic Differential Equation (SDE) is a powerful tool for time series and sequence modeling. However, training Latent SDEs typically relies on adjoint sensitivity methods, which depend on simulation and backpropagation through approximate SDE solutions, which limit scalability. In this work, we propose SDE Matching, a new simulation-free method for training Latent SDEs. Inspired by modern Score- and Flow Matching algorithms for learning generative dynamics, we extend these ideas to the domain of stochastic dynamics for time series and sequence modeling, eliminating the need for costly numerical simulations. Our results demonstrate that SDE Matching achieves performance comparable to adjoint sensitivity methods while drastically reducing computational complexity.
Authors: Chenlu Ye, Yujia Jin, Alekh Agarwal, Tong Zhang
Abstract: Typical contextual bandit algorithms assume that the rewards at each round lie in some fixed range $[0, R]$, and their regret scales polynomially with this reward range $R$. However, many practical scenarios naturally involve heavy-tailed rewards or rewards where the worst-case range can be substantially larger than the variance. In this paper, we develop an algorithmic approach building on Catoni's estimator from robust statistics, and apply it to contextual bandits with general function approximation. When the variance of the reward at each round is known, we use a variance-weighted regression approach and establish a regret bound that depends only on the cumulative reward variance and logarithmically on the reward range $R$ as well as the number of rounds $T$. For the unknown-variance case, we further propose a careful peeling-based algorithm and remove the need for cumbersome variance estimation. With additional dependence on the fourth moment, our algorithm also enjoys a variance-based bound with logarithmic reward-range dependence. Moreover, we demonstrate the optimality of the leading-order term in our regret bound through a matching lower bound.
Authors: Edward Ellis, Andrew Bulpitt, Nasim Parsa, Michael F Byrne, Sharib Ali
Abstract: Ultrasound (US) imaging is clinically invaluable due to its noninvasive and safe nature. However, interpreting US images is challenging, requires significant expertise, and time, and is often prone to errors. Deep learning offers assistive solutions such as segmentation. Supervised methods rely on large, high-quality, and consistently labeled datasets, which are challenging to curate. Moreover, these methods tend to underperform on out-of-distribution data, limiting their clinical utility. Self-supervised learning (SSL) has emerged as a promising alternative, leveraging unlabeled data to enhance model performance and generalisability. We introduce a contrastive SSL approach tailored for B-mode US images, incorporating a novel Relation Contrastive Loss (RCL). RCL encourages learning of distinct features by differentiating positive and negative sample pairs through a learnable metric. Additionally, we propose spatial and frequency-based augmentation strategies for the representation learning on US images. Our approach significantly outperforms traditional supervised segmentation methods across three public breast US datasets, particularly in data-limited scenarios. Notable improvements on the Dice similarity metric include a 4% increase on 20% and 50% of the BUSI dataset, nearly 6% and 9% improvements on 20% and 50% of the BrEaST dataset, and 6.4% and 3.7% improvements on 20% and 50% of the UDIAT dataset, respectively. Furthermore, we demonstrate superior generalisability on the out-of-distribution UDIAT dataset with performance boosts of 20.6% and 13.6% compared to the supervised baseline using 20% and 50% of the BUSI and BrEaST training data, respectively. Our research highlights that domain-inspired SSL can improve US segmentation, especially under data-limited conditions.
Authors: Antoni Kowalczuk, Jan Dubi\'nski, Franziska Boenisch, Adam Dziedzic
Abstract: Image autoregressive (IAR) models have surpassed diffusion models (DMs) in both image quality (FID: 1.48 vs. 1.58) and generation speed. However, their privacy risks remain largely unexplored. To address this, we conduct a comprehensive privacy analysis comparing IARs to DMs. We develop a novel membership inference attack (MIA) that achieves a significantly higher success rate in detecting training images (TPR@FPR=1%: 86.38% for IARs vs. 4.91% for DMs). Using this MIA, we perform dataset inference (DI) and find that IARs require as few as six samples to detect dataset membership, compared to 200 for DMs, indicating higher information leakage. Additionally, we extract hundreds of training images from an IAR (e.g., 698 from VAR-d30). Our findings highlight a fundamental privacy-utility trade-off: while IARs excel in generation quality and speed, they are significantly more vulnerable to privacy attacks. This suggests that incorporating techniques from DMs, such as per-token probability modeling using diffusion, could help mitigate IARs' privacy risks. Our code is available at https://github.com/sprintml/privacy_attacks_against_iars.
URLs: https://github.com/sprintml/privacy_attacks_against_iars.
Authors: Huiqun Huang, Cong Chen, Jean-Philippe Monteuuis, Jonathan Petit, Fei Miao
Abstract: Collaborative Object Detection (COD) and collaborative perception can integrate data or features from various entities, and improve object detection accuracy compared with individual perception. However, adversarial attacks pose a potential threat to the deep learning COD models, and introduce high output uncertainty. With unknown attack models, it becomes even more challenging to improve COD resiliency and quantify the output uncertainty for highly dynamic perception scenes such as autonomous vehicles. In this study, we propose the Trusted Uncertainty Quantification in Collaborative Perception framework (TUQCP). TUQCP leverages both adversarial training and uncertainty quantification techniques to enhance the adversarial robustness of existing COD models. More specifically, TUQCP first adds perturbations to the shared information of randomly selected agents during object detection collaboration by adversarial training. TUQCP then alleviates the impacts of adversarial attacks by providing output uncertainty estimation through learning-based module and uncertainty calibration through conformal prediction. Our framework works for early and intermediate collaboration COD models and single-agent object detection models. We evaluate TUQCP on V2X-Sim, a comprehensive collaborative perception dataset for autonomous driving, and demonstrate a 80.41% improvement in object detection accuracy compared to the baselines under the same adversarial attacks. TUQCP demonstrates the importance of uncertainty quantification to COD under adversarial attacks.
Authors: Ron Benchetrit, Idan Lev-Yehudi, Andrey Zhitnikov, Vadim Indelman
Abstract: Partially Observable Markov Decision Processes (POMDPs) provide a robust framework for decision-making under uncertainty in applications such as autonomous driving and robotic exploration. Their extension, $\rho$POMDPs, introduces belief-dependent rewards, enabling explicit reasoning about uncertainty. Existing online $\rho$POMDP solvers for continuous spaces rely on fixed belief representations, limiting adaptability and refinement - critical for tasks such as information-gathering. We present $\rho$POMCPOW, an anytime solver that dynamically refines belief representations, with formal guarantees of improvement over time. To mitigate the high computational cost of updating belief-dependent rewards, we propose a novel incremental computation approach. We demonstrate its effectiveness for common entropy estimators, reducing computational cost by orders of magnitude. Experimental results show that $\rho$POMCPOW outperforms state-of-the-art solvers in both efficiency and solution quality.
Authors: Sam Young, Yeon-jae Jwa, Kazuhiro Terao
Abstract: Effective self-supervised learning (SSL) techniques have been key to unlocking large datasets for representation learning. While many promising methods have been developed using online corpora and captioned photographs, their application to scientific domains, where data encodes highly specialized knowledge, remains in its early stages. We present a self-supervised masked modeling framework for 3D particle trajectory analysis in Time Projection Chambers (TPCs). These detectors produce globally sparse (<1% occupancy) but locally dense point clouds, capturing meter-scale particle trajectories at millimeter resolution. Starting with PointMAE, this work proposes volumetric tokenization to group sparse ionization points into resolution-agnostic patches, as well as an auxiliary energy infilling task to improve trajectory semantics. This approach -- which we call Point-based Liquid Argon Masked Autoencoder (PoLAr-MAE) -- achieves 99.4% track and 97.7% shower classification F-scores, matching that of supervised baselines without any labeled data. While the model learns rich particle trajectory representations, it struggles with sub-token phenomena like overlapping or short-lived particle trajectories. To support further research, we release PILArNet-M -- the largest open LArTPC dataset (1M+ events, 5.2B labeled points) -- to advance SSL in high energy physics (HEP). Project site: https://youngsm.com/polarmae/
Authors: Thijs Overmeer, Tim Janssen, Wim P. M. Nuijten
Abstract: This paper introduces the first Expected Possession Value (EPV) benchmark and a new and improved EPV model for football. Through the introduction of the OJN-Pass-EPV benchmark, we present a novel method to quantitatively assess the quality of EPV models by using pairs of game states with given relative EPVs. Next, we attempt to replicate the results of Fern\'andez et al. (2021) using a dataset containing Dutch Eredivisie and World Cup matches. Following our failure to do so, we propose a new architecture based on U-net-type convolutional neural networks, achieving good results in model loss and Expected Calibration Error. Finally, we present an improved pass model that incorporates ball height and contains a new dual-component pass value model that analyzes reward and risk. The resulting EPV model correctly identifies the higher value state in 78% of the game state pairs in the OJN-Pass-EPV benchmark, demonstrating its ability to accurately assess goal-scoring potential. Our findings can help assess the quality of EPV models, improve EPV predictions, help assess potential reward and risk of passing decisions, and improve player and team performance.
Authors: Xavier Aramayo Carrasco, Maksim Nekrashevich, Petr Mokrov, Evgeny Burnaev, Alexander Korotin
Abstract: Recently, the Gromov-Wasserstein Optimal Transport (GWOT) problem has attracted the special attention of the ML community. In this problem, given two distributions supported on two (possibly different) spaces, one has to find the most isometric map between them. In the discrete variant of GWOT, the task is to learn an assignment between given discrete sets of points. In the more advanced continuous formulation, one aims at recovering a parametric mapping between unknown continuous distributions based on i.i.d. samples derived from them. The clear geometrical intuition behind the GWOT makes it a natural choice for several practical use cases, giving rise to a number of proposed solvers. Some of them claim to solve the continuous version of the problem. At the same time, GWOT is notoriously hard, both theoretically and numerically. Moreover, all existing continuous GWOT solvers still heavily rely on discrete techniques. Natural questions arise: to what extent do existing methods unravel the GWOT problem, what difficulties do they encounter, and under which conditions they are successful? Our benchmark paper is an attempt to answer these questions. We specifically focus on the continuous GWOT as the most interesting and debatable setup. We crash-test existing continuous GWOT approaches on different scenarios, carefully record and analyze the obtained results, and identify issues. Our findings experimentally testify that the scientific community is still missing a reliable continuous GWOT solver, which necessitates further research efforts. As the first step in this direction, we propose a new continuous GWOT method which does not rely on discrete techniques and partially solves some of the problems of the competitors.
Authors: Sunder Ali Khowaja, Kapal Dev, Syed Muhammad Anwar, Marius George Linguraru
Abstract: Self-supervised learning in the federated learning paradigm has been gaining a lot of interest both in industry and research due to the collaborative learning capability on unlabeled yet isolated data. However, self-supervised based federated learning strategies suffer from performance degradation due to label scarcity and diverse data distributions, i.e., data heterogeneity. In this paper, we propose the SelfFed framework for medical images to overcome data heterogeneity and label scarcity issues. The first phase of the SelfFed framework helps to overcome the data heterogeneity issue by leveraging the pre-training paradigm that performs augmentative modeling using Swin Transformer-based encoder in a decentralized manner. The label scarcity issue is addressed by fine-tuning paradigm that introduces a contrastive network and a novel aggregation strategy. We perform our experimental analysis on publicly available medical imaging datasets to show that SelfFed performs better when compared to existing baselines and works. Our method achieves a maximum improvement of 8.8% and 4.1% on Retina and COVID-FL datasets on non-IID datasets. Further, our proposed method outperforms existing baselines even when trained on a few (10%) labeled instances.
Authors: Anh T. Hoang, Zsolt J. Viharos
Abstract: Machine learning is a powerful tool for extracting valuable information and making various predictions from diverse datasets. Traditional machine learning algorithms rely on well-defined input and output variables; however, there are scenarios where the separation between the input and output variables and the underlying, associated input and output layers of the model are unknown. Feature Selection (FS) and Neural Architecture Search (NAS) have emerged as promising solutions in such scenarios. This paper proposes MICS-EFS, a Model Input-Output Configuration Search with Embedded Feature Selection. The methodology explores internal dependencies in the complete input parameter space for classification tasks involving both 1D sensor time-series and 2D image data. MICS-EFS employs a modified encoder-decoder model and the Sequential Forward Search (SFS) algorithm, combining input-output configuration search with embedded feature selection. Experimental results demonstrate the superior performance of MICS-EFS compared to other FS algorithms. Across all tested datasets, MICS-EFS delivered an average accuracy improvement of 1.5% over baseline models, with the accuracy gains ranging from 0.5% to 5.9%. Moreover, the algorithm reduced feature dimensionality to just 2-5% of the original data, significantly enhancing computational efficiency. These results highlight the potential of MICS-EFS to improve model accuracy and efficiency in various machine learning tasks. Furthermore, the proposed method has been validated in a real-world industrial application focused on machining processes, underscoring its effectiveness and practicality in addressing complex input-output challenges.
Authors: John Martin, Hanspeter Schaub
Abstract: Scientific machine learning and the advent of the Physics-Informed Neural Network (PINN) have shown high potential in their ability to solve complex differential equations. One example is the use of PINNs to solve the gravity field modeling problem -- learning convenient representations of the gravitational potential from position and acceleration data. These PINN gravity models, or PINN-GMs, have demonstrated advantages in model compactness, robustness to noise, and sample efficiency when compared to popular alternatives; however, further investigation has revealed various failure modes for these and other machine learning gravity models which this manuscript aims to address. Specifically, this paper introduces the third generation Physics-Informed Neural Network Gravity Model (PINN-GM-III) which includes design changes that solve the problems of feature divergence, bias towards low-altitude samples, numerical instability, and extrapolation error. Six evaluation metrics are proposed to expose these past pitfalls and illustrate the PINN-GM-III's robustness to them. This study concludes by evaluating the PINN-GM-III modeling accuracy on a heterogeneous density asteroid, and comparing its performance to other analytic and machine learning gravity models.
Authors: Andrew Bai, Chih-Kuan Yeh, Cho-Jui Hsieh, Ankur Taly
Abstract: Incrementally fine-tuning foundational models on new tasks or domains is now the de facto approach in NLP. A known pitfall of this approach is the \emph{catastrophic forgetting} of prior knowledge that happens during fine-tuning. A common approach to alleviate such forgetting is to rehearse samples from prior tasks during fine-tuning. Several existing works assume a fixed memory buffer to store prior task examples, while relying on inferences (forward passes) with the model at hand for choosing examples for rehearsal from the buffer. However, given the increasing computational cost of model inference, and decreasing cost of data storage, we focus on the setting to rehearse samples with a fixed computational budget instead of a fixed memory budget. We propose a sampling scheme, \texttt{\bf mix-cd}, that prioritizes rehearsal of ``collateral damage'' samples, which are samples predicted correctly by the prior model but forgotten by the incrementally tuned one. The crux of our scheme is a procedure to efficiently estimate the density of collateral damage samples without incurring additional model inferences. Our approach is computationally efficient, easy to implement, and outperforms several leading continual learning methods in compute-constrained settings. All the code will be publicly available at https://github.com/jybai/mix-cd-rehearsal.
Authors: Hao Liu, Wilson Yan, Matei Zaharia, Pieter Abbeel
Abstract: Enabling long-context understanding remains a key challenge in scaling existing sequence models -- a crucial component in developing generally intelligent models that can process and operate over long temporal horizons that potentially consist of millions of tokens. In this paper, we aim to address these challenges by providing a comprehensive exploration of the full development process for producing 1M context language models and video-language models, setting new benchmarks in language retrieval and new capabilities in long video understanding. We detail our long context data curation process, progressive context extension from 4K to 1M tokens, and present an efficient open-source implementation for scalable training on long sequences. Additionally, we open-source a family of 7B parameter models capable of processing long text documents and videos exceeding 1M tokens.
Authors: Jan Henrik Bertrand, David B. Hoffmann, Jacopo Pio Gargano, Laurent Mombaerts, Jonathan Taws
Abstract: Recent advances in representation learning have successfully leveraged the underlying domain-specific structure of data across various fields. However, representing diverse and complex entities stored in tabular format within a latent space remains challenging. In this paper, we introduce DEEPCAE, a novel method for calculating the regularization term for multi-layer contractive autoencoders (CAEs). Additionally, we formalize a general-purpose entity embedding framework and use it to empirically show that DEEPCAE outperforms all other tested autoencoder variants in both reconstruction performance and downstream prediction performance. Notably, when compared to a stacked CAE across 13 datasets, DEEPCAE achieves a 34% improvement in reconstruction error.
Authors: Dimitris Bertsimas, Vassilis Digalakis Jr, Yu Ma, Phevos Paschalidis
Abstract: We consider the problem of retraining machine learning (ML) models when new batches of data become available. Existing approaches greedily optimize for predictive power independently at each batch, without considering the stability of the model's structure or analytical insights across retraining iterations. We propose a model-agnostic framework for finding sequences of models that are stable across retraining iterations. We develop a mixed-integer optimization formulation that is guaranteed to recover Pareto optimal models (in terms of the predictive power-stability trade-off) with good generalization properties, as well as an efficient polynomial-time algorithm that performs well in practice. We focus on retaining consistent analytical insights-which is important to model interpretability, ease of implementation, and fostering trust with users-by using custom-defined distance metrics that can be directly incorporated into the optimization problem. We evaluate our framework across models (regression, decision trees, boosted trees, and neural networks) and application domains (healthcare, vision, and language), including deployment in a production pipeline at a major US hospital. We find that, on average, a 2% reduction in predictive power leads to a 30% improvement in stability.
Authors: Ray Zirui Zhang, Xiaohui Xie, John S. Lowengrub
Abstract: We propose a new neural network based method for solving inverse problems for partial differential equations (PDEs) by formulating the PDE inverse problem as a bilevel optimization problem. At the upper level, we minimize the data loss with respect to the PDE parameters. At the lower level, we train a neural network to locally approximate the PDE solution operator in the neighborhood of a given set of PDE parameters, which enables an accurate approximation of the descent direction for the upper level optimization problem. The lower level loss function includes the L2 norms of both the residual and its derivative with respect to the PDE parameters. We apply gradient descent simultaneously on both the upper and lower level optimization problems, leading to an effective and fast algorithm. The method, which we refer to as BiLO (Bilevel Local Operator learning), is also able to efficiently infer unknown functions in the PDEs through the introduction of an auxiliary variable. Through extensive experiments over multiple PDE systems, we demonstrate that our method enforces strong PDE constraints, is robust to sparse and noisy data, and eliminates the need to balance the residual and the data loss, which is inherent to the soft PDE constraints in many existing methods.
Authors: Adam S. Shai, Sarah E. Marzen, Lucas Teixeira, Alexander Gietelink Oldenziel, Paul M. Riechers
Abstract: What computational structure are we building into large language models when we train them on next-token prediction? Here, we present evidence that this structure is given by the meta-dynamics of belief updating over hidden states of the data-generating process. Leveraging the theory of optimal prediction, we anticipate and then find that belief states are linearly represented in the residual stream of transformers, even in cases where the predicted belief state geometry has highly nontrivial fractal structure. We investigate cases where the belief state geometry is represented in the final residual stream or distributed across the residual streams of multiple layers, providing a framework to explain these observations. Furthermore we demonstrate that the inferred belief states contain information about the entire future, beyond the local next-token prediction that the transformers are explicitly trained on. Our work provides a general framework connecting the structure of training data to the geometric structure of activations inside transformers.
Authors: Linglong Qian, Yiyuan Yang, Wenjie Du, Jun Wang, Richard Dobsoni, Zina Ibrahim
Abstract: This study investigates the impact of masking strategies on time series imputation models in healthcare settings. While current approaches predominantly rely on random masking for model evaluation, this practice fails to capture the structured nature of missing patterns in clinical data. Using the PhysioNet Challenge 2012 dataset, we analyse how different masking implementations affect both imputation accuracy and downstream clinical predictions across eleven imputation methods. Our results demonstrate that masking choices significantly influence model performance, while recurrent architectures show more consistent performance across strategies. Analysis of downstream mortality prediction reveals that imputation accuracy doesn't necessarily translate to optimal clinical prediction capabilities. Our findings emphasise the need for clinically-informed masking strategies that better reflect real-world missing patterns in healthcare data, suggesting current evaluation frameworks may need reconsideration for reliable clinical deployment.
Authors: Andreas Schlaginhaufen, Maryam Kamgarpour
Abstract: Inverse reinforcement learning (IRL) aims to infer a reward from expert demonstrations, motivated by the idea that the reward, rather than the policy, is the most succinct and transferable description of a task [Ng et al., 2000]. However, the reward corresponding to an optimal policy is not unique, making it unclear if an IRL-learned reward is transferable to new transition laws in the sense that its optimal policy aligns with the optimal policy corresponding to the expert's true reward. Past work has addressed this problem only under the assumption of full access to the expert's policy, guaranteeing transferability when learning from two experts with the same reward but different transition laws that satisfy a specific rank condition [Rolland et al., 2022]. In this work, we show that the conditions developed under full access to the expert's policy cannot guarantee transferability in the more practical scenario where we have access only to demonstrations of the expert. Instead of a binary rank condition, we propose principal angles as a more refined measure of similarity and dissimilarity between transition laws. Based on this, we then establish two key results: 1) a sufficient condition for transferability to any transition laws when learning from at least two experts with sufficiently different transition laws, and 2) a sufficient condition for transferability to local changes in the transition law when learning from a single expert. Furthermore, we also provide a probably approximately correct (PAC) algorithm and an end-to-end analysis for learning transferable rewards from demonstrations of multiple experts.
Authors: Hojoon Lee, Hyeonseo Cho, Hyunseung Kim, Donghu Kim, Dugki Min, Jaegul Choo, Clare Lyle
Abstract: This study investigates the loss of generalization ability in neural networks, revisiting warm-starting experiments from Ash & Adams. Our empirical analysis reveals that common methods designed to enhance plasticity by maintaining trainability provide limited benefits to generalization. While reinitializing the network can be effective, it also risks losing valuable prior knowledge. To this end, we introduce the Hare & Tortoise, inspired by the brain's complementary learning system. Hare & Tortoise consists of two components: the Hare network, which rapidly adapts to new information analogously to the hippocampus, and the Tortoise network, which gradually integrates knowledge akin to the neocortex. By periodically reinitializing the Hare network to the Tortoise's weights, our method preserves plasticity while retaining general knowledge. Hare & Tortoise can effectively maintain the network's ability to generalize, which improves advanced reinforcement learning algorithms on the Atari-100k benchmark. The code is available at https://github.com/dojeon-ai/hare-tortoise.
Authors: Ilia Sucholutsky, Katherine M. Collins, Maya Malaviya, Nori Jacoby, Weiyang Liu, Theodore R. Sumers, Michalis Korakakis, Umang Bhatt, Mark Ho, Joshua B. Tenenbaum, Brad Love, Zachary A. Pardos, Adrian Weller, Thomas L. Griffiths
Abstract: A good teacher should not only be knowledgeable, but should also be able to communicate in a way that the student understands -- to share the student's representation of the world. In this work, we introduce a new controlled experimental setting, GRADE, to study pedagogy and representational alignment. We use GRADE through a series of machine-machine and machine-human teaching experiments to characterize a utility curve defining a relationship between representational alignment, teacher expertise, and student learning outcomes. We find that improved representational alignment with a student improves student learning outcomes (i.e., task accuracy), but that this effect is moderated by the size and representational diversity of the class being taught. We use these insights to design a preliminary classroom matching procedure, GRADE-Match, that optimizes the assignment of students to teachers. When designing machine teachers, our results suggest that it is important to focus not only on accuracy, but also on representational alignment with human learners.
Authors: Pankayaraj Pathmanathan, Souradip Chakraborty, Xiangyu Liu, Yongyuan Liang, Furong Huang
Abstract: Recent advancements in Reinforcement Learning with Human Feedback (RLHF) have significantly impacted the alignment of Large Language Models (LLMs). The sensitivity of reinforcement learning algorithms such as Proximal Policy Optimization (PPO) has led to new line work on Direct Policy Optimization (DPO), which treats RLHF in a supervised learning framework. The increased practical use of these RLHF methods warrants an analysis of their vulnerabilities. In this work, we investigate the vulnerabilities of DPO to poisoning attacks under different scenarios and compare the effectiveness of preference poisoning, a first of its kind. We comprehensively analyze DPO's vulnerabilities under different types of attacks, i.e., backdoor and non-backdoor attacks, and different poisoning methods across a wide array of language models, i.e., LLama 7B, Mistral 7B, and Gemma 7B. We find that unlike PPO-based methods, which, when it comes to backdoor attacks, require at least 4\% of the data to be poisoned to elicit harmful behavior, we exploit the true vulnerabilities of DPO more simply so we can poison the model with only as much as 0.5\% of the data. We further investigate the potential reasons behind the vulnerability and how well this vulnerability translates into backdoor vs non-backdoor attacks.
Authors: Huiqun Huang, Sihong He, Fei Miao
Abstract: Trajectory prediction models that can infer both finite future trajectories and their associated uncertainties of the target vehicles in an online setting (e.g., real-world application scenarios) is crucial for ensuring the safe and robust navigation and path planning of autonomous vehicle motion. However, the majority of existing trajectory prediction models have neither considered reducing the uncertainty as one objective during the training stage nor provided reliable uncertainty quantification during inference stage under potential distribution shift. Therefore, in this paper, we propose the Conformal Uncertainty Quantification under Distribution Shift framework, CUQDS, to quantify the uncertainty of the predicted trajectories of existing trajectory prediction models under potential data distribution shift, while considering improving the prediction accuracy of the models and reducing the estimated uncertainty during the training stage. Specifically, CUQDS includes 1) a learning-based Gaussian process regression module that models the output distribution of the base model (any existing trajectory prediction or time series forecasting neural networks) and reduces the estimated uncertainty by additional loss term, and 2) a statistical-based Conformal P control module to calibrate the estimated uncertainty from the Gaussian process regression module in an online setting under potential distribution shift between training and testing data.
Authors: Xisen Jin, Xiang Ren
Abstract: Large Language models (LLMs) suffer from forgetting of upstream data when fine-tuned. Despite efforts on mitigating forgetting, few have investigated whether, and how forgotten upstream examples are dependent on newly learned tasks. Insights on such dependencies enable efficient and targeted mitigation of forgetting. In this paper, we empirically analyze forgetting that occurs in $N$ upstream examples of language modeling or instruction-tuning after fine-tuning LLMs on one of $M$ new tasks, visualized in $M\times N$ matrices. We show that the matrices are often well-approximated with low-rank matrices, indicating the dominance of simple associations between the learned tasks and forgotten upstream examples. Leveraging the analysis, we predict forgetting of upstream examples when fine-tuning on unseen tasks with matrix completion over the empirical associations. This enables fast identification of most forgotten examples without expensive inference on the entire upstream data. The approach, despite simplicity, outperforms prior approaches that learn semantic relationships of learned tasks and upstream examples with LMs for predicting forgetting. We demonstrate the practical utility of our analysis by showing statistically significantly reduced forgetting as we upweight predicted examples for replay at fine-tuning. Project page: https://inklab.usc.edu/lm-forgetting-prediction/
Authors: Gianluca Drappo, Alberto Maria Metelli, Marcello Restelli
Abstract: Hierarchical Reinforcement Learning (HRL) approaches have shown successful results in solving a large variety of complex, structured, long-horizon problems. Nevertheless, a full theoretical understanding of this empirical evidence is currently missing. In the context of the \emph{option} framework, prior research has devised efficient algorithms for scenarios where options are fixed, and the high-level policy selecting among options only has to be learned. However, the fully realistic scenario in which both the high-level and the low-level policies are learned is surprisingly disregarded from a theoretical perspective. This work makes a step towards the understanding of this latter scenario. Focusing on the finite-horizon problem, we present a meta-algorithm alternating between regret minimization algorithms instanced at different (high and low) temporal abstractions. At the higher level, we treat the problem as a Semi-Markov Decision Process (SMDP), with fixed low-level policies, while at a lower level, inner option policies are learned with a fixed high-level policy. The bounds derived are compared with the lower bound for non-hierarchical finite-horizon problems, allowing to characterize when a hierarchical approach is provably preferable, even without pre-trained options.
Authors: Linglong Qian, Tao Wang, Jun Wang, Hugh Logan Ellis, Robin Mitra, Richard Dobson, Zina Ibrahim
Abstract: We present a comprehensive analysis of deep learning approaches for Electronic Health Record (EHR) time-series imputation, examining how architectural and framework biases combine to influence model performance. Our investigation reveals varying capabilities of deep imputers in capturing complex spatiotemporal dependencies within EHRs, and that model effectiveness depends on how its combined biases align with medical time-series characteristics. Our experimental evaluation challenges common assumptions about model complexity, demonstrating that larger models do not necessarily improve performance. Rather, carefully designed architectures can better capture the complex patterns inherent in clinical data. The study highlights the need for imputation approaches that prioritise clinically meaningful data reconstruction over statistical accuracy. Our experiments show imputation performance variations of up to 20\% based on preprocessing and implementation choices, emphasising the need for standardised benchmarking methodologies. Finally, we identify critical gaps between current deep imputation methods and medical requirements, highlighting the importance of integrating clinical insights to achieve more reliable imputation approaches for healthcare applications.
Authors: Hengyu Fu, Zehao Dou, Jiawei Guo, Mengdi Wang, Minshuo Chen
Abstract: Diffusion Transformer, the backbone of Sora for video generation, successfully scales the capacity of diffusion models, pioneering new avenues for high-fidelity sequential data generation. Unlike static data such as images, sequential data consists of consecutive data frames indexed by time, exhibiting rich spatial and temporal dependencies. These dependencies represent the underlying dynamic model and are critical to validate the generated data. In this paper, we make the first theoretical step towards bridging diffusion transformers for capturing spatial-temporal dependencies. Specifically, we establish score approximation and distribution estimation guarantees of diffusion transformers for learning Gaussian process data with covariance functions of various decay patterns. We highlight how the spatial-temporal dependencies are captured and affect learning efficiency. Our study proposes a novel transformer approximation theory, where the transformer acts to unroll an algorithm. We support our theoretical results by numerical experiments, providing strong evidence that spatial-temporal dependencies are captured within attention layers, aligning with our approximation theory.
Authors: Xiaoyu Wang, Guojun Xiong, Houwei Cao, Jian Li, Yong Liu
Abstract: Federated Learning (FL) trains a shared model using data and computation power on distributed agents coordinated by a central server. Decentralized FL (DFL) utilizes local model exchange and aggregation between agents to reduce the communication and computation overheads on the central server. However, when agents are mobile, the communication opportunity between agents can be sporadic, largely hindering the convergence and accuracy of DFL. In this paper, we propose Cached Decentralized Federated Learning (Cached-DFL) to investigate delay-tolerant model spreading and aggregation enabled by model caching on mobile agents. Each agent stores not only its own model, but also models of agents encountered in the recent past. When two agents meet, they exchange their own models as well as the cached models. Local model aggregation utilizes all models stored in the cache. We theoretically analyze the convergence of Cached-DFL, explicitly taking into account the model staleness introduced by caching. We design and compare different model caching algorithms for different DFL and mobility scenarios. We conduct detailed case studies in a vehicular network to systematically investigate the interplay between agent mobility, cache staleness, and model convergence. In our experiments, Cached-DFL converges quickly, and significantly outperforms DFL without caching.
Authors: Ming Li, Pengcheng Xu, Junjie Hu, Zeyu Tang, Guang Yang
Abstract: Federated learning holds great potential for enabling large-scale healthcare research and collaboration across multiple centres while ensuring data privacy and security are not compromised. Although numerous recent studies suggest or utilize federated learning based methods in healthcare, it remains unclear which ones have potential clinical utility. This review paper considers and analyzes the most recent studies up to May 2024 that describe federated learning based methods in healthcare. After a thorough review, we find that the vast majority are not appropriate for clinical use due to their methodological flaws and/or underlying biases which include but are not limited to privacy concerns, generalization issues, and communication costs. As a result, the effectiveness of federated learning in healthcare is significantly compromised. To overcome these challenges, we provide recommendations and promising opportunities that might be implemented to resolve these problems and improve the quality of model development in federated learning with healthcare.
Authors: Mohsen Ghaffari, Mahsa Varshosaz, Einar Broch Johnsen, Andrzej W\k{a}sowski
Abstract: Tabular reinforcement learning methods cannot operate directly on continuous state spaces. One solution for this problem is to partition the state space. A good partitioning enables generalization during learning and more efficient exploitation of prior experiences. Consequently, the learning process becomes faster and produces more reliable policies. However, partitioning introduces approximation, which is particularly harmful in the presence of nonlinear relations between state components. An ideal partition should be as coarse as possible, while capturing the key structure of the state space for the given problem. This work extracts partitions from the environment dynamics by symbolic execution. We show that symbolic partitioning improves state space coverage with respect to environmental behavior and allows reinforcement learning to perform better for sparse rewards. We evaluate symbolic state space partitioning with respect to precision, scalability, learning agent performance and state space coverage for the learnt policies.
Authors: Eric Mochiutti, Eric Aislan Antonelo, Eduardo Camponogara
Abstract: Echo State Networks (ESNs) are recurrent neural networks usually employed for modeling nonlinear dynamic systems with relatively ease of training. By incorporating physical laws into the training of ESNs, Physics-Informed ESNs (PI-ESNs) were proposed initially to model chaotic dynamic systems without external inputs. They require less data for training since Ordinary Differential Equations (ODEs) of the considered system help to regularize the ESN. In this work, the PI-ESN is extended with external inputs to model controllable nonlinear dynamic systems. Additionally, an existing self-adaptive balancing loss method is employed to balance the contributions of the residual regression term and the physics-informed loss term in the total loss function. The experiments with two nonlinear systems modeled by ODEs, the Van der Pol oscillator and the four-tank system, and with one differential-algebraic (DAE) system, an electric submersible pump, revealed that the proposed PI-ESN outperforms the conventional ESN, especially in scenarios with limited data availability, showing that PI-ESNs can regularize an ESN model with external inputs previously trained on just a few datapoints, reducing its overfitting and improving its generalization error (up to 92% relative reduction in the test error). Further experiments demonstrated that the proposed PI-ESN is robust to parametric uncertainties in the ODE equations and that model predictive control using PI-ESN outperforms the one using plain ESN, particularly when training data is scarce.
Authors: Ethan Blaser, Shangtong Zhang
Abstract: Stochastic approximation is an important class of algorithms, and a large body of previous analysis focuses on stochastic approximations driven by contractive operators, which is not applicable in some important reinforcement learning settings. This work instead investigates stochastic approximations with merely nonexpansive operators. In particular, we study nonexpansive stochastic approximations with Markovian noise, providing both asymptotic and finite sample analysis. Key to our analysis are a few novel bounds of noise terms resulting from the Poisson equation. As an application, we prove, for the first time, that the classical tabular average reward temporal difference learning converges to a sample path dependent fixed point.
Authors: Ethan Ewer, Daewon Chae, Thomas Zeng, Jinkyu Kim, Kangwook Lee
Abstract: Next-token prediction is conventionally done using decoder-only Transformers with causal attention, as this approach allows for efficient reuse of keys and values. What if we were not compute-limited, should we still use decoder-only Transformers? In this work, we introduce Encoder-only Next Token Prediction (ENTP). We explore the differences between ENTP and decoder-only Transformers in expressive power and complexity, highlighting potential advantages of ENTP in settings with unbounded compute. We introduce the $\operatorname{Count3}$ task and show, both theoretically and experimentally, that while ENTP can perform this task easily, a decoder-only Transformer cannot. Finally, we empirically demonstrate the superior performance of ENTP across representative tasks where next-token prediction based Transformers can be evaluated, including addition, in-context learning, and language modeling.
Authors: Sergei Kholkin, Grigoriy Ksenofontov, David Li, Nikita Kornilov, Nikita Gushchin, Alexandra Suvorikova, Alexey Kroshnin, Evgeny Burnaev, Alexander Korotin
Abstract: The Iterative Markovian Fitting (IMF) procedure, which iteratively projects onto the space of Markov processes and their reciprocal class, successfully solves the Schr\"odinger Bridge problem. However, an efficient practical implementation requires a heuristic modification - alternating between fitting forward and backward time diffusion at each iteration. This modification is crucial for stabilizing training and achieving reliable results in applications such as unpaired domain translation. Our work reveals a close connection between the modified version of IMF and the Iterative Proportional Fitting (IPF) procedure - a foundational method for the Schr\"odinger Bridge problem, also known as Sinkhorn's algorithm. Specifically, we demonstrate that this heuristic modification of the IMF effectively integrates both IMF and IPF procedures. We refer to this combined approach as the Iterative Proportional Markovian Fitting (IPMF) procedure. Through theoretical and empirical analysis, we establish the convergence of IPMF procedure under various settings, contributing to developing a unified framework for solving Schr\"odinger Bridge problems.
Authors: Dengsheng Chen, Jie Hu, Xiaoming Wei, Enhua Wu
Abstract: Joint-embedding predictive architectures (JEPAs) have shown substantial promise in self-supervised representation learning, yet their application in generative modeling remains underexplored. Conversely, diffusion models have demonstrated significant efficacy in modeling arbitrary probability distributions. In this paper, we introduce Denoising with a Joint-Embedding Predictive Architecture (D-JEPA), pioneering the integration of JEPA within generative modeling. By recognizing JEPA as a form of masked image modeling, we reinterpret it as a generalized next-token prediction strategy, facilitating data generation in an auto-regressive manner. Furthermore, we incorporate diffusion loss to model the per-token probability distribution, enabling data generation in a continuous space. We also adapt flow matching loss as an alternative to diffusion loss, thereby enhancing the flexibility of D-JEPA. Empirically, with increased GFLOPs, D-JEPA consistently achieves lower FID scores with fewer training epochs, indicating its good scalability. Our base, large, and huge models outperform all previous generative models across all scales on ImageNet conditional generation benchmarks. Beyond image generation, D-JEPA is well-suited for other continuous data modeling, including video and audio.
Authors: Jaeyeon Kim, Sehyun Kwon, Joo Young Choi, Jongho Park, Jaewoong Cho, Jason D. Lee, Ernest K. Ryu
Abstract: In-context learning (ICL) describes a language model's ability to generate outputs based on a set of input demonstrations and a subsequent query. To understand this remarkable capability, researchers have studied simplified, stylized models. These studies have consistently observed long loss plateaus, during which models exhibit minimal improvement, followed by a sudden, rapid surge of learning. In this work, we reveal that training on multiple diverse ICL tasks simultaneously shortens the loss plateaus, making each task easier to learn. This finding is surprising as it contradicts the natural intuition that the combined complexity of multiple ICL tasks would lengthen the learning process, not shorten it. Our result suggests that the recent success in large-scale training of language models may be attributed not only to the richness of the data at scale but also to the easier optimization (training) induced by the diversity of natural language training data.
Authors: Alberto Huertas Celdr\'an, Chao Feng, Sabyasachi Banik, Gerome Bovet, Gregorio Martinez Perez, Burkhard Stiller
Abstract: Federated Learning (FL), introduced in 2016, was designed to enhance data privacy in collaborative model training environments. Among the FL paradigm, horizontal FL, where clients share the same set of features but different data samples, has been extensively studied in both centralized and decentralized settings. In contrast, Vertical Federated Learning (VFL), which is crucial in real-world decentralized scenarios where clients possess different, yet sensitive, data about the same entity, remains underexplored. Thus, this work introduces De-VertiFL, a novel solution for training models in a decentralized VFL setting. De-VertiFL contributes by introducing a new network architecture distribution, an innovative knowledge exchange scheme, and a distributed federated training process. Specifically, De-VertiFL enables the sharing of hidden layer outputs among federation clients, allowing participants to benefit from intermediate computations, thereby improving learning efficiency. De-VertiFL has been evaluated using a variety of well-known datasets, including both image and tabular data, across binary and multiclass classification tasks. The results demonstrate that De-VertiFL generally surpasses state-of-the-art methods in F1-score performance, while maintaining a decentralized and privacy-preserving framework.
Authors: Yingxu Wang, Mengzhu Wang, Siwei Liu, Nan Yin
Abstract: Spiking Graph Networks (SGNs) have garnered significant attraction from researchers and industry due to their ability to address energy consumption challenges in graph classification. However, SGNs are only effective for in-distribution data and cannot tackle out-of-distribution data. In this paper, we first propose the domain adaptation problem in SGNs, and introduce a novel framework named Degree-aware Spiking Graph Domain Adaptation for Classification (DeSGDA). The proposed DeSGDA addresses the spiking graph domain adaptation problem in three aspects: node degree-aware personalized spiking representation, adversarial feature distribution alignment, and pseudo-label distillation. First, we introduce the personalized spiking representation method for generating degree-dependent spiking signals. Specifically, the node degree determines the threshold of triggering a spike, allowing this personalized approach to capture more expressive information for classification. Then, we propose the graph feature distribution alignment module that is adversarially trained using membrane potential against a domain discriminator. Such an alignment module can efficiently maintain high performance and low energy consumption in the case of inconsistent distribution. Additionally, we extract consistent predictions across two spaces to create reliable pseudo-labels, effectively leveraging unlabeled data to enhance classification performance. Extensive experiments on benchmark datasets validate the superiority of the DeSGDA compared with competitive baselines.
Authors: Sagi Shaier, Francisco Pereira, Katharina von der Wense, Lawrence E Hunter, Matt Jones
Abstract: The evolution of biological neural systems has led to both modularity and sparse coding, which enables energy efficiency and robustness across the diversity of tasks in the lifespan. In contrast, standard neural networks rely on dense, non-specialized architectures, where all model parameters are simultaneously updated to learn multiple tasks, leading to interference. Current sparse neural network approaches aim to alleviate this issue but are hindered by limitations such as 1) trainable gating functions that cause representation collapse, 2) disjoint experts that result in redundant computation and slow learning, and 3) reliance on explicit input or task IDs that limit flexibility and scalability. In this paper we propose Conditionally Overlapping Mixture of ExperTs (COMET), a general deep learning method that addresses these challenges by inducing a modular, sparse architecture with an exponential number of overlapping experts. COMET replaces the trainable gating function used in Sparse Mixture of Experts with a fixed, biologically inspired random projection applied to individual input representations. This design causes the degree of expert overlap to depend on input similarity, so that similar inputs tend to share more parameters. This results in faster learning per update step and improved out-of-sample generalization. We demonstrate the effectiveness of COMET on a range of tasks, including image classification, language modeling, and regression, using several popular deep learning architectures.
Authors: Shao-Ting Chiu, Junyuan Hong, Ulisses Braga-Neto
Abstract: We introduce DeepSets Operator Networks (DeepOSets), an efficient, non-autoregressive neural network architecture for in-context learning of permutation-invariant operators. DeepOSets combines the operator learning capabilities of Deep Operator Networks (DeepONets) with the set learning capabilities of DeepSets. Here, we present the application of DeepOSets to the problem of learning supervised learning algorithms, which are continuous permutation-invariant operators. We show that DeepOSets are universal approximators for this class of operators. In an empirical comparison with a popular autoregressive (transformer-based) model for in-context learning of linear regression, DeepOSets reduced the number of model weights by several orders of magnitude and required a fraction of training and inference time, in addition to significantly outperforming the transformer model in noisy settings. We also demonstrate the multiple operator learning capabilities of DeepOSets with a polynomial regression experiment where the order of the polynomial is learned in-context from the prompt.
Authors: Yarden As, Bhavya Sukhija, Lenart Treven, Carmelo Sferrazza, Stelian Coros, Andreas Krause
Abstract: Reinforcement learning (RL) is ubiquitous in the development of modern AI systems. However, state-of-the-art RL agents require extensive, and potentially unsafe, interactions with their environments to learn effectively. These limitations confine RL agents to simulated environments, hindering their ability to learn directly in real-world settings. In this work, we present ActSafe, a novel model-based RL algorithm for safe and efficient exploration. ActSafe learns a well-calibrated probabilistic model of the system and plans optimistically w.r.t. the epistemic uncertainty about the unknown dynamics, while enforcing pessimism w.r.t. the safety constraints. Under regularity assumptions on the constraints and dynamics, we show that ActSafe guarantees safety during learning while also obtaining a near-optimal policy in finite time. In addition, we propose a practical variant of ActSafe that builds on latest model-based RL advancements and enables safe exploration even in high-dimensional settings such as visual control. We empirically show that ActSafe obtains state-of-the-art performance in difficult exploration tasks on standard safe deep RL benchmarks while ensuring safety during learning.
Authors: Mohammad Mozaffari, Amir Yazdanbakhsh, Maryam Mehri Dehnavi
Abstract: Conventional model compression techniques for LLMs address high memory consumption and slow inference challenges but typically require computationally expensive retraining to preserve accuracy. In contrast, one-shot compression methods eliminate retraining cost, but struggle to achieve accuracy comparable to dense models. This paper presents SLIM, a new one-shot compression framework that holistically integrates hardware-friendly quantization, sparsity, and low-rank approximation into a unified process. First, we formulate the quantization process using a probabilistic approach (SLIM-Quant) that enables us to apply uniform quantization. Then, we use an existing one-shot pruning method to apply semi-structured sparsity on top of the quantized weights. Finally, to compensate for the introduced aggregated quantization and sparsity error, we use a novel saliency function with unique invertible and additive features that enables us to mathematically compute the value of low-rank adapters. SLIM improves model accuracy by up to 5.66% (LLaMA-2-7B) for 2:4 sparsity with 4-bit weight quantization, outperforming prior methods. Models compressed with SLIM achieve up to 3.78x and 3.75x layer-wise speedup on Nvidia RTX3060 and A100 GPUs, respectively. We also propose an optional PEFT recipe that further improves accuracy by up to 1.66% (LLaMA-2-13B) compared to SLIM without fine-tuning
Authors: Tengfei Cheng, Yangdi Huang, Yunxuan Dong
Abstract: Tidal energy is one of the key components in increasing the penetration rate of renewable energy. The penetration of tidal energy in the electrical grid depends on the accuracy of tidal current speed forecasting. Modeling inaccuracies hinder forecast accuracy. Previous research has primarily used physical models to forecast tidal current speed. However, tidal current variations influenced by the orbital periods of celestial bodies make accurate physical modeling challenging. Researching the multi-periodicity of tides is crucial for accurately forecasting tidal current speed. In this article, we propose the Wavelet-Enhanced Convolutional Network (WCN) to learn multi-periodicity. The framework embeds intra-period and inter-period variations of one-dimensional tidal current data into the rows and columns of a two-dimensional tensor. Then, the two-dimensional variations of the sequence can be processed by convolutional kernels. We integrate a time-frequency analysis method into the framework to further address local periodic features. Additionally, to enhance the framework's stability, we optimize the framework's hyperparameters with the Tree-structured Parzen Estimator algorithm. The proposed framework avoids the lack of learning multi-periodicity. Compared with benchmarks, the proposed framework reduces the mean absolute error and mean square error in 10-step forecasting by, at most, 90.36% and 97.56%, respectively.
Authors: Zixuan Yang, Jiaqi Zheng, Guihai Chen
Abstract: Ensuring verifiable and interpretable safety of deep reinforcement learning (DRL) is crucial for its deployment in real-world applications. Existing approaches like verification-in-the-loop training, however, face challenges such as difficulty in deployment, inefficient training, lack of interpretability, and suboptimal performance in property satisfaction and reward performance. In this work, we propose a novel verification-driven interpretation-in-the-loop framework Reintrainer to develop trustworthy DRL models, which are guaranteed to meet the expected constraint properties. Specifically, in each iteration, this framework measures the gap between the on-training model and predefined properties using formal verification, interprets the contribution of each input feature to the model's output, and then generates the training strategy derived from the on-the-fly measure results, until all predefined properties are proven. Additionally, the low reusability of existing verifiers and interpreters motivates us to develop Reinfier, a general and fundamental tool within Reintrainer for DRL verification and interpretation. Reinfier features breakpoints searching and verification-driven interpretation, associated with a concise constraint-encoding language DRLP. Evaluations demonstrate that Reintrainer outperforms the state-of-the-art on six public benchmarks in both performance and property guarantees. Our framework can be accessed at https://github.com/Kurayuri/Reinfier.
Authors: Daehwan Kim, Haejun Chung, Ikbeom Jang
Abstract: Recent studies have shown that deep neural networks are not well-calibrated and often produce over-confident predictions. The miscalibration issue primarily stems from using cross-entropy in classifications, which aims to align predicted softmax probabilities with one-hot labels. In ordinal regression tasks, this problem is compounded by an additional challenge: the expectation that softmax probabilities should exhibit unimodal distribution is not met with cross-entropy. The ordinal regression literature has focused on learning orders and overlooked calibration. To address both issues, we propose a novel loss function that introduces ordinal-aware calibration, ensuring that prediction confidence adheres to ordinal relationships between classes. It incorporates soft ordinal encoding and ordinal-aware regularization to enforce both calibration and unimodality. Extensive experiments across four popular ordinal regression benchmarks demonstrate that our approach achieves state-of-the-art calibration without compromising classification accuracy.
Authors: Gathika Ratnayaka, James Nichols, Qing Wang
Abstract: Partial graph matching extends traditional graph matching by allowing some nodes to remain unmatched, enabling applications in more complex scenarios. However, this flexibility introduces additional complexity, as both the subset of nodes to match and the optimal mapping must be determined. While recent studies have explored deep learning techniques for partial graph matching, a significant limitation remains: the absence of an optimization objective that fully captures the problem's intrinsic nature while enabling efficient solutions. In this paper, we propose a novel optimization framework for partial graph matching, inspired by optimal partial transport. Our approach formulates an objective that enables partial assignments while incorporating matching biases, using weighted total variation as the divergence function to guarantee optimal partial assignments. Our method can achieve efficient, exact solutions within cubic worst case time complexity. Our contributions are threefold: (i) we introduce a novel optimization objective that balances matched and unmatched nodes; (ii) we establish a connection between partial graph matching and linear sum assignment problem, enabling efficient solutions; (iii) we propose a deep graph matching architecture with a novel partial matching loss, providing an end-to-end solution. The empirical evaluations on standard graph matching benchmarks demonstrate the efficacy of the proposed approach.
Authors: Xi Zhang, Yuan Pu, Yuki Kawamura, Andrew Loza, Yoshua Bengio, Dennis L. Shung, Alexander Tong
Abstract: Modeling stochastic and irregularly sampled time series is a challenging problem found in a wide range of applications, especially in medicine. Neural stochastic differential equations (Neural SDEs) are an attractive modeling technique for this problem, which parameterize the drift and diffusion terms of an SDE with neural networks. However, current algorithms for training Neural SDEs require backpropagation through the SDE dynamics, greatly limiting their scalability and stability. To address this, we propose Trajectory Flow Matching (TFM), which trains a Neural SDE in a simulation-free manner, bypassing backpropagation through the dynamics. TFM leverages the flow matching technique from generative modeling to model time series. In this work we first establish necessary conditions for TFM to learn time series data. Next, we present a reparameterization trick which improves training stability. Finally, we adapt TFM to the clinical time series setting, demonstrating improved performance on three clinical time series datasets both in terms of absolute performance and uncertainty prediction.
Authors: Bohan Lyu, Yadi Cao, Duncan Watson-Parris, Leon Bergen, Taylor Berg-Kirkpatrick, Rose Yu
Abstract: Large Language Models (LLMs) demonstrate promising capabilities in solving simple scientific problems but, even with domain-specific fine-tuning, often produce hallucinations for complex ones. While integrating LLMs with tools can mitigate this reliability issue, models finetuned on tool usage only often over-rely on them, incurring unnecessary costs from resource-intensive scientific tools even for simpler problems. Inspired by how human experts assess the complexity of the problem before choosing the solutions, we propose a novel two-component fine-tuning method, Adapting While Learning (AWL). In the first component, World Knowledge Learning (WKL), LLMs internalize scientific knowledge by learning from tools-generated solutions. In the second component, Tool Usage Adaptation (TUA), we classify questions as easy or hard based on the WKL-trained model's accuracy, and train it to maintain direct reasoning for simple problems while switching to tools for challenging ones. We validate our method on 6 scientific benchmark datasets in climate science, epidemiology, and mathematics. Compared to the base 8B model, our trained models achieve 28.27% higher answer accuracy and 13.76% better tool usage accuracy, even surpassing state-of-the-art models including GPT-4 and Claude-3.5 on 4 custom-created datasets.
Authors: Nikola Milosevic, Johannes M\"uller, Nico Scherf
Abstract: Reinforcement Learning (RL) agents can solve diverse tasks but often exhibit unsafe behavior. Constrained Markov Decision Processes (CMDPs) address this by enforcing safety constraints, yet existing methods either sacrifice reward maximization or allow unsafe training. We introduce Constrained Trust Region Policy Optimization (C-TRPO), which reshapes the policy space geometry to ensure trust regions contain only safe policies, guaranteeing constraint satisfaction throughout training. We analyze its theoretical properties and connections to TRPO, Natural Policy Gradient (NPG), and Constrained Policy Optimization (CPO). Experiments show that C-TRPO reduces constraint violations while maintaining competitive returns.
Authors: Yang Hu, Xiao Wang, Zezhen Ding, Lirong Wu, Huatian Zhang, Stan Z. Li, Sheng Wang, Jiheng Zhang, Ziyun Li, Tianlong Chen
Abstract: Diffusion-based models have significant achievements in time series generation but suffer from inefficient computation: solving high-dimensional ODEs/SDEs via iterative numerical solvers demands hundreds to thousands of drift function evaluations per sample, incurring prohibitive costs. To resolve this, we propose FlowTS, an ODE-based model that leverages rectified flow with straight-line transport in probability space. By learning geodesic paths between distributions, FlowTS achieves computational efficiency through exact linear trajectory simulation, accelerating training and generation while improving performances. We further introduce an adaptive sampling strategy inspired by the exploration-exploitation trade-off, balancing noise adaptation and precision. Notably, FlowTS enables seamless adaptation from unconditional to conditional generation without retraining, ensuring efficient real-world deployment. Also, to enhance generation authenticity, FlowTS integrates trend and seasonality decomposition, attention registers (for global context aggregation), and Rotary Position Embedding (RoPE) (for position information). For unconditional setting, extensive experiments demonstrate that FlowTS achieves state-of-the-art performance, with context FID scores of 0.019 and 0.011 on Stock and ETTh datasets (prev. best: 0.067, 0.061). For conditional setting, we have achieved superior performance in solar forecasting (MSE 213, prev. best: 375) and MuJoCo imputation tasks (MSE 7e-5, prev. best 2.7e-4). The code is available at https://github.com/UNITES-Lab/FlowTS.
Authors: Vinay Kumar Sankarapu, Chintan Chitroda, Yashwardhan Rathore, Neeraj Kumar Singh, Pratinav Seth
Abstract: The rapid growth of AI has led to more complex deep learning models, often operating as opaque "black boxes" with limited transparency in their decision-making. This lack of interpretability poses challenges, especially in high-stakes applications where understanding model output is crucial. This work highlights the importance of interpretability in fostering trust, accountability, and responsible deployment. To address these challenges, we introduce DLBacktrace, a novel, model-agnostic technique designed to provide clear insights into deep learning model decisions across a wide range of domains and architectures, including MLPs, CNNs, and Transformer-based LLM models. We present a comprehensive overview of DLBacktrace and benchmark its performance against established interpretability methods such as SHAP, LIME, and GradCAM. Our results demonstrate that DLBacktrace effectively enhances understanding of model behavior across diverse tasks. DLBacktrace is compatible with models developed in both PyTorch and TensorFlow, supporting architectures such as BERT, ResNet, U-Net, and custom DNNs for tabular data. The library is open-sourced and available at https://github.com/AryaXAI/DLBacktrace .
Authors: Jie Zhang, Christian Schlarmann, Kristina Nikoli\'c, Nicholas Carlini, Francesco Croce, Matthias Hein, Florian Tram\`er
Abstract: Ensemble everything everywhere is a defense to adversarial examples that was recently proposed to make image classifiers robust. This defense works by ensembling a model's intermediate representations at multiple noisy image resolutions, producing a single robust classification. This defense was shown to be effective against multiple state-of-the-art attacks. Perhaps even more convincingly, it was shown that the model's gradients are perceptually aligned: attacks against the model produce noise that perceptually resembles the targeted class. In this short note, we show that this defense is not robust to adversarial attack. We first show that the defense's randomness and ensembling method cause severe gradient masking. We then use standard adaptive attack techniques to reduce the defense's robust accuracy from 48% to 14% on CIFAR-100 and from 62% to 11% on CIFAR-10, under the $\ell_\infty$-norm threat model with $\varepsilon=8/255$.
Authors: Tae-Hoon Lee, Min-Soo Kim
Abstract: Mixed-integer linear programming (MILP) is a widely used optimization technique across various fields. Existing $\textit{end-to-end learning}$ methods for MILP generate values for a subset of decision variables and delegate the remaining problem to traditional MILP solvers. However, this approach often fails to guarantee solution feasibility (i.e., satisfying all constraints) due to inaccurate predictions and primarily focuses on binary decision variables. Satisfying all constraints is a prerequisite for obtaining the optimal solution, and the feasibility issue becomes even more critical with non-binary integer (integer, for short) variables. Thus, addressing the feasibility of MILP involving integer variables is crucial. To address these challenges, we propose a novel reinforcement learning (RL)-based solver that not only finds the first feasible solution but also incrementally discovers better feasible solutions without delegating the remainder to off-the-shelf solvers. Our experimental results demonstrate that the proposed method achieves (near-)optimal solutions.
Authors: Jacob Adkins, Michael Bowling, Adam White
Abstract: The performance of modern reinforcement learning algorithms critically relies on tuning ever-increasing numbers of hyperparameters. Often, small changes in a hyperparameter can lead to drastic changes in performance, and different environments require very different hyperparameter settings to achieve state-of-the-art performance reported in the literature. We currently lack a scalable and widely accepted approach to characterizing these complex interactions. This work proposes a new empirical methodology for studying, comparing, and quantifying the sensitivity of an algorithm's performance to hyperparameter tuning for a given set of environments. We then demonstrate the utility of this methodology by assessing the hyperparameter sensitivity of several commonly used normalization variants of PPO. The results suggest that several algorithmic performance improvements may, in fact, be a result of an increased reliance on hyperparameter tuning.
Authors: Kristof Meding, Christoph Sorge
Abstract: When does a digital image resemble reality? The relevance of this question increases as the generation of synthetic images -- so called deep fakes -- becomes increasingly popular. Deep fakes have gained much attention for a number of reasons -- among others, due to their potential to disrupt the political climate. In order to mitigate these threats, the EU AI Act implements specific transparency regulations for generating synthetic content or manipulating existing content. However, the distinction between real and synthetic images is -- even from a computer vision perspective -- far from trivial. We argue that the current definition of deep fakes in the AI act and the corresponding obligations are not sufficiently specified to tackle the challenges posed by deep fakes. By analyzing the life cycle of a digital photo from the camera sensor to the digital editing features, we find that: (1.) Deep fakes are ill-defined in the EU AI Act. The definition leaves too much scope for what a deep fake is. (2.) It is unclear how editing functions like Google's ``best take'' feature can be considered as an exception to transparency obligations. (3.) The exception for substantially edited images raises questions about what constitutes substantial editing of content and whether or not this editing must be perceptible by a natural person. Our results demonstrate that complying with the current AI Act transparency obligations is difficult for providers and deployers. As a consequence of the unclear provisions, there is a risk that exceptions may be either too broad or too limited. We intend our analysis to foster the discussion on what constitutes a deep fake and to raise awareness about the pitfalls in the current AI Act transparency obligations.
Authors: Bikram Khanal, Pablo Rivas
Abstract: Quantum machine learning offers a transformative approach to solving complex problems, but the inherent noise hinders its practical implementation in near-term quantum devices. This obstacle makes it difficult to understand the generalizability of quantum circuit models. Designing robust quantum machine learning models under noise requires a principled understanding of complexity and generalization, extending beyond classical capacity measures. This study investigates the generalization properties of parameterized quantum machine learning models under the influence of noise. We present a data-dependent generalization bound grounded in the quantum Fisher information matrix. We leverage statistical learning theory to relate the parameter space volumes and training sizes to estimate the generalization capability of the trained model. We provide a structured characterization of complexity in quantum models by integrating local parameter neighborhoods and effective dimensions defined through quantum Fisher information matrix eigenvalues. We also analyze the tightness of the bound and discuss the tradeoff between model expressiveness and generalization performance.
Authors: Dipam Goswami, Simone Magistri, Kai Wang, Bart{\l}omiej Twardowski, Andrew D. Bagdanov, Joost van de Weijer
Abstract: Using pre-trained models has been found to reduce the effect of data heterogeneity and speed up federated learning algorithms. Recent works have investigated the use of first-order statistics and second-order statistics to aggregate local client data distributions at the server and achieve very high performance without any training. In this work we propose a training-free method based on an unbiased estimator of class covariance matrices. Our method, which only uses first-order statistics in the form of class means communicated by clients to the server, incurs only a fraction of the communication costs required by methods based on communicating second-order statistics. We show how these estimated class covariances can be used to initialize a linear classifier, thus exploiting the covariances without actually sharing them. When compared to state-of-the-art methods which also share only class means, our approach improves performance in the range of 4-26\% with exactly the same communication cost. Moreover, our method achieves performance competitive or superior to sharing second-order statistics with dramatically less communication overhead. Finally, using our method to initialize classifiers and then performing federated fine-tuning yields better and faster convergence. Code is available at https://github.com/dipamgoswami/FedCOF.
Authors: Utkarsh Saxena, Sayeh Sharify, Kaushik Roy, Xin Wang
Abstract: Post-training quantization (PTQ) of large language models (LLMs) holds the promise in reducing the prohibitive computational cost at inference time. Quantization of all weight, activation and key-value (KV) cache tensors to 4-bit without significantly degrading generalizability is challenging, due to the high quantization error caused by extreme outliers in activations. To tackle this problem, we propose ResQ, a PTQ method that pushes further the state-of-the-art. By means of principal component analysis (PCA), it identifies a low-rank subspace (in practice 1/8 of the hidden dimension) in which activation variances are highest, and keep the coefficients within this subspace in high precision, e.g. 8-bit, while quantizing the rest to 4-bit. Within each subspace, invariant random rotation is applied to further suppress outliers. We show that this is a provably optimal mixed precision quantization scheme that minimizes error. With the Llama and Qwen2.5 families of models, we demonstrate that ResQ outperforms recent uniform and mixed precision PTQ methods on a variety of benchmarks, achieving up to 33\% lower perplexity on Wikitext than the next best method SpinQuant, and upto 3\times speedup over 16-bit baseline. Code is available at https://github.com/utkarsh-dmx/project-resq.
Authors: Arseniy Andreyev, Pierfrancesco Beneventano
Abstract: Recent findings by Cohen et al., 2021, demonstrate that when training neural networks with full-batch gradient descent with a step size of $\eta$, the largest eigenvalue $\lambda_{\max}$ of the full-batch Hessian consistently stabilizes at $\lambda_{\max} = 2/\eta$. These results have significant implications for convergence and generalization. This, however, is not the case of mini-batch stochastic gradient descent (SGD), limiting the broader applicability of its consequences. We show that SGD trains in a different regime we term Edge of Stochastic Stability (EoSS). In this regime, what stabilizes at $2/\eta$ is *Batch Sharpness*: the expected directional curvature of mini-batch Hessians along their corresponding stochastic gradients. As a consequence $\lambda_{\max}$--which is generally smaller than Batch Sharpness--is suppressed, aligning with the long-standing empirical observation that smaller batches and larger step sizes favor flatter minima. We further discuss implications for mathematical modeling of SGD trajectories.
Authors: Yikang Gui, Prashant Doshi
Abstract: Inverse reinforcement learning (IRL) has progressed significantly toward accurately learning the underlying rewards in both discrete and continuous domains from behavior data. The next advance is to learn {\em intrinsic} preferences in ways that produce useful behavior in settings or tasks which are different but aligned with the observed ones. In the context of robotic applications, this helps integrate robots into processing lines involving new tasks (with shared intrinsic preferences) without programming from scratch. We introduce a method to inversely learn an abstract reward function from behavior trajectories in two or more differing instances of a domain. The abstract reward function is then used to learn task behavior in another separate instance of the domain. This step offers evidence of its transferability and validates its correctness. We evaluate the method on trajectories in tasks from multiple domains in OpenAI's Gym testbed and AssistiveGym and show that the learned abstract reward functions can successfully learn task behaviors in instances of the respective domains, which have not been seen previously.
Authors: Xiaoye Miao, Bin Li, Yanzhang, Xinkui Zhao, Yangyang Wu
Abstract: Privacy concerns arise as sensitive data proliferate. Despite decentralized federated learning (DFL) aggregating gradients from neighbors to avoid direct data transmission, it still poses indirect data leaks from the transmitted gradients. Existing privacy-preserving methods for DFL add noise to gradients. They either diminish the model predictive accuracy or suffer from ineffective gradient protection. In this paper, we propose a novel lossless privacy-preserving aggregation rule named LPPA to enhance gradient protection as much as possible but without loss of DFL model predictive accuracy. LPPA subtly injects the noise difference between the sent and received noise into transmitted gradients for gradient protection. The noise difference incorporates neighbors' randomness for each client, effectively safeguarding against data leaks. LPPA employs the noise flow conservation theory to ensure that the noise impact can be globally eliminated. The global sum of all noise differences remains zero, ensuring that accurate gradient aggregation is unaffected and the model accuracy remains intact. We theoretically prove that the privacy-preserving capacity of LPPA is \sqrt{2} times greater than that of noise addition, while maintaining comparable model accuracy to the standard DFL aggregation without noise injection. Experimental results verify the theoretical findings and show that LPPA achieves a 14% mean improvement in accuracy over noise addition. We also demonstrate the effectiveness of LPPA in protecting raw data and guaranteeing lossless model accuracy.
Authors: Willian T. Lunardi, Abdulrahman Banabila, Dania Herzalla, Martin Andreoni
Abstract: Distance-based anomaly detection methods rely on compact and separable in-distribution (ID) embeddings to effectively delineate anomaly boundaries. Single-positive contrastive formulations suffer from class collision, promoting unnecessary intra-class variance within ID samples. While multi-positive formulations can improve inlier compactness, they fail to preserve the diversity among synthetic outliers. We address these limitations by proposing a contrastive pretext task for anomaly detection that enforces three key properties: (1) compact ID clustering to reduce intra-class variance, (2) inlier-outlier separation to enhance inter-class separation, and (3) outlier-outlier separation to maintain diversity among synthetic outliers and prevent representation collapse. These properties work together to ensure a more robust and discriminative feature space for anomaly detection. Our approach achieves approximately 12x faster convergence than NT-Xent and 7x faster than Rot-SupCon, with superior performance. On CIFAR-10, it delivers an average performance boost of 6.2% over NT-Xent and 2% over Rot-SupCon, with class-specific improvements of up to 16.9%. Our code is available at https://anonymous.4open.science/r/firm-98B6.
Authors: Olabode T. Ajenifujah, Amir Barati Farimani
Abstract: Additive manufacturing (AM) is a rapidly evolving technology that has attracted applications across a wide range of fields due to its ability to fabricate complex geometries. However, one of the key challenges in AM is achieving consistent print quality. This inconsistency is often attributed to uncontrolled melt pool dynamics, partly caused by spatter which can lead to defects. Therefore, capturing and controlling the evolution of the melt pool is crucial for enhancing process stability and part quality. In this study, we developed a framework to support decision-making towards efficient AM process operations, capable of facilitating quality control and minimizing defects via machine learning (ML) and polynomial symbolic regression models. We implemented experimentally validated computational tools, specifically for laser powder bed fusion (LPBF) processes as a cost-effective approach to collect large datasets. For a dataset consisting of 281 varying process conditions, parameters such as melt pool dimensions (length, width, depth), melt pool geometry (area, volume), and volume indicated as spatter were extracted. Using machine learning (ML) and polynomial symbolic regression models, a high R2 of over 95 % was achieved in predicting the melt pool dimensions and geometry features on both the training and testing datasets, with either process conditions (power and velocity) or melt pool dimensions as the model inputs. In the case of volume indicated as spatter the value of the R2 improved after logarithmic transforming the model inputs, which were either the process conditions or the melt pool dimensions. Among the investigated ML models, the ExtraTree model achieved the highest R2 values of 96.7 % and 87.5 %.
Authors: Zihan Qiu, Zeyu Huang, Bo Zheng, Kaiyue Wen, Zekun Wang, Rui Men, Ivan Titov, Dayiheng Liu, Jingren Zhou, Junyang Lin
Abstract: This paper revisits the implementation of $\textbf{L}$oad-$\textbf{b}$alancing $\textbf{L}$oss (LBL) when training Mixture-of-Experts (MoEs) models. Specifically, LBL for MoEs is defined as $N_E \sum_{i=1}^{N_E} f_i p_i$, where $N_E$ is the total number of experts, $f_i$ represents the frequency of expert $i$ being selected, and $p_i$ denotes the average gating score of the expert $i$. Existing MoE training frameworks usually employ the parallel training strategy so that $f_i$ and the LBL are calculated within a $\textbf{micro-batch}$ and then averaged across parallel groups. In essence, a micro-batch for training billion-scale LLMs normally contains very few sequences. So, the micro-batch LBL is almost at the sequence level, and the router is pushed to distribute the token evenly within each sequence. Under this strict constraint, even tokens from a domain-specific sequence ($\textit{e.g.}$, code) are uniformly routed to all experts, thereby inhibiting expert specialization. In this work, we propose calculating LBL using a $\textbf{global-batch}$ to loose this constraint. Because a global-batch contains much more diverse sequences than a micro-batch, which will encourage load balance at the corpus level. Specifically, we introduce an extra communication step to synchronize $f_i$ across micro-batches and then use it to calculate the LBL. Through experiments on training MoEs-based LLMs (up to $\textbf{42.8B}$ total parameters and $\textbf{400B}$ tokens), we surprisingly find that the global-batch LBL strategy yields excellent performance gains in both pre-training perplexity and downstream tasks. Our analysis reveals that the global-batch LBL also greatly improves the domain specialization of MoE experts.
Authors: Jiaqi Yang, Wei Xie, Xiaohua Xu
Abstract: Quantum neural networks (QNNs) play an important role as an emerging technology in the rapidly growing field of quantum machine learning. While their empirical success is evident, the theoretical explorations of QNNs, particularly their generalization properties, are less developed and primarily focus on the uniform convergence approach. In this paper, we exploit an advanced tool in classical learning theory, i.e., algorithmic stability, to study the generalization of QNNs. We first establish high-probability generalization bounds for QNNs via uniform stability. Our bounds shed light on the key factors influencing the generalization performance of QNNs and provide practical insights into both the design and training processes. We next explore the generalization of QNNs on near-term noisy intermediate-scale quantum (NISQ) devices, highlighting the potential benefits of quantum noise. Moreover, we argue that our previous analysis characterizes worst-case generalization guarantees, and we establish a refined optimization-dependent generalization bound for QNNs via on-average stability. Numerical experiments on various real-world datasets support our theoretical findings.
Authors: Zhihao Yao, Jixuan Yin, Bo Li
Abstract: Short text clustering has gained significant attention in the data mining community. However, the limited valuable information contained in short texts often leads to low-discriminative representations, increasing the difficulty of clustering. This paper proposes a novel short text clustering framework, called Reliable \textbf{P}seudo-labeling via \textbf{O}ptimal \textbf{T}ransport with \textbf{A}ttention for Short Text Clustering (\textbf{POTA}), that generate reliable pseudo-labels to aid discriminative representation learning for clustering. Specially, \textbf{POTA} first implements an instance-level attention mechanism to capture the semantic relationships among samples, which are then incorporated as a semantic consistency regularization term into an optimal transport problem. By solving this OT problem, we can yield reliable pseudo-labels that simultaneously account for sample-to-sample semantic consistency and sample-to-cluster global structure information. Additionally, the proposed OT can adaptively estimate cluster distributions, making \textbf{POTA} well-suited for varying degrees of imbalanced datasets. Then, we utilize the pseudo-labels to guide contrastive learning to generate discriminative representations and achieve efficient clustering. Extensive experiments demonstrate \textbf{POTA} outperforms state-of-the-art methods. The code is available at: \href{https://github.com/YZH0905/POTA-STC/tree/main}{https://github.com/YZH0905/POTA-STC/tree/main}.
URLs: https://github.com/YZH0905/POTA-STC/tree/main, https://github.com/YZH0905/POTA-STC/tree/main
Authors: Lin William Cong, Guanhao Feng, Jingyu He, Xin He
Abstract: We introduce a new class of tree-based models, P-Trees, for analyzing (unbalanced) panel of individual asset returns, generalizing high-dimensional sorting with economic guidance and interpretability. Under the mean-variance efficient framework, P-Trees construct test assets that significantly advance the efficient frontier compared to commonly used test assets, with alphas unexplained by benchmark pricing models. P-Tree tangency portfolios also constitute traded factors, recovering the pricing kernel and outperforming popular observable and latent factor models for investments and cross-sectional pricing. Finally, P-Trees capture the complexity of asset returns with sparsity, achieving out-of-sample Sharpe ratios close to those attained only by over-parameterized large models.
Authors: Kasper Green Larsen, Markus Engelund Mathiasen, Clement Svendsen
Abstract: We introduce a new replicable boosting algorithm which significantly improves the sample complexity compared to previous algorithms. The algorithm works by doing two layers of majority voting, using an improved version of the replicable boosting algorithm introduced by Impagliazzo et al. [2022] in the bottom layer.
Authors: Devansh Bhardwaj, Naman Mishra
Abstract: Fingerprinting refers to the process of identifying underlying Machine Learning (ML) models of AI Systemts, such as Large Language Models (LLMs), by analyzing their unique characteristics or patterns, much like a human fingerprint. The fingerprinting of Large Language Models (LLMs) has become essential for ensuring the security and transparency of AI-integrated applications. While existing methods primarily rely on access to direct interactions with the application to infer model identity, they often fail in real-world scenarios involving multi-agent systems, frequent model updates, and restricted access to model internals. In this paper, we introduce a novel fingerprinting framework designed to address these challenges by integrating static and dynamic fingerprinting techniques. Our approach identifies architectural features and behavioral traits, enabling accurate and robust fingerprinting of LLMs in dynamic environments. We also highlight new threat scenarios where traditional fingerprinting methods are ineffective, bridging the gap between theoretical techniques and practical application. To validate our framework, we present an extensive evaluation setup that simulates real-world conditions and demonstrate the effectiveness of our methods in identifying and monitoring LLMs in Gen-AI applications. Our results highlight the framework's adaptability to diverse and evolving deployment contexts.
Authors: Minh Le, Anh Nguyen, Huy Nguyen, Chau Nguyen, Nhat Ho
Abstract: Visual Prompt Tuning (VPT) has recently emerged as a powerful method for adapting pre-trained vision models to downstream tasks. By introducing learnable prompt tokens as task-specific instructions, VPT effectively guides pre-trained transformer models with minimal overhead. Despite its empirical success, a comprehensive theoretical understanding of VPT remains an active area of research. Building on recent insights into the connection between mixture of experts and prompt-based approaches, we identify a key limitation in VPT: the restricted functional expressiveness in prompt formulation. To address this limitation, we propose Visual Adaptive Prompt Tuning (VAPT), a new generation of prompts that redefines prompts as adaptive functions of the input. Our theoretical analysis shows that this simple yet intuitive approach achieves optimal sample efficiency. Empirical results on VTAB-1K and FGVC further demonstrate VAPT's effectiveness, with performance gains of 7.34% and 1.04% over fully fine-tuning baselines, respectively. Notably, VAPT also surpasses VPT by a substantial margin while using fewer parameters. These results highlight both the effectiveness and efficiency of our method and pave the way for future research to explore the potential of adaptive prompts.
Authors: Diyuan Wu, Marco Mondelli
Abstract: Neural Collapse is a phenomenon where the last-layer representations of a well-trained neural network converge to a highly structured geometry. In this paper, we focus on its first (and most basic) property, known as NC1: the within-class variability vanishes. While prior theoretical studies establish the occurrence of NC1 via the data-agnostic unconstrained features model, our work adopts a data-specific perspective, analyzing NC1 in a three-layer neural network, with the first two layers operating in the mean-field regime and followed by a linear layer. In particular, we establish a fundamental connection between NC1 and the loss landscape: we prove that points with small empirical loss and gradient norm (thus, close to being stationary) approximately satisfy NC1, and the closeness to NC1 is controlled by the residual loss and gradient norm. We then show that (i) gradient flow on the mean squared error converges to NC1 solutions with small empirical loss, and (ii) for well-separated data distributions, both NC1 and vanishing test loss are achieved simultaneously. This aligns with the empirical observation that NC1 emerges during training while models attain near-zero test error. Overall, our results demonstrate that NC1 arises from gradient training due to the properties of the loss landscape, and they show the co-occurrence of NC1 and small test error for certain data distributions.
Authors: Abhijeet Mulgund, Chirag Pabbaraju
Abstract: The paradigm of weak-to-strong generalization constitutes the training of a strong AI model on data labeled by a weak AI model, with the goal that the strong model nevertheless outperforms its weak supervisor on the target task of interest. For the setting of real-valued regression with the squared loss, recent work quantitatively characterizes the gain in performance of the strong model over the weak model in terms of the misfit between the strong and weak model. We generalize such a characterization to learning tasks whose loss functions correspond to arbitrary Bregman divergences when the strong class is convex. This extends the misfit-based characterization of performance gain in weak-to-strong generalization to classification tasks, as the cross-entropy loss can be expressed in terms of a Bregman divergence. In most practical scenarios, however, the strong model class may not be convex. We therefore weaken this assumption and study weak-to-strong generalization for convex combinations of $k$ strong models in the strong class, in the concrete setting of classification. This allows us to obtain a similar misfit-based characterization of performance gain, upto an additional error term that vanishes as $k$ gets large. Our theoretical findings are supported by thorough experiments on synthetic as well as real-world datasets.
Authors: Yuchun Miao, Sen Zhang, Liang Ding, Yuqi Zhang, Lefei Zhang, Dacheng Tao
Abstract: This work identifies the Energy Loss Phenomenon in Reinforcement Learning from Human Feedback (RLHF) and its connection to reward hacking. Specifically, energy loss in the final layer of a Large Language Model (LLM) gradually increases during the RL process, with an excessive increase in energy loss characterizing reward hacking. Beyond empirical analysis, we further provide a theoretical foundation by proving that, under mild conditions, the increased energy loss reduces the upper bound of contextual relevance in LLMs, which is a critical aspect of reward hacking as the reduced contextual relevance typically indicates overfitting to reward model-favored patterns in RL. To address this issue, we propose an Energy loss-aware PPO algorithm (EPPO) which penalizes the increase in energy loss in the LLM's final layer during reward calculation to prevent excessive energy loss, thereby mitigating reward hacking. We theoretically show that EPPO can be conceptually interpreted as an entropy-regularized RL algorithm, which provides deeper insights into its effectiveness. Extensive experiments across various LLMs and tasks demonstrate the commonality of the energy loss phenomenon, as well as the effectiveness of EPPO in mitigating reward hacking and improving RLHF performance.
Authors: Kefan Dong, Tengyu Ma
Abstract: A fundamental challenge in formal theorem proving by LLMs is the lack of high-quality training data. Although reinforcement learning or expert iteration partially mitigates this issue by alternating between LLM generating proofs and finetuning them on correctly generated ones, performance quickly plateaus due to the scarcity of correct proofs (sparse rewards). To keep improving the models with limited data, we draw inspiration from mathematicians, who continuously develop new results, partly by proposing novel conjectures or exercises (which are often variants of known results) and attempting to solve them. We design the Self-play Theorem Prover (STP) that simultaneously takes on two roles, conjecturer and prover, each providing training signals to the other. The conjecturer is trained iteratively on previously generated conjectures that are barely provable by the current prover, which incentivizes it to generate increasingly challenging conjectures over time. The prover attempts to prove the conjectures with standard expert iteration. We evaluate STP with both Lean and Isabelle formal versifiers. With 19.8 billion tokens generated during the training in Lean, STP proves 26.3% of the statements in the LeanWorkbook dataset, doubling the previous best result of 13.2% achieved through expert iteration. The final model achieves state-of-the-art performance among whole-proof generation methods on miniF2F-test (61.1%, pass@3200), Proofnet-test (23.1%, pass@3200) and PutnamBench (8/644, pass@64).
Authors: Alexander Moreno, Justin Xiao, Jonathan Mei
Abstract: Structured Kernel Interpolation (SKI) (Wilson et al. 2015) helps scale Gaussian Processes (GPs) by approximating the kernel matrix via interpolation at inducing points, achieving linear computational complexity. However, it lacks rigorous theoretical error analysis. This paper bridges the gap: we prove error bounds for the SKI Gram matrix and examine the error's effect on hyperparameter estimation and posterior inference. We further provide a practical guide to selecting the number of inducing points under convolutional cubic interpolation: they should grow as $n^{d/3}$ for error control. Crucially, we identify two dimensionality regimes governing the trade-off between SKI Gram matrix spectral norm error and computational complexity. For $d \leq 3$, any error tolerance can achieve linear time for sufficiently large sample size. For $d > 3$, the error must increase with sample size to maintain linear time. Our analysis provides key insights into SKI's scalability-accuracy trade-offs, establishing precise conditions for achieving linear-time GP inference with controlled approximation error.
Authors: Teng Xiao, Yige Yuan, Zhengyu Chen, Mingxiao Li, Shangsong Liang, Zhaochun Ren, Vasant G Honavar
Abstract: Existing preference optimization objectives for language model alignment require additional hyperparameters that must be extensively tuned to achieve optimal performance, increasing both the complexity and time required for fine-tuning large language models. In this paper, we propose a simple yet effective hyperparameter-free preference optimization algorithm for alignment. We observe that promising performance can be achieved simply by optimizing inverse perplexity, which is calculated as the inverse of the exponentiated average log-likelihood of the chosen and rejected responses in the preference dataset. The resulting simple learning objective, SimPER, is easy to implement and eliminates the need for expensive hyperparameter tuning and a reference model, making it both computationally and memory efficient. Extensive experiments on widely used real-world benchmarks, including MT-Bench, AlpacaEval 2, and 10 key benchmarks of the Open LLM Leaderboard with 5 base models, demonstrate that SimPER consistently and significantly outperforms existing approaches-even without any hyperparameters or a reference model . For example, despite its simplicity, SimPER outperforms state-of-the-art methods by up to 5.7 points on AlpacaEval 2 and achieves the highest average ranking across 10 benchmarks on the Open LLM Leaderboard. The source code for SimPER is publicly available at: https://github.com/tengxiao1/SimPER.
Authors: Peixuan Han, Cheng Qian, Xiusi Chen, Yuji Zhang, Denghui Zhang, Heng Ji
Abstract: Large language models (LLMs) have demonstrated exceptional capabilities across a wide range of tasks but also pose significant risks due to their potential to generate harmful content. Although existing safety mechanisms can improve model safety, they often lead to overly cautious behavior and fail to fully utilize LLMs' internal cognitive processes. Drawing inspiration from cognitive science, where humans rely on reflective reasoning (System 2 thinking) to regulate language and behavior, we empirically demonstrate that LLMs also possess a similar capacity for internal assessment and regulation, which can be actively detected. Building on this insight, we introduce SafeSwitch, a framework that dynamically regulates unsafe outputs by monitoring and utilizing the model's internal states. Our empirical results show that SafeSwitch reduces harmful outputs by over 80% on safety benchmarks while maintaining strong utility. Compared to traditional safety alignment methods, SafeSwitch delivers more informative and context-aware refusals, demonstrates resilience to unseen queries, and achieves these benefits while only tuning less than 6% of the original parameters. These features make SafeSwitch a promising approach for implementing nuanced safety controls in LLMs.
Authors: Kevin Chen, Marco Cusumano-Towner, Brody Huval, Aleksei Petrenko, Jackson Hamburger, Vladlen Koltun, Philipp Kr\"ahenb\"uhl
Abstract: Interactive digital agents (IDAs) leverage APIs of stateful digital environments to perform tasks in response to user requests. While IDAs powered by instruction-tuned large language models (LLMs) can react to feedback from interface invocations in multi-step exchanges, they have not been trained in their respective digital environments. Prior methods accomplish less than half of tasks in sophisticated benchmarks such as AppWorld. We present a reinforcement learning (RL) approach that trains IDAs directly in their target environments. We formalize this training as a partially observable Markov decision process and derive LOOP, a data- and memory-efficient variant of proximal policy optimization. LOOP uses no value network and maintains exactly one copy of the underlying LLM in memory, making its implementation straightforward and as memory-efficient as fine-tuning a single LLM. A 32-billion-parameter agent trained with LOOP in the AppWorld environment outperforms the much larger OpenAI o1 agent by 9 percentage points (15% relative). To our knowledge, this is the first reported application of RL to IDAs that interact with a stateful, multi-domain, multi-app environment via direct API calls. Our analysis sheds light on the effectiveness of RL in this area, showing that the agent learns to consult the API documentation, avoid unwarranted assumptions, minimize confabulation, and recover from setbacks.
Authors: Isha Puri, Shivchander Sudalairaj, Guangxuan Xu, Kai Xu, Akash Srivastava
Abstract: Large language models (LLMs) have achieved significant performance gains via scaling up model sizes and/or data. However, recent evidence suggests diminishing returns from such approaches, motivating scaling the computation spent at inference time. Existing inference-time scaling methods, usually with reward models, cast the task as a search problem, which tends to be vulnerable to reward hacking as a consequence of approximation errors in reward models. In this paper, we instead cast inference-time scaling as a probabilistic inference task and leverage sampling-based techniques to explore the typical set of the state distribution of a state-space model with an approximate likelihood, rather than optimize for its mode directly. We propose a novel inference-time scaling approach by adapting particle-based Monte Carlo methods to this task. Our empirical evaluation demonstrates that our methods have a 4-16x better scaling rate over our deterministic search counterparts on various challenging mathematical reasoning tasks. Using our approach, we show that Qwen2.5-Math-1.5B-Instruct can surpass GPT-4o accuracy in only 4 rollouts, while Qwen2.5-Math-7B-Instruct scales to o1 level accuracy in only 32 rollouts. Our work not only presents an effective method to inference-time scaling, but also connects the rich literature in probabilistic inference with inference-time scaling of LLMs to develop more robust algorithms in future work. Code and further information is available at https://probabilistic-inference-scaling.github.io.
Authors: Simon Briend, Luc Devroye, Gabor Lugosi
Abstract: A uniform $k$-{\sc dag} generalizes the uniform random recursive tree by picking $k$ parents uniformly at random from the existing nodes. It starts with $k$ ''roots''. Each of the $k$ roots is assigned a bit. These bits are propagated by a noisy channel. The parents' bits are flipped with probability $p$, and a majority vote is taken. When all nodes have received their bits, the $k$-{\sc dag} is shown without identifying the roots. The goal is to estimate the majority bit among the roots. We identify the threshold for $p$ as a function of $k$ below which the majority rule among all nodes yields an error $c+o(1)$ with $c<1/2$. Above the threshold the majority rule errs with probability $1/2+o(1)$.
Authors: Haoxian Chen, Henry Lam
Abstract: Bayesian Optimization is a popular approach for optimizing expensive black-box functions. Its key idea is to use a surrogate model to approximate the objective and, importantly, quantify the associated uncertainty that allows a sequential search of query points that balance exploitation-exploration. Gaussian process (GP) has been a primary candidate for the surrogate model, thanks to its Bayesian-principled uncertainty quantification power and modeling flexibility. However, its challenges have also spurred an array of alternatives whose convergence properties could be more opaque. Motivated by these, we study in this paper an axiomatic framework that elicits the minimal requirements to guarantee black-box optimization convergence that could apply beyond GP-based methods. Moreover, we leverage the design freedom in our framework, which we call Pseudo-Bayesian Optimization, to construct empirically superior algorithms. In particular, we show how using simple local regression, and a suitable "randomized prior" construction to quantify uncertainty, not only guarantees convergence but also consistently outperforms state-of-the-art benchmarks in examples ranging from high-dimensional synthetic experiments to realistic hyperparameter tuning and robotic applications.
Authors: Hao Wang, Shangwei Guo, Jialing He, Hangcheng Liu, Tianwei Zhang, Tao Xiang
Abstract: Pre-trained models (PTMs) are widely adopted across various downstream tasks in the machine learning supply chain. Adopting untrustworthy PTMs introduces significant security risks, where adversaries can poison the model supply chain by embedding hidden malicious behaviors (backdoors) into PTMs. However, existing backdoor attacks to PTMs can only achieve partially task-agnostic and the embedded backdoors are easily erased during the fine-tuning process. This makes it challenging for the backdoors to persist and propagate through the supply chain. In this paper, we propose a novel and severer backdoor attack, TransTroj, which enables the backdoors embedded in PTMs to efficiently transfer in the model supply chain. In particular, we first formalize this attack as an indistinguishability problem between poisoned and clean samples in the embedding space. We decompose embedding indistinguishability into pre- and post-indistinguishability, representing the similarity of the poisoned and reference embeddings before and after the attack. Then, we propose a two-stage optimization that separately optimizes triggers and victim PTMs to achieve embedding indistinguishability. We evaluate TransTroj on four PTMs and six downstream tasks. Experimental results show that our method significantly outperforms SOTA task-agnostic backdoor attacks -- achieving nearly 100% attack success rate on most downstream tasks -- and demonstrates robustness under various system settings. Our findings underscore the urgent need to secure the model supply chain against such transferable backdoor attacks. The code is available at https://github.com/haowang-cqu/TransTroj .
Authors: Andrei A. Klishin, Joseph Bakarji, J. Nathan Kutz, Krithika Manohar
Abstract: Recovering dynamical equations from observed noisy data is the central challenge of system identification. We develop a statistical mechanics approach to analyze sparse equation discovery algorithms, which typically balance data fit and parsimony via hyperparameter tuning. In this framework, statistical mechanics offers tools to analyze the interplay between complexity and fitness similarly to that of entropy and energy in physical systems. To establish this analogy, we define the hyperparameter optimization procedure as a two-level Bayesian inference problem that separates variable selection from coefficient inference and enables the computation of the posterior parameter distribution in closed form. Our approach provides uncertainty quantification, crucial in the low-data limit that is frequently encountered in real-world applications. A key advantage of employing statistical mechanical concepts, such as free energy and the partition function, is to connect the large data limit to thermodynamic limit and characterize the sparsity- and noise-induced phase transitions that delineate correct from incorrect identification. We thus provide a method for closed-loop inference, estimating the noise in a given model and checking if the model is tolerant to that noise amount. This perspective of sparse equation discovery is versatile and can be adapted to various other equation discovery algorithms.
Authors: Guoxiong Gao, Haocheng Ju, Jiedong Jiang, Zihan Qin, Bin Dong
Abstract: The interactive theorem prover Lean enables the verification of formal mathematical proofs and is backed by an expanding community. Central to this ecosystem is its mathematical library, mathlib4, which lays the groundwork for the formalization of an expanding range of mathematical theories. However, searching for theorems in mathlib4 can be challenging. To successfully search in mathlib4, users often need to be familiar with its naming conventions or documentation strings. Therefore, creating a semantic search engine that can be used easily by individuals with varying familiarity with mathlib4 is very important. In this paper, we present a semantic search engine (https://leansearch.net/) for mathlib4 that accepts informal queries and finds the relevant theorems. We also establish a benchmark for assessing the performance of various search engines for mathlib4.
URLs: https://leansearch.net/)
Authors: Johannes M\"uller, Semih \c{C}ayc{\i}, Guido Mont\'ufar
Abstract: Kakade's natural policy gradient method has been studied extensively in recent years, showing linear convergence with and without regularization. We study another natural gradient method based on the Fisher information matrix of the state-action distributions which has received little attention from the theoretical side. Here, the state-action distributions follow the Fisher-Rao gradient flow inside the state-action polytope with respect to a linear potential. Therefore, we study Fisher-Rao gradient flows of linear programs more generally and show linear convergence with a rate that depends on the geometry of the linear program. Equivalently, this yields an estimate on the error induced by entropic regularization of the linear program which improves existing results. We extend these results and show sublinear convergence for perturbed Fisher-Rao gradient flows and natural gradient flows up to an approximation error. In particular, these general results cover the case of state-action natural policy gradients.
Authors: Zihan Guan, Mengxuan Hu, Sheng Li, Anil Vullikanti
Abstract: Diffusion models are vulnerable to backdoor attacks, where malicious attackers inject backdoors by poisoning certain training samples during the training stage. This poses a significant threat to real-world applications in the Model-as-a-Service (MaaS) scenario, where users query diffusion models through APIs or directly download them from the internet. To mitigate the threat of backdoor attacks under MaaS, black-box input-level backdoor detection has drawn recent interest, where defenders aim to build a firewall that filters out backdoor samples in the inference stage, with access only to input queries and the generated results from diffusion models. Despite some preliminary explorations on the traditional classification tasks, these methods cannot be directly applied to the generative tasks due to two major challenges: (1) more diverse failures and (2) a multi-modality attack surface. In this paper, we propose a black-box input-level backdoor detection framework on diffusion models, called UFID. Our defense is motivated by an insightful causal analysis: Backdoor attacks serve as the confounder, introducing a spurious path from input to target images, which remains consistent even when we perturb the input samples with Gaussian noise. We further validate the intuition with theoretical analysis. Extensive experiments across different datasets on both conditional and unconditional diffusion models show that our method achieves superb performance on detection effectiveness and run-time efficiency.
Authors: Yimu Wang, Shuai Yuan, Bo Xue, Xiangru Jian, Wei Pang, Mushi Wang, Ning Yu
Abstract: Recent progress in video-text retrieval has been driven largely by advancements in model architectures and training strategies. However, the representation learning capabilities of videotext retrieval models remain constrained by lowquality and limited training data annotations. To address this issue, we present a novel ViDeoText Retrieval Paradigm with RElevance-based AugMentation, namely DREAM, which enhances video and text data using large foundation models to learn more generalized features. Specifically, we first adopt a simple augmentation method, which generates self-similar data by randomly duplicating or dropping subwords and frames. In addition, inspired by the recent advancement in visual and language generative models, we propose a more robust augmentation method through textual paraphrasing and video stylization using large language models (LLMs) and visual generative models (VGMs). To further enrich video and text information, we propose a relevance-based augmentation method, where LLMs and VGMs generate and integrate new relevant information into the original data. Leveraging this enriched data, extensive experiments on several video-text retrieval benchmarks demonstrate the superiority of DREAM over existing methods.
Authors: Hao Yang, Ayberk Acar, Keshuai Xu, Anton Deguet, Peter Kazanzides, Jie Ying Wu
Abstract: Robot-assisted minimally invasive surgery, such as through the da Vinci systems, improves precision and patient outcomes. However, da Vinci systems prior to da Vinci 5, lacked direct force-sensing capabilities, forcing surgeons to operate without the haptic feedback they get through laparoscopy. Our prior work restored force sensing through machine learning-based force estimation for the da Vinci Research Kit (dVRK) Classic. This study extends our previous method to the newer dVRK system, the dVRK-Si. Additionally, we benchmark the performance of the learning-based algorithm against baseline methods (which make simplifying assumptions on the torque) to study how the two systems differ. Results show the learning-based method achieves an average root-mean-square-error (RMSE) of 5.21\%, for the dVRK-Si, which is comparable to the dVRK Classic. In both systems, the learning-based method outperforms baselines, but the difference is much larger in the dVRK-Si. Nonetheless, dVRK-Si force estimation accuracy lags behind the dVRK Classic, with RMSE 2 to 3 times higher. Further analysis reveals poor PID control in the dVRK-Si. We hypothesize that this is due to the lack of gravity compensation, as unlike the dVRK Classic, the dVRK-Si is not mechanically balanced. This study advances the understanding of learning-based force estimation and is the first work to characterize the dynamics of the new dVRK-Si system.
Authors: Khai Nguyen, Hai Nguyen, Nhat Ho
Abstract: The sliced Wasserstein barycenter (SWB) is a widely acknowledged method for efficiently generalizing the averaging operation within probability measure spaces. However, achieving marginal fairness SWB, ensuring approximately equal distances from the barycenter to marginals, remains unexplored. The uniform weighted SWB is not necessarily the optimal choice to obtain the desired marginal fairness barycenter due to the heterogeneous structure of marginals and the non-optimality of the optimization. As the first attempt to tackle the problem, we define the marginal fairness sliced Wasserstein barycenter (MFSWB) as a constrained SWB problem. Due to the computational disadvantages of the formal definition, we propose two hyperparameter-free and computationally tractable surrogate MFSWB problems that implicitly minimize the distances to marginals and encourage marginal fairness at the same time. To further improve the efficiency, we perform slicing distribution selection and obtain the third surrogate definition by introducing a new slicing distribution that focuses more on marginally unfair projecting directions. We discuss the relationship of the three proposed problems and their relationship to sliced multi-marginal Wasserstein distance. Finally, we conduct experiments on finding 3D point-clouds averaging, color harmonization, and training of sliced Wasserstein autoencoder with class-fairness representation to show the favorable performance of the proposed surrogate MFSWB problems.
Authors: Zachariah Malik, Yu-Jui Huang
Abstract: This paper proposes a new theoretical lens to view Wasserstein generative adversarial networks (WGANs). To minimize the Wasserstein-1 distance between the true data distribution and our estimate of it, we derive a distribution-dependent ordinary differential equation (ODE) which represents the gradient flow of the Wasserstein-1 loss, and show that a forward Euler discretization of the ODE converges. This inspires a new class of generative models that naturally integrates persistent training (which we call W1-FE). When persistent training is turned off, we prove that W1-FE reduces to WGAN. When we intensify persistent training, W1-FE is shown to outperform WGAN in training experiments from low to high dimensions, in terms of both convergence speed and training results. Intriguingly, one can reap the benefits only when persistent training is carefully integrated through our ODE perspective. As demonstrated numerically, a naive inclusion of persistent training in WGAN (without relying on our ODE framework) can significantly worsen training results.
Authors: Maya Anderson, Guy Amit, Abigail Goldsteen
Abstract: Retrieval Augmented Generation (RAG) systems have shown great promise in natural language processing. However, their reliance on data stored in a retrieval database, which may contain proprietary or sensitive information, introduces new privacy concerns. Specifically, an attacker may be able to infer whether a certain text passage appears in the retrieval database by observing the outputs of the RAG system, an attack known as a Membership Inference Attack (MIA). Despite the significance of this threat, MIAs against RAG systems have yet remained under-explored. This study addresses this gap by introducing an efficient and easy-to-use method for conducting MIA against RAG systems. We demonstrate the effectiveness of our attack using two benchmark datasets and multiple generative models, showing that the membership of a document in the retrieval database can be efficiently determined through the creation of an appropriate prompt in both black-box and gray-box settings. Moreover, we introduce an initial defense strategy based on adding instructions to the RAG template, which shows high effectiveness for some datasets and models. Our findings highlight the importance of implementing security countermeasures in deployed RAG systems and developing more advanced defenses to protect the privacy and security of retrieval databases.
Authors: Vincent P. Grande, Michael T. Schaub
Abstract: Topological Data Analysis (TDA) allows us to extract powerful topological and higher-order information on the global shape of a data set or point cloud. Tools like Persistent Homology or the Euler Transform give a single complex description of the global structure of the point cloud. However, common machine learning applications like classification require point-level information and features to be available. In this paper, we bridge this gap and propose a novel method to extract node-level topological features from complex point clouds using discrete variants of concepts from algebraic topology and differential geometry. We verify the effectiveness of these topological point features (TOPF) on both synthetic and real-world data and study their robustness under noise and heterogeneous sampling.
Authors: Jackie Lok, Rishi Sonthalia, Elizaveta Rebrova
Abstract: We study the discrete dynamics of mini-batch gradient descent with random reshuffling for least squares regression. We show that the training and generalization errors depend on a sample cross-covariance matrix $Z$ between the original features $X$ and a set of new features $\widetilde{X}$ in which each feature is modified by the mini-batches that appear before it during the learning process in an averaged way. Using this representation, we establish that the dynamics of mini-batch and full-batch gradient descent agree up to leading order with respect to the step size using the linear scaling rule. However, mini-batch gradient descent with random reshuffling exhibits a subtle dependence on the step size that a gradient flow analysis cannot detect, such as converging to a limit that depends on the step size. By comparing $Z$, a non-commutative polynomial of random matrices, with the sample covariance matrix of $X$ asymptotically, we demonstrate that batching affects the dynamics by resulting in a form of shrinkage on the spectrum.
Authors: Ryota Watanabe, Takanori Ishii, Yuji Hirono, Hirokazu Maruoka
Abstract: Finding self-similarity is a key step for understanding the governing law behind complex physical phenomena. Traditional methods for identifying self-similarity often rely on specific models, which can introduce significant bias. In this paper, we present a novel neural network-based approach that discovers self-similarity directly from observed data, without presupposing any models. The presence of self-similar solutions in a physical problem signals that the governing law contains a function whose arguments are given by power-law monomials of physical parameters, which are characterized by power-law exponents. The basic idea is to enforce such particular forms structurally in a neural network in a parametrized way. We train the neural network model using the observed data, and when the training is successful, we can extract the power exponents that characterize scale-transformation symmetries of the physical problem. We demonstrate the effectiveness of our method with both synthetic and experimental data, validating its potential as a robust, model-independent tool for exploring self-similarity in complex systems.
Authors: Chufan Gao, Jathurshan Pradeepkumar, Trisha Das, Shivashankar Thati, Jimeng Sun
Abstract: Background: The global cost of drug discovery and development exceeds $200 billion annually, with clinical trial outcomes playing a critical role in the regulatory approval of new drugs and impacting patient outcomes. Despite their significance, large-scale, high-quality clinical trial outcome data are not readily available to the public, limiting advances in trial outcome predictive modeling. Methods: We introduce the Clinical Trial Outcome (CTO) knowledge base, a fully reproducible, large-scale (around 125K drug and biologics trials), open-source of clinical trial information including large language model (LLM) interpretations of publications, matched trials over phases, sentiment analysis from news, stock prices of trial sponsors, and other trial-related metrics. From this knowledge base, we additionally performed manual annotation of a set of recent clinical trials from 2020-2024. Results: We evaluated the quality of our knowledge base by generating high-quality trial outcome labels that demonstrate strong agreement with previously published expert annotations, achieving an F1 score of 94 for Phase 3 trials and 91 across all phases. Additionally, we benchmarked a suite of standard machine learning models on our manually annotated set, highlighting the distribution shift of recent trials and the need for continuously updated labeling methods. Conclusions: By analyzing CTO's performance on recent trials, we showed a need for recent, high-quality trial outcome labels. We release our knowledge base and labels to the public at https://chufangao.github.io/CTOD, which will also be regularly updated to support ongoing research in clinical trial outcomes, offering insights that could optimize the drug development process.
Authors: Efthymios Costa, Ioanna Papatsouma, Angelos Markos
Abstract: In this paper, we present an information-theoretic method for clustering mixed-type data, that is, data consisting of both continuous and categorical variables. The proposed approach is built on the deterministic variant of the Information Bottleneck algorithm, designed to optimally compress data while preserving its relevant structural information. We evaluate the performance of our method against four well-established clustering techniques for mixed-type data -- KAMILA, K-Prototypes, Factor Analysis for Mixed Data with K-Means, and Partitioning Around Medoids using Gower's dissimilarity -- using both simulated and real-world datasets. The results highlight that the proposed approach offers a competitive alternative to traditional clustering techniques, particularly under specific conditions where heterogeneity in data poses significant challenges.
Authors: Matan Kleiner, Lior Michaeli, Tomer Michaeli
Abstract: Diffractive neural networks hold great promise for applications requiring intensive computational processing. Considerable attention has focused on diffractive networks for either spatially coherent or spatially incoherent illumination. Here we illustrate that, as opposed to imaging systems, in diffractive networks the degree of spatial coherence has a dramatic effect. In particular, we show that when the spatial coherence length on the object is comparable to the minimal feature size preserved by the optical system, neither the incoherent nor the coherent extremes serve as acceptable approximations. Importantly, this situation is inherent to many settings involving active illumination, including reflected light microscopy, autonomous vehicles and smartphones. Following this observation, we propose a general framework for training diffractive networks for any specified degree of spatial and temporal coherence, supporting all types of linear and nonlinear layers. Using our method, we numerically optimize networks for image classification, and thoroughly investigate their performance dependence on the illumination coherence properties. We further introduce the concept of coherence-blind networks, which have enhanced resilience to changes in illumination conditions. Our findings serve as a steppingstone toward adopting all-optical neural networks in real-world applications, leveraging nothing but natural light.
Authors: Max D. Champneys, Timothy J. Rogers
Abstract: Model parsimony is an important \emph{cognitive bias} in data-driven modelling that aids interpretability and helps to prevent over-fitting. Sparse identification of nonlinear dynamics (SINDy) methods are able to learn sparse representations of complex dynamics directly from data, given a basis of library functions. In this work, a novel Bayesian treatment of dictionary learning system identification, as an alternative to SINDy, is envisaged. The proposed method -- Bayesian identification of nonlinear dynamics (BINDy) -- is distinct from previous approaches in that it targets the full joint posterior distribution over both the terms in the library and their parameterisation in the model. This formulation confers the advantage that an arbitrary prior may be placed over the model structure to produce models that are sparse in the model space rather than in parameter space. Because this posterior is defined over parameter vectors that can change in dimension, the inference cannot be performed by standard techniques. Instead, a Gibbs sampler based on reversible-jump Markov-chain Monte-Carlo is proposed. BINDy is shown to compare favourably to ensemble SINDy in three benchmark case-studies. In particular, it is seen that the proposed method is better able to assign high probability to correct model terms.
Authors: Mouxiang Chen, Lefei Shen, Zhuo Li, Xiaoyun Joy Wang, Jianling Sun, Chenghao Liu
Abstract: Foundation models have emerged as a promising approach in time series forecasting (TSF). Existing approaches either repurpose large language models (LLMs) or build large-scale time series datasets to develop TSF foundation models for universal forecasting. However, these methods face challenges due to the severe cross-domain gap or in-domain heterogeneity. This paper explores a new road to building a TSF foundation model from rich, high-quality natural images. Our key insight is that a visual masked autoencoder, pre-trained on the ImageNet dataset, can naturally be a numeric series forecaster. By reformulating TSF as an image reconstruction task, we bridge the gap between image pre-training and TSF downstream tasks. Surprisingly, without further adaptation in the time series domain, the proposed VisionTS could achieve better zero-shot forecast performance than existing TSF foundation models. With fine-tuning for one epoch, VisionTS could further improve the forecasting and achieve state-of-the-art performance in most cases. Extensive experiments reveal intrinsic similarities between images and real-world time series, suggesting that visual models may offer a "free lunch" for TSF and highlight the potential for future cross-modality research. Our code is publicly available at https://github.com/Keytoyze/VisionTS.
Authors: Constantin Waubert de Puiseau, Fabian Wolz, Merlin Montag, Jannik Peters, Hasan Tercan, Tobias Meisen
Abstract: The job shop scheduling problem (JSSP) and its solution algorithms have been of enduring interest in both academia and industry for decades. In recent years, machine learning (ML) is playing an increasingly important role in advancing existing and building new heuristic solutions for the JSSP, aiming to find better solutions in shorter computation times. In this paper we build on top of a state-of-the-art deep reinforcement learning (DRL) agent, called Neural Local Search (NLS), which can efficiently and effectively control a large local neighborhood search on the JSSP. In particular, we develop a method for training the decision transformer (DT) algorithm on search trajectories taken by a trained NLS agent to further improve upon the learned decision-making sequences. Our experiments show that the DT successfully learns local search strategies that are different and, in many cases, more effective than those of the NLS agent itself. In terms of the tradeoff between solution quality and acceptable computational time needed for the search, the DT is particularly superior in application scenarios where longer computational times are acceptable. In this case, it makes up for the longer inference times required per search step, which are caused by the larger neural network architecture, through better quality decisions per step. Thereby, the DT achieves state-of-the-art results for solving the JSSP with ML-enhanced search.
Authors: Bikram Khanal, Pablo Rivas, Arun Sanjel, Korn Sooksatra, Ernesto Quevedo, Alejandro Rodriguez
Abstract: Despite the mounting anticipation for the quantum revolution, the success of Quantum Machine Learning (QML) in the Noisy Intermediate-Scale Quantum (NISQ) era hinges on a largely unexplored factor: the generalization error bound, a cornerstone of robust and reliable machine learning models. Current QML research, while exploring novel algorithms and applications extensively, is predominantly situated in the context of noise-free, ideal quantum computers. However, Quantum Circuit (QC) operations in NISQ-era devices are susceptible to various noise sources and errors. In this article, we conduct a Systematic Mapping Study (SMS) to explore the state-of-the-art generalization bound for supervised QML in NISQ-era and analyze the latest practices in the field. Our study systematically summarizes the existing computational platforms with quantum hardware, datasets, optimization techniques, and the common properties of the bounds found in the literature. We further present the performance accuracy of various approaches in classical benchmark datasets like the MNIST and IRIS datasets. The SMS also highlights the limitations and challenges in QML in the NISQ era and discusses future research directions to advance the field. Using a detailed Boolean operators query in five reliable indexers, we collected 544 papers and filtered them to a small set of 37 relevant articles. This filtration was done following the best practice of SMS with well-defined research questions and inclusion and exclusion criteria.
Authors: Daewoong Kim, Hao-Wen Dong, Dasaem Jeong
Abstract: Modeling the natural contour of fundamental frequency (F0) plays a critical role in music audio synthesis. However, transcribing and managing multiple F0 contours in polyphonic music is challenging, and explicit F0 contour modeling has not yet been explored for polyphonic instrumental synthesis. In this paper, we present ViolinDiff, a two-stage diffusion-based synthesis framework. For a given violin MIDI file, the first stage estimates the F0 contour as pitch bend information, and the second stage generates mel spectrogram incorporating these expressive details. The quantitative metrics and listening test results show that the proposed model generates more realistic violin sounds than the model without explicit pitch bend modeling. Audio samples are available online: daewoung.github.io/ViolinDiff-Demo.
Authors: Blessed Guda, Gabrial Zencha Ashungafac, Lawrence Francis, Carlee Joe-Wong
Abstract: Large Language models (LLMs) have brought about substantial advancements in the field of Question Answering (QA) systems. These models do remarkably well in addressing intricate inquiries in a variety of disciplines. However, because of domain-specific vocabulary, complex technological concepts, and the requirement for exact responses applying LLMs to specialized sectors like telecommunications presents additional obstacles. GPT-3.5 has been used in recent work, to obtain noteworthy accuracy for telecom-related questions in a Retrieval Augmented Generation (RAG) framework. Notwithstanding these developments, the practical use of models such as GPT-3.5 is restricted by their proprietary nature and high computing demands. This paper introduces QMOS, an innovative approach which uses a Question-Masked loss and Option Shuffling trick to enhance the performance of LLMs in answering Multiple-Choice Questions in the telecommunications domain. Our focus was on using opensource, smaller language models (Phi-2 and Falcon-7B) within an enhanced RAG framework. Our multi-faceted approach involves several enhancements to the whole LLM-RAG pipeline of finetuning, retrieval, prompt engineering and inference. Our approaches significantly outperform existing results, achieving accuracy improvements from baselines of 24.70% to 49.30% with Falcon-7B and from 42.07% to 84.65% with Phi-2.
Authors: Mengze Hong, Chen Jason Zhang, Lingxiao Yang, Yuanfeng Song, Di Jiang
Abstract: Understanding the meaning of infant cries is a significant challenge for young parents in caring for their newborns. The presence of background noise and the lack of labeled data present practical challenges in developing systems that can detect crying and analyze its underlying reasons. In this paper, we present a novel data-driven framework, "InfantCryNet," for accomplishing these tasks. To address the issue of data scarcity, we employ pre-trained audio models to incorporate prior knowledge into our model. We propose the use of statistical pooling and multi-head attention pooling techniques to extract features more effectively. Additionally, knowledge distillation and model quantization are applied to enhance model efficiency and reduce the model size, better supporting industrial deployment in mobile devices. Experiments on real-life datasets demonstrate the superior performance of the proposed framework, outperforming state-of-the-art baselines by 4.4% in classification accuracy. The model compression effectively reduces the model size by 7% without compromising performance and by up to 28% with only an 8% decrease in accuracy, offering practical insights for model selection and system design.
Authors: Catherine Yeh, Donghao Ren, Yannick Assogba, Dominik Moritz, Fred Hohman
Abstract: Data augmentation is crucial to make machine learning models more robust and safe. However, augmenting data can be challenging as it requires generating diverse data points to rigorously evaluate model behavior on edge cases and mitigate potential harms. Creating high-quality augmentations that cover these "unknown unknowns" is a time- and creativity-intensive task. In this work, we introduce Amplio, an interactive tool to help practitioners navigate "unknown unknowns" in unstructured text datasets and improve data diversity by systematically identifying empty data spaces to explore. Amplio includes three human-in-the-loop data augmentation techniques: Augment With Concepts, Augment by Interpolation, and Augment with Large Language Model. In a user study with 18 professional red teamers, we demonstrate the utility of our augmentation methods in helping generate high-quality, diverse, and relevant model safety prompts. We find that Amplio enabled red teamers to augment data quickly and creatively, highlighting the transformative potential of interactive augmentation workflows.
Authors: Dwipam Katariya, Juan Manuel Origgi, Yage Wang, Thomas Caputo
Abstract: Users engage with financial services companies through multiple channels, often interacting with mobile applications, web platforms, call centers, and physical locations to service their accounts. The resulting interactions are recorded at heterogeneous temporal resolutions across these domains. This multi-channel data can be combined and encoded to create a comprehensive representation of the customer's journey for accurate intent prediction. This demands sequential learning solutions. NMT transformers achieve state-of-the-art sequential representation learning by encoding context and decoding for the next best action to represent long-range dependencies. However, three major challenges exist while combining multi-domain sequences within an encoder-decoder transformers architecture for intent prediction applications: a) aligning sequences with different sampling rates b) learning temporal dynamics across multi-variate, multi-domain sequences c) combining dynamic and static sequences. We propose an encoder-decoder transformer model to address these challenges for contextual and sequential intent prediction in financial servicing applications. Our experiments show significant improvement over the existing tabular method.
Authors: Xuan Zhang, Fengzhuo Zhang, Cunxiao Du, Chao Du, Tianyu Pang, Wei Gao, Min Lin
Abstract: Scaling language models to handle longer contexts introduces substantial memory challenges due to the growing cost of key-value (KV) caches. Motivated by the efficiency gains of hybrid models and the broad availability of pretrained large transformer backbones, we explore transitioning transformer models into hybrid architectures for a more efficient generation. In this work, we propose LightTransfer, a lightweight method that transforms models such as LLaMA into hybrid variants. Our approach identifies lazy layers -- those focusing on recent or initial tokens -- and replaces their full attention with streaming attention. This transformation can be performed without any training for long-context understanding tasks or with minimal fine-tuning for o1-like long reasoning generation tasks that require stronger reasoning capabilities. Experiments across diverse benchmarks and models (e.g., LLaMA, Mistral, QwQ-STILL) demonstrate that, even when half of the layers are identified as lazy, LightTransfer achieves up to 2.17$\times$ throughput improvement with minimal performance loss ($<1.5\%$ on LongBench) and achieves 53.3\% on math benchmark AIME24 of advanced o1-like long reasoning model QwQ-STILL.
Authors: Zheng Wang, Wanwan Wang, Yimin Huang, Zhaopeng Peng, Ziqi Yang, Ming Yao, Cheng Wang, Xiaoliang Fan
Abstract: In recent years, graph neural networks (GNNs) have been commonly utilized for social recommendation systems. However, real-world scenarios often present challenges related to user privacy and business constraints, inhibiting direct access to valuable social information from other platforms. While many existing methods have tackled matrix factorization-based social recommendations without direct social data access, developing GNN-based federated social recommendation models under similar conditions remains largely unexplored. To address this issue, we propose a novel vertical federated social recommendation method leveraging privacy-preserving two-party graph convolution networks (P4GCN) to enhance recommendation accuracy without requiring direct access to sensitive social information. First, we introduce a Sandwich-Encryption module to ensure comprehensive data privacy during the collaborative computing process. Second, we provide a thorough theoretical analysis of the privacy guarantees, considering the participation of both curious and honest parties. Extensive experiments on four real-world datasets demonstrate that P4GCN outperforms state-of-the-art methods in terms of recommendation accuracy.
Authors: Prakhar Verma, Sukruta Prakash Midigeshi, Gaurav Sinha, Arno Solin, Nagarajan Natarajan, Amit Sharma
Abstract: We introduce Plan*RAG, a novel framework that enables structured multi-hop reasoning in retrieval-augmented generation (RAG) through test-time reasoning plan generation. While existing approaches such as ReAct maintain reasoning chains within the language model's context window, we observe that this often leads to plan fragmentation and execution failures. Our key insight is that by isolating the reasoning plan as a directed acyclic graph (DAG) outside the LM's working memory, we can enable (1) systematic exploration of reasoning paths, (2) atomic subqueries enabling precise retrievals and grounding, and (3) efficiency through parallel execution and bounded context window utilization. Moreover, Plan*RAG's modular design allows it to be integrated with existing RAG methods, thus providing a practical solution to improve current RAG systems. On standard multi-hop reasoning benchmarks, Plan*RAG consistently achieves improvements over recently proposed methods such as RQ-RAG and Self-RAG, while maintaining comparable computational costs.
Authors: Yunuo Wang, Ningning Yang, Jialin Li
Abstract: Generative Adversarial Networks (GANs) have surfaced as a revolutionary element within the domain of low-dose computed tomography (LDCT) imaging, providing an advanced resolution to the enduring issue of reconciling radiation exposure with image quality. This comprehensive review synthesizes the rapid advancements in GAN-based LDCT denoising techniques, examining the evolution from foundational architectures to state-of-the-art models incorporating advanced features such as anatomical priors, perceptual loss functions, and innovative regularization strategies. We critically analyze various GAN architectures, including conditional GANs (cGANs), CycleGANs, and Super-Resolution GANs (SRGANs), elucidating their unique strengths and limitations in the context of LDCT denoising. The evaluation provides both qualitative and quantitative results related to the improvements in performance in benchmark and clinical datasets with metrics such as PSNR, SSIM, and LPIPS. After highlighting the positive results, we discuss some of the challenges preventing a wider clinical use, including the interpretability of the images generated by GANs, synthetic artifacts, and the need for clinically relevant metrics. The review concludes by highlighting the essential significance of GAN-based methodologies in the progression of precision medicine via tailored LDCT denoising models, underlining the transformative possibilities presented by artificial intelligence within contemporary radiological practice.
Authors: Shuntaro Okada, Kenji Doi, Ryota Yoshihashi, Hirokatsu Kataoka, Tomohiro Tanaka
Abstract: We propose a noise schedule that ensures a constant rate of change in the probability distribution of diffused data throughout the diffusion process. To obtain this schedule, we measure the probability-distributional change of diffused data by simulating the forward process and use it to determine the noise schedule before training diffusion models. The functional form of the noise schedule is automatically determined and tailored to each dataset and type of diffusion model, such as pixel space or latent space. We evaluate the effectiveness of our noise schedule on unconditional and class-conditional image generation tasks using the LSUN (Bedroom, Church, Cat, Horse), ImageNet, and FFHQ datasets. Through extensive experiments, we confirmed that our noise schedule broadly improves the performance of the pixel-space and latent-space diffusion models regardless of the dataset, sampler, and number of function evaluations.
Authors: Sara Pohland, Claire Tomlin
Abstract: Convolutional neural networks (CNNs) are extremely popular and effective for image classification tasks but tend to be overly confident in their predictions. Various works have sought to quantify uncertainty associated with these models, detect out-of-distribution (OOD) inputs, or identify anomalous regions in an image, but limited work has sought to develop a holistic approach that can accurately estimate perception model confidence across various sources of uncertainty. We develop a probabilistic and reconstruction-based competency estimation (PaRCE) method and compare it to existing approaches for uncertainty quantification and OOD detection. We find that our method can best distinguish between correctly classified, misclassified, and OOD samples with anomalous regions, as well as between samples with visual image modifications resulting in high, medium, and low prediction accuracy. We describe how to extend our approach for anomaly localization tasks and demonstrate the ability of our approach to distinguish between regions in an image that are familiar to the perception model from those that are unfamiliar. We find that our method generates interpretable scores that most reliably capture a holistic notion of perception model confidence.
Authors: Zijian Zhang, Kaiyuan Zheng, Zhaorun Chen, Joel Jang, Yi Li, Siwei Han, Chaoqi Wang, Mingyu Ding, Dieter Fox, Huaxiu Yao
Abstract: Despite the recent advancements of vision-language-action (VLA) models on a variety of robotics tasks, they suffer from critical issues such as poor generalizability to unseen tasks, due to their reliance on behavior cloning exclusively from successful rollouts. Furthermore, they are typically fine-tuned to replicate demonstrations collected by experts under different settings, thus introducing distribution bias and limiting their adaptability to diverse manipulation objectives, such as efficiency, safety, and task completion. To bridge this gap, we introduce GRAPE: Generalizing Robot Policy via Preference Alignment. Specifically, GRAPE aligns VLAs on a trajectory level and implicitly models reward from both successful and failure trials to boost generalizability to diverse tasks. Moreover, GRAPE breaks down complex manipulation tasks to independent stages and automatically guides preference modeling through customized spatiotemporal constraints with keypoints proposed by a large vision-language model. Notably, these constraints are flexible and can be customized to align the model with varying objectives, such as safety, efficiency, or task success. We evaluate GRAPE across a diverse array of tasks in both real-world and simulated environments. Experimental results demonstrate that GRAPE enhances the performance of state-of-the-art VLA models, increasing success rates on in-domain and unseen manipulation tasks by 51.79% and 58.20%, respectively. Additionally, GRAPE can be aligned with various objectives, such as safety and efficiency, reducing collision rates by 37.44% and rollout step-length by 11.15%, respectively. All code, models, and data are available at https://grape-vla.github.io/
Authors: Anian Ruoss, Fabio Pardo, Harris Chan, Bonnie Li, Volodymyr Mnih, Tim Genewein
Abstract: In this paper, we present a benchmark to pressure-test today's frontier models' multimodal decision-making capabilities in the very long-context regime (up to one million tokens) and investigate whether these models can learn from large numbers of expert demonstrations in their context. We evaluate the performance of Claude 3.5 Sonnet, Gemini 1.5 Flash, Gemini 1.5 Pro, Gemini 2.0 Flash Experimental, GPT-4o, o1-mini, o1-preview, and o1 as policies across a battery of simple interactive decision-making tasks: playing tic-tac-toe, chess, and Atari, navigating grid worlds, solving crosswords, and controlling a simulated cheetah. We study increasing amounts of expert demonstrations in the context $\unicode{x2013}$ from no demonstrations to 512 full episodes. Across our tasks, models rarely manage to fully reach expert performance, and often, presenting more demonstrations has little effect. Some models steadily improve with more demonstrations on a few tasks. We investigate the effect of encoding observations as text or images and the impact of chain-of-thought prompting. To help quantify the impact of other approaches and future innovations, we open source our benchmark that covers the zero-, few-, and many-shot regimes in a unified evaluation.
Authors: Yaron Veksler, Sharon Hornstein, Han Wang, Maria Laura Delle Monache, Daniel Urieli
Abstract: The proliferation of connected automated vehicles represents an unprecedented opportunity for improving driving efficiency and alleviating traffic congestion. However, existing research fails to address realistic multi-lane highway scenarios without assuming connectivity, perception, and control capabilities that are typically unavailable in current vehicles. This paper proposes a novel AI system that is the first to improve highway traffic efficiency compared with human-like traffic in realistic, simulated multi-lane scenarios, while relying on existing connectivity, perception, and control capabilities. At the core of our approach is a reinforcement learning based controller that dynamically communicates time-headways to automated vehicles near bottlenecks based on real-time traffic conditions. These desired time-headways are then used by adaptive cruise control (ACC) systems to adjust their following distance. By (i) integrating existing traffic estimation technology and low-bandwidth vehicle-to-infrastructure connectivity, (ii) leveraging safety-certified ACC systems, and (iii) targeting localized bottleneck challenges that can be addressed independently in different locations, we propose a potentially practical, safe, and scalable system that can positively impact numerous road users.
Authors: Lucky Li
Abstract: We study the discrete-time linear-quadratic (LQ) control model using reinforcement learning (RL). Using entropy to measure the cost of exploration, we prove that the optimal feedback policy for the problem must be Gaussian type. Then, we apply the results of the discrete-time LQ model to solve the discrete-time mean-variance asset-liability management problem and prove our RL algorithm's policy improvement and convergence. Finally, a numerical example sheds light on the theoretical results established using simulations.
Authors: Trishita Tiwari, G. Edward Suh
Abstract: This work performs an analysis of sequence-level probabilities for quantifying the of risk training data extraction from Large Language Models (LLMs). Per-sequence extraction probabilities provide finer-grained information than has been studied in prior work. We re-analyze the effects of decoding schemes, model sizes, prefix lengths, partial sequence leakages, and token positions to uncover new insights that were not possible in previous works due to their choice of metrics. We perform this study on two pre-trained models, Llama and OPT, trained on the Common Crawl and The Pile respectively. We discover that 1) Extraction Rate, the predominant metric used in prior quantification work, underestimates the threat of leakage of training data in randomized LLMs by as much as 2.14X. 2) Although on average, larger models and longer prefixes can extract more data, this is not true for a substantial portion of individual sequences. 30.4-41.5% of our sequences are easier to extract with either shorter prefixes or smaller models. 3) Contrary to previous beliefs, partial leakage in commonly used decoding schemes like top-k and top-p is not easier than leaking verbatim training data. 4) Extracting later tokens in a sequence is as much as 10.12X easier than extracting earlier tokens. The insights gained from our analysis shed light on the nature of memorization of training data on a per-sequence basis.
Authors: Guoxuan Chen, Han Shi, Jiawei Li, Yihang Gao, Xiaozhe Ren, Yimeng Chen, Xin Jiang, Zhenguo Li, Weiyang Liu, Chao Huang
Abstract: Large Language Models (LLMs) have exhibited exceptional performance across a spectrum of natural language processing tasks. However, their substantial sizes pose considerable challenges, particularly in computational demands and inference speed, due to their quadratic complexity. In this work, we have identified a key pattern: certain seemingly meaningless special tokens (i.e., separators) contribute disproportionately to attention scores compared to semantically meaningful tokens. This observation suggests that information of the segments between these separator tokens can be effectively condensed into the separator tokens themselves without significant information loss. Guided by this insight, we introduce SepLLM, a plug-and-play framework that accelerates inference by compressing these segments and eliminating redundant tokens. Additionally, we implement efficient kernels for training acceleration. Experimental results across training-free, training-from-scratch, and post-training settings demonstrate SepLLM's effectiveness. Notably, using the Llama-3-8B backbone, SepLLM achieves over 50% reduction in KV cache on the GSM8K-CoT benchmark while maintaining comparable performance. Furthermore, in streaming settings, SepLLM effectively processes sequences of up to 4 million tokens or more while maintaining consistent language modeling capabilities.
Authors: Hengrui Zhang, Alexandru B. Georgescu, Suraj Yerramilli, Christopher Karpovich, Daniel W. Apley, Elsa A. Olivetti, James M. Rondinelli, Wei Chen
Abstract: The increasing demands of sustainable energy, electronics, and biomedical applications call for next-generation functional materials with unprecedented properties. Of particular interest are emerging materials that display exceptional physical properties, making them promising candidates in energy-efficient microelectronic devices. As the conventional Edisonian approach becomes significantly outpaced by growing societal needs, emerging computational modeling and machine learning (ML) methods are employed for the rational design of materials. However, the complex physical mechanisms, cost of first-principles calculations, and the dispersity and scarcity of data pose challenges to both physics-based and data-driven materials modeling. Moreover, the combinatorial composition-structure design space is high-dimensional and often disjoint, making design optimization nontrivial. In this Account, we review a team effort toward establishing a framework that integrates data-driven and physics-based methods to address these challenges and accelerate materials design. We begin by presenting our integrated materials design framework and its three components in a general context. We then provide an example of applying this materials design framework to metal-insulator transition (MIT) materials, a specific type of emerging materials with practical importance in next-generation memory technologies. We identify multiple new materials which may display this property and propose pathways for their synthesis. Finally, we identify some outstanding challenges in data-driven materials design, such as materials data quality issues and property-performance mismatch. We seek to raise awareness of these overlooked issues hindering materials design, thus stimulating efforts toward developing methods to mitigate the gaps.
Authors: Zhenglai Li, Jun Wang, Chang Tang, Xinzhong Zhu, Wei Zhang, Xinwang Liu
Abstract: Multi-view clustering (MvC) aims to integrate information from different views to enhance the capability of the model in capturing the underlying data structures. The widely used joint training paradigm in MvC is potentially not fully leverage the multi-view information, since the imbalanced and under-optimized view-specific features caused by the uniform learning objective for all views. For instance, particular views with more discriminative information could dominate the learning process in the joint training paradigm, leading to other views being under-optimized. To alleviate this issue, we first analyze the imbalanced phenomenon in the joint-training paradigm of multi-view clustering from the perspective of gradient descent for each view-specific feature extractor. Then, we propose a novel balanced multi-view clustering (BMvC) method, which introduces a view-specific contrastive regularization (VCR) to modulate the optimization of each view. Concretely, VCR preserves the sample similarities captured from the joint features and view-specific ones into the clustering distributions corresponding to view-specific features to enhance the learning process of view-specific feature extractors. Additionally, a theoretical analysis is provided to illustrate that VCR adaptively modulates the magnitudes of gradients for updating the parameters of view-specific feature extractors to achieve a balanced multi-view learning procedure. In such a manner, BMvC achieves a better trade-off between the exploitation of view-specific patterns and the exploration of view-invariance patterns to fully learn the multi-view information for the clustering task. Finally, a set of experiments are conducted to verify the superiority of the proposed method compared with state-of-the-art approaches on eight benchmark MvC datasets.
Authors: Guoxuan Chen, Lianghao Xia, Chao Huang
Abstract: Graph neural networks (GNNs) have demonstrated superior performance in collaborative recommendation through their ability to conduct high-order representation smoothing, effectively capturing structural information within users' interaction patterns. However, existing GNN paradigms face significant challenges in scalability and robustness when handling large-scale, noisy, and real-world datasets. To address these challenges, we present LightGNN, a lightweight and distillation-based GNN pruning framework designed to substantially reduce model complexity while preserving essential collaboration modeling capabilities. Our LightGNN framework introduces a computationally efficient pruning module that adaptively identifies and removes redundant edges and embedding entries for model compression. The framework is guided by a resource-friendly hierarchical knowledge distillation objective, whose intermediate layer augments the observed graph to maintain performance, particularly in high-rate compression scenarios. Extensive experiments on public datasets demonstrate LightGNN's effectiveness, significantly improving both computational efficiency and recommendation accuracy. Notably, LightGNN achieves an 80% reduction in edge count and 90% reduction in embedding entries while maintaining performance comparable to more complex state-of-the-art baselines. The implementation of our LightGNN framework is available at the github repository: https://github.com/HKUDS/LightGNN.
Authors: Sanchit Sinha, Guangzhi Xiong, Aidong Zhang
Abstract: As Vision Transformers (ViTs) are increasingly adopted in sensitive vision applications, there is a growing demand for improved interpretability. This has led to efforts to forward-align these models with carefully annotated abstract, human-understandable semantic entities - concepts. Concepts provide global rationales to the model predictions and can be quickly understood/intervened on by domain experts. Most current research focuses on designing model-agnostic, plug-and-play generic concept-based explainability modules that do not incorporate the inner workings of foundation models (e.g., inductive biases, scale invariance, etc.) during training. To alleviate this issue for ViTs, in this paper, we propose ASCENT-ViT, an attention-based, concept learning framework that effectively composes scale and position-aware representations from multiscale feature pyramids and ViT patch representations, respectively. Further, these representations are aligned with concept annotations through attention matrices - which incorporate spatial and global (semantic) concepts. ASCENT-ViT can be utilized as a classification head on top of standard ViT backbones for improved predictive performance and accurate and robust concept explanations as demonstrated on five datasets, including three widely used benchmarks (CUB, Pascal APY, Concept-MNIST) and 2 real-world datasets (AWA2, KITS).
Authors: Sara Abdali, Can Goksen, Saeed Amizadeh, Kazuhito Koishida
Abstract: Investigating NLP through a philosophical lens has recently caught researcher's eyes as it connects computational methods with classical schools of philosophy. This paper introduces a philosophical approach inspired by the Hegelian Dialectic for LLMs' self-reflection, utilizing a self-dialectical approach to emulate internal critiques and then synthesize new ideas by resolving the contradicting points. Moreover, this paper investigates the effect of LLMs' temperature for generation by establishing a dynamic annealing approach, which promotes the creativity in the early stages and gradually refines it by focusing on the nuances, as well as a fixed temperature strategy for generation. Our proposed approach is examined to determine its ability to generate novel ideas from an initial proposition. Additionally, a Multi Agent Majority Voting (MAMV) strategy is leveraged to assess the validity and novelty of the generated ideas, which proves beneficial in the absence of domain experts. Our experiments show promise in generating new ideas and provide a stepping stone for future research.
Authors: Ying Chen, Paul Griffin, Paolo Recchia, Lei Zhou, Hongrui Zhang
Abstract: Recovery rate prediction plays a pivotal role in bond investment strategies, enhancing risk assessment, optimizing portfolio allocation, improving pricing accuracy, and supporting effective credit risk management. However, forecasting faces challenges like high-dimensional features, small sample sizes, and overfitting. We propose a hybrid Quantum Machine Learning model incorporating Parameterized Quantum Circuits (PQC) within a neural network framework. PQCs inherently preserve unitarity, avoiding computationally costly orthogonality constraints, while amplitude encoding enables exponential data compression, reducing qubit requirements logarithmically. Applied to a global dataset of 1,725 observations (1996-2023), our method achieved superior accuracy (RMSE 0.228) compared to classical neural networks (0.246) and quantum models with angle encoding (0.242), with efficient computation times. This work highlights the potential of hybrid quantum-classical architectures in advancing recovery rate forecasting.
Authors: Philippe Pasquier, Jeff Ens, Nathan Fradet, Paul Triana, Davide Rizzotti, Jean-Baptiste Rolland, Maryam Safi
Abstract: We present and release MIDI-GPT, a generative system based on the Transformer architecture that is designed for computer-assisted music composition workflows. MIDI-GPT supports the infilling of musical material at the track and bar level, and can condition generation on attributes including: instrument type, musical style, note density, polyphony level, and note duration. In order to integrate these features, we employ an alternative representation for musical material, creating a time-ordered sequence of musical events for each track and concatenating several tracks into a single sequence, rather than using a single time-ordered sequence where the musical events corresponding to different tracks are interleaved. We also propose a variation of our representation allowing for expressiveness. We present experimental results that demonstrate that MIDI-GPT is able to consistently avoid duplicating the musical material it was trained on, generate music that is stylistically similar to the training dataset, and that attribute controls allow enforcing various constraints on the generated material. We also outline several real-world applications of MIDI-GPT, including collaborations with industry partners that explore the integration and evaluation of MIDI-GPT into commercial products, as well as several artistic works produced using it.
Authors: David Salinas, Omar Swelam, Frank Hutter
Abstract: Evaluating Large Language Models (LLMs) often requires costly human annotations. To address this, LLM-based judges have been proposed, which compare the outputs of two LLMs enabling the ranking of models without human intervention. While several approaches have been proposed, many confounding factors are present between different papers. For instance the model, the prompt and other hyperparameters are typically changed at the same time making apple-to-apple comparisons challenging. In this paper, we propose to systematically analyze and tune hyperparameter of LLM judges. To alleviate the high cost of evaluating a judge, we propose to leverage multi-objective multi-fidelity which allows to find judges that trades accuracy for cost and also reduce significantly the cost of the search. Our method identifies judges that not only outperform existing benchmarks in accuracy and cost-efficiency but also utilize open-weight models, ensuring greater accessibility and reproducibility.
Authors: Hyunju Kang, Geonhee Han, Yoonjae Jeong, Hogun Park
Abstract: Text-to-audio generation models (TAG) have achieved significant advances in generating audio conditioned on text descriptions. However, a critical challenge lies in the lack of transparency regarding how each textual input impacts the generated audio. To address this issue, we introduce AudioGenX, an Explainable AI (XAI) method that provides explanations for text-to-audio generation models by highlighting the importance of input tokens. AudioGenX optimizes an Explainer by leveraging factual and counterfactual objective functions to provide faithful explanations at the audio token level. This method offers a detailed and comprehensive understanding of the relationship between text inputs and audio outputs, enhancing both the explainability and trustworthiness of TAG models. Extensive experiments demonstrate the effectiveness of AudioGenX in producing faithful explanations, benchmarked against existing methods using novel evaluation metrics specifically designed for audio generation tasks.
Authors: Shuyuan Zheng, Sudong Cai, Chuan Xiao, Yang Cao, Jianbin Qin, Masatoshi Yoshikawa, Makoto Onizuka
Abstract: In collaborative machine learning, data valuation, i.e., evaluating the contribution of each client' data to the machine learning model, has become a critical task for incentivizing and selecting positive data contributions. However, existing studies often assume that clients engage in data valuation truthfully, overlooking the practical motivation for clients to exaggerate their contributions. To unlock this threat, this paper introduces the first data overvaluation attack, enabling strategic clients to have their data significantly overvalued. Furthermore, we propose a truthful data valuation metric, named Truth-Shapley. Truth-Shapley is the unique metric that guarantees some promising axioms for data valuation while ensuring that clients' optimal strategy is to perform truthful data valuation. Our experiments demonstrate the vulnerability of existing data valuation metrics to the data overvaluation attack and validate the robustness and effectiveness of Truth-Shapley.
Authors: Yang Cao, Zhao Song, Chiwun Yang
Abstract: This paper considers an efficient video modeling process called Video Latent Flow Matching (VLFM). Unlike prior works, which randomly sampled latent patches for video generation, our method relies on current strong pre-trained image generation models, modeling a certain caption-guided flow of latent patches that can be decoded to time-dependent video frames. We first speculate multiple images of a video are differentiable with respect to time in some latent space. Based on this conjecture, we introduce the HiPPO framework to approximate the optimal projection for polynomials to generate the probability path. Our approach gains the theoretical benefits of the bounded universal approximation error and timescale robustness. Moreover, VLFM processes the interpolation and extrapolation abilities for video generation with arbitrary frame rates. We conduct experiments on several text-to-video datasets to showcase the effectiveness of our method.
Authors: Kunal Pai, Premkumar Devanbu, Toufique Ahmed
Abstract: One of the central tasks in software maintenance is being able to understand and develop code changes. Thus, given a natural language description of the desired new operation of a function, an agent (human or AI) might be asked to generate the set of edits to that function to implement the desired new operation; likewise, given a set of edits to a function, an agent might be asked to generate a changed description, of that function's new workings. Thus, there is an incentive to train a neural model for change-related tasks. Motivated by this, we offer a new, "natural", large dataset of coupled changes to code and documentation mined from actual high-quality GitHub projects, where each sample represents a single commit where the code and the associated docstring were changed together. We present the methodology for gathering the dataset, and some sample, challenging (but realistic) tasks where our dataset provides opportunities for both learning and evaluation. We find that current models (specifically Llama-3.1 405B, Mixtral 8$\times$22B) do find these maintenance-related tasks challenging.