Authors: Yurun Wu, Yousong Sun, Burkhard Wunsche, Jia Wang, Elliott Wen
Abstract: Virtual Reality (VR) has rapidly become a mainstream platform for gaming and interactive experiences, yet ensuring the quality, safety, and appropriateness of VR content remains a pressing challenge. Traditional human-based quality assurance is labor-intensive and cannot scale with the industry's rapid growth. While automated testing has been applied to traditional 2D and 3D games, extending it to VR introduces unique difficulties due to high-dimensional sensory inputs and strict real-time performance requirements. We present VRScout, a deep learning-based agent capable of autonomously navigating VR environments and interacting with virtual objects in a human-like and real-time manner. VRScout learns from human demonstrations using an enhanced Action Chunking Transformer that predicts multi-step action sequences. This enables our agent to capture higher-level strategies and generalize across diverse environments. To balance responsiveness and precision, we introduce a dynamically adjustable sliding horizon that adapts the agent's temporal context at runtime. We evaluate VRScout on commercial VR titles and show that it achieves expert-level performance with only limited training data, while maintaining real-time inference at 60 FPS on consumer-grade hardware. These results position VRScout as a practical and scalable framework for automated VR game testing, with direct applications in both quality assurance and safety auditing.
Authors: Samaksh Bhargav, Zining Zhu
Abstract: Large Language Model (LLM) deployment requires guiding the LLM to recognize and not answer unsafe prompts while complying with safe prompts. Previous methods for achieving this require adjusting model weights along with other expensive procedures. While recent advances in Sparse Autoencoders (SAEs) have enabled interpretable feature extraction from LLMs, existing approaches lack systematic feature selection methods and principled evaluation of safety-utility tradeoffs. We explored using different steering features and steering strengths using Sparse Auto Encoders (SAEs) to provide a solution. Using an accurate and innovative contrasting prompt method with the AI-Generated Prompts Dataset from teknium/OpenHermes-2p5-Mistral-7B and Air Bench eu-dataset to efficiently choose the best features in the model to steer, we tested this method on Llama-3 8B. We conclude that using this method, our approach achieves an 18.9% improvement in safety performance while simultaneously increasing utility by 11.1%, demonstrating that targeted SAE steering can overcome traditional safety-utility tradeoffs when optimal features are identified through principled selection methods.
Authors: Myeongseob Ko, Hoang Anh Just, Charles Fleming, Ming Jin, Ruoxi Jia
Abstract: Machine unlearning has emerged as a prevalent technical solution for selectively removing unwanted knowledge absorbed during pre-training, without requiring full retraining. While recent unlearning techniques can effectively remove undesirable content without severely compromising performance on standard benchmarks, we find that they may inadvertently create ``knowledge holes'' -- unintended losses of benign knowledge that standard benchmarks fail to capture. To probe where unlearned models reveal knowledge holes, we propose a test case generation framework that explores both immediate neighbors of unlearned content and broader areas of potential failures. Our evaluation demonstrates significant hidden costs of unlearning: up to 98.7\% of the test cases yield irrelevant or nonsensical responses from unlearned models, despite being answerable by the pretrained model. These findings necessitate rethinking the conventional approach to evaluating knowledge preservation in unlearning, moving beyond standard, static benchmarks.
Authors: Lei Liu, Zhongyi Yu, Hong Wang, Huanshuo Dong, Haiyang Xin, Hongwei Zhao, Bin Li
Abstract: In recent years, Neural Operators(NO) have gradually emerged as a popular approach for solving Partial Differential Equations (PDEs). However, their application to large-scale engineering tasks suffers from significant computational overhead. And the fact that current models impose a uniform computational cost while physical fields exhibit vastly different complexities constitutes a fundamental mismatch, which is the root of this inefficiency. For instance, in turbulence flows, intricate vortex regions require deeper network processing compared to stable flows. To address this, we introduce a framework: Skip-Block Routing (SBR), a general framework designed for Transformer-based neural operators, capable of being integrated into their multi-layer architectures. First, SBR uses a routing mechanism to learn the complexity and ranking of tokens, which is then applied during inference. Then, in later layers, it decides how many tokens are passed forward based on this ranking. This way, the model focuses more processing capacity on the tokens that are more complex. Experiments demonstrate that SBR is a general framework that seamlessly integrates into various neural operators. Our method reduces computational cost by approximately 50% in terms of Floating Point Operations (FLOPs), while still delivering up to 2x faster inference without sacrificing accuracy.
Authors: Georg Velev, Stefan Lessmann
Abstract: The dynamic energy sector requires both predictive accuracy and runtime efficiency for short-term forecasting of energy generation under operational constraints, where timely and precise predictions are crucial. The manual configuration of complex methods, which can generate accurate global multi-step predictions without suffering from a computational bottleneck, represents a procedure with significant time requirements and high risk for human-made errors. A further intricacy arises from the temporal dynamics present in energy-related data. Additionally, the generalization to unseen data is imperative for continuously deploying forecasting techniques over time. To overcome these challenges, in this research, we design a neural architecture search (NAS)-based framework for the automated discovery of time series models that strike a balance between computational efficiency, predictive performance, and generalization power for the global, multi-step short-term forecasting of energy production time series. In particular, we introduce a search space consisting only of efficient components, which can capture distinctive patterns of energy time series. Furthermore, we formulate a novel objective function that accounts for performance generalization in temporal context and the maximal exploration of different regions of our high-dimensional search space. The results obtained on energy production time series show that an ensemble of lightweight architectures discovered with NAS outperforms state-of-the-art techniques, such as Transformers, as well as pre-trained forecasting models, in terms of both efficiency and accuracy.
Authors: Seonggyun Lee, Sungjun Lim, Seojin Park, Soeun Cheon, Kyungwoo Song
Abstract: The field of preference optimization has made outstanding contributions to the alignment of language models with human preferences. Despite these advancements, recent methods still rely heavily on substantial paired (labeled) feedback data, leading to substantial resource expenditures. To address these challenges, we study the problem of Semi-Supervised Preference Optimization (SSPO) in which the idea is to learn from both a small number of pairwise preference labels and a large pool of unpaired samples simultaneously. Our key theoretical contribution proves the existence of an optimal reward threshold capable of separating winning and losing responses with high probability, which enables a principled pseudo-labeling of unpaired data. By leveraging these pseudo-labels, SSPO effectively distills latent preferences from large-scale unpaired data, thus maintaining human alignment while drastically reducing acquisition costs. Extensive experiments across datasets validate this remarkable data efficiency; for instance, SSPO trained with Llama3-8B-Instruct on just 1% of UltraFeedback consistently surpasses strong baselines trained on 10% of UltraFeedback.
Authors: Tyrus Whitman, Andrew Particka, Christopher Diers, Ian Griffin, Charuka Wickramasinghe, Pradeep Ranaweera
Abstract: In this study, we present and validate the predictive capability of the Physics-Informed Neural Networks (PINNs) methodology for solving a variety of engineering and biological dynamical systems governed by ordinary differential equations (ODEs). While traditional numerical methods a re effective for many ODEs, they often struggle to achieve convergence in problems involving high stiffness, shocks, irregular domains, singular perturbations, high dimensions, or boundary discontinuities. Alternatively, PINNs offer a powerful approach for handling challenging numerical scenarios. In this study, classical ODE problems are employed as controlled testbeds to systematically evaluate the accuracy, training efficiency, and generalization capability under controlled conditions of the PINNs framework. Although not a universal solution, PINNs can achieve superior results by embedding physical laws directly into the learning process. We first analyze the existence and uniqueness properties of several benchmark problems and subsequently validate the PINNs methodology on these model systems. Our results demonstrate that for complex problems to converge to correct solutions, the loss function components data loss, initial condition loss, and residual loss must be appropriately balanced through careful weighting. We further establish that systematic tuning of hyperparameters, including network depth, layer width, activation functions, learning rate, optimization algorithms, w eight initialization schemes, and collocation point sampling, plays a crucial role in achieving accurate solutions. Additionally, embedding prior knowledge and imposing hard constraints on the network architecture, without loss the generality of the ODE system, significantly enhances the predictive capability of PINNs.
Authors: Kohei Tsuchiyama, Andre Roehm, Takatomo Mihana, Ryoichi Horisaki
Abstract: Physical Neural Networks (PNN) are promising platforms for next-generation computing systems. However, recent advances in digital neural network performance are largely driven by the rapid growth in the number of trainable parameters and, so far, demonstrated PNNs are lagging behind by several orders of magnitude in terms of scale. This mirrors size and performance constraints found in early digital neural networks. In that period, efficient reuse of parameters contributed to the development of parameter-efficient architectures such as convolutional neural networks. In this work, we numerically investigate hardware-friendly weight-tying for PNNs. Crucially, with many PNN systems, there is a time-scale separation between the fast dynamic active elements of the forward pass and the only slowly trainable elements implementing weights and biases. With this in mind,we propose the Reuse of Layers for eXpanding a Neural Network (ReLaX-Net) architecture, which employs a simple layer-by-layer time-multiplexing scheme to increase the effective network depth and efficiently use the number of parameters. We only require the addition of fast switches for existing PNNs. We validate ReLaX-Nets via numerical experiments on image classification and natural language processing tasks. Our results show that ReLaX-Net improves computational performance with only minor modifications to a conventional PNN. We observe a favorable scaling, where ReLaX-Nets exceed the performance of equivalent traditional RNNs or DNNs with the same number of parameters.
Authors: Omkar Kulkarni, Rohitash Chandra
Abstract: Financial fraud detection is critical for maintaining the integrity of financial systems, particularly in decentralised environments such as cryptocurrency networks. Although Graph Convolutional Networks (GCNs) are widely used for financial fraud detection, graph Transformer models such as Graph-BERT are gaining prominence due to their Transformer-based architecture, which mitigates issues such as over-smoothing. Graph-BERT is designed for static graphs and primarily evaluated on citation networks with undirected edges. However, financial transaction networks are inherently dynamic, with evolving structures and directed edges representing the flow of money. To address these challenges, we introduce DynBERG, a novel architecture that integrates Graph-BERT with a Gated Recurrent Unit (GRU) layer to capture temporal evolution over multiple time steps. Additionally, we modify the underlying algorithm to support directed edges, making DynBERG well-suited for dynamic financial transaction analysis. We evaluate our model on the Elliptic dataset, which includes Bitcoin transactions, including all transactions during a major cryptocurrency market event, the Dark Market Shutdown. By assessing DynBERG's resilience before and after this event, we analyse its ability to adapt to significant market shifts that impact transaction behaviours. Our model is benchmarked against state-of-the-art dynamic graph classification approaches, such as EvolveGCN and GCN, demonstrating superior performance, outperforming EvolveGCN before the market shutdown and surpassing GCN after the event. Additionally, an ablation study highlights the critical role of incorporating a time-series deep learning component, showcasing the effectiveness of GRU in modelling the temporal dynamics of financial transactions.
Authors: Yao Liu
Abstract: Accurate and robust weather forecasting remains a fundamental challenge due to the inherent spatio-temporal complexity of atmospheric systems. In this paper, we propose a novel self-supervised learning framework that leverages spatio-temporal structures to improve multi-variable weather prediction. The model integrates a graph neural network (GNN) for spatial reasoning, a self-supervised pretraining scheme for representation learning, and a spatio-temporal adaptation mechanism to enhance generalization across varying forecasting horizons. Extensive experiments on both ERA5 and MERRA-2 reanalysis datasets demonstrate that our approach achieves superior performance compared to traditional numerical weather prediction (NWP) models and recent deep learning methods. Quantitative evaluations and visual analyses in Beijing and Shanghai confirm the model's capability to capture fine-grained meteorological patterns. The proposed framework provides a scalable and label-efficient solution for future data-driven weather forecasting systems.
Authors: Dhananjaya Gowda, Seoha Song, Junhyun Lee, Harshith Goka
Abstract: As the large language models (LLMs) grow in size each day, efficient training and fine-tuning has never been as important as nowadays. This resulted in the great interest in parameter efficient fine-tuning (PEFT), and effective methods including low-rank adapters (LoRA) has emerged. Although the various PEFT methods have been studied extensively in the recent years, the greater part of the subject remains unexplored with the huge degree of freedom. In this paper, we propose FLoRA, a family of fused forward-backward adapters (FFBA) for parameter-efficient fine-tuning of LLMs on downstream tasks. The FFBA combine ideas from the popular LoRA and parallel adapters to improve the overall fine-tuning accuracies. At the same time, latencies are minimized by fusing the forward and backward adapters into existing projection layers of the base model. Experimental results show that the proposed FFB adapters perform significantly better than the popularly used LoRA in both accuracy and latency for a similar parameter budget.
Authors: Da Chang, Peng Xue, Yu Li, Yongxiang Liu, Pengxiang Xu, Shixun Zhang
Abstract: Parameter-Efficient Fine-Tuning (PEFT) methods are crucial for adapting large pre-trained models. Among these, LoRA is considered a foundational approach. Building on this, the influential DoRA method enhances performance by decomposing weight updates into magnitude and direction. However, its underlying mechanism remains unclear, and it introduces significant computational overhead. In this work, we first identify that DoRA's success stems from its capacity to increase the singular value entropy of the weight update matrix, which promotes a more uniform update distribution akin to full fine-tuning. We then reformulate DoRA into a mathematically equivalent and more efficient matrix form, revealing it as a learnable weight conditioning method. Based on this insight, we propose a unified framework for designing advanced PEFT methods by exploring two orthogonal dimensions: the architectural placement and the transformation type of the conditioning matrix. Within this framework, we introduce two novel methods: (1) \textbf{Pre-Diag}, which applies a diagonal conditioning matrix before the LoRA update to efficiently calibrate the pre-trained weights, thereby enhancing performance while reducing training time; and (2) \textbf{S}kewed \textbf{O}rthogonal \textbf{R}otation \textbf{A}daptation (\textbf{SORA}), which employs a parameter-efficient orthogonal rotation to perform a more powerful, norm-preserving transformation of the feature space. Extensive experiments on natural language understanding and generation tasks demonstrate that our proposed methods achieve superior performance and efficiency compared to both LoRA and DoRA. The code is available at https://github.com/MaeChd/SORA.
Authors: Federico Formica, Stefano Gregis, Aurora Francesca Zanenga, Andrea Rota, Mark Lawford, Claudio Menghi
Abstract: Understanding why neural networks make certain decisions is pivotal for their use in safety-critical applications. Feature-Guided Analysis (FGA) extracts slices of neural networks relevant to their tasks. Existing feature-guided approaches typically monitor the activation of the neural network neurons to extract the relevant rules. Preliminary results are encouraging and demonstrate the feasibility of this solution by assessing the precision and recall of Feature-Guided Analysis on two pilot case studies. However, the applicability in industrial contexts needs additional empirical evidence. To mitigate this need, this paper assesses the applicability of FGA on a benchmark made by the MNIST and LSC datasets. We assessed the effectiveness of FGA in computing rules that explain the behavior of the neural network. Our results show that FGA has a higher precision on our benchmark than the results from the literature. We also evaluated how the selection of the neural network architecture, training, and feature selection affect the effectiveness of FGA. Our results show that the selection significantly affects the recall of FGA, while it has a negligible impact on its precision.
Authors: Hao Wang, Licheng Pan, Yuan Lu, Zhichao Chen, Tianqiao Liu, Shuting He, Zhixuan Chu, Qingsong Wen, Haoxuan Li, Zhouchen Lin
Abstract: The design of training objective is central to training time-series forecasting models. Existing training objectives such as mean squared error mostly treat each future step as an independent, equally weighted task, which we found leading to the following two issues: (1) overlook the label autocorrelation effect among future steps, leading to biased training objective; (2) fail to set heterogeneous task weights for different forecasting tasks corresponding to varying future steps, limiting the forecasting performance. To fill this gap, we propose a novel quadratic-form weighted training objective, addressing both of the issues simultaneously. Specifically, the off-diagonal elements of the weighting matrix account for the label autocorrelation effect, whereas the non-uniform diagonals are expected to match the most preferable weights of the forecasting tasks with varying future steps. To achieve this, we propose a Quadratic Direct Forecast (QDF) learning algorithm, which trains the forecast model using the adaptively updated quadratic-form weighting matrix. Experiments show that our QDF effectively improves performance of various forecast models, achieving state-of-the-art results. Code is available at https://anonymous.4open.science/r/QDF-8937.
Authors: Gio Huh, Dhruv Sheth, Rayhan Zirvi, Frank Xiao
Abstract: While Vision-Language Models (VLMs) excel in many areas, they struggle with complex spatial reasoning, which requires problem decomposition and strategic tool use. Fine-tuning smaller, more deployable models offers an efficient path to strong performance, but this is hampered by a major bottleneck: the absence of high-quality, step-by-step reasoning data. To address this data-efficiency gap, we introduce SpatialTraceGen, a framework to distill the reasoning processes of a large teacher model into a high-quality dataset of multi-hop, multi-tool reasoning traces. A key innovation is our automated Verifier, which scalably ensures the fidelity of each reasoning step, providing a cost-effective alternative to manual human annotation. On the CLEVR-Humans benchmark, this verifier-guided process improves the average quality score of traces by 17\% while reducing quality variance by over 40\%. SpatialTraceGen delivers a dataset of expert traces, providing the structured, step-by-step examples of tool use necessary for effective fine-tuning and sample-efficient offline reinforcement learning.
Authors: Leonhard Duda, Khadijeh Alibabaei, Elena Vollmer, Leon Klug, Valentin Kozlov, Lisana Berberi, Mishal Benz, Rebekka Volk, Juan Pedro Guti\'errez Hermosillo Muriedas, Markus G\"otz, Judith S\'a\'inz-Pardo D\'iaz, \'Alvaro L\'opez Garc\'ia, Frank Schultmann, Achim Streit
Abstract: Federated Learning (FL) is an approach for training a shared Machine Learning (ML) model with distributed training data and multiple participants. FL allows bypassing limitations of the traditional Centralized Machine Learning CL if data cannot be shared or stored centrally due to privacy or technical restrictions -- the participants train the model locally with their training data and do not need to share it among the other participants. This paper investigates the practical implementation and effectiveness of FL in a real-world scenario, specifically focusing on unmanned aerial vehicle (UAV)-based thermal images for common thermal feature detection in urban environments. The distributed nature of the data arises naturally and makes it suitable for FL applications, as images captured in two German cities are available. This application presents unique challenges due to non-identical distribution and feature characteristics of data captured at both locations. The study makes several key contributions by evaluating FL algorithms in real deployment scenarios rather than simulation. We compare several FL approaches with a centralized learning baseline across key performance metrics such as model accuracy, training time, communication overhead, and energy usage. This paper also explores various FL workflows, comparing client-controlled workflows and server-controlled workflows. The findings of this work serve as a valuable reference for understanding the practical application and limitations of the FL methods in segmentation tasks in UAV-based imaging.
Authors: Yuxi Liu, Renjia Deng, Yutong He, Xue Wang, Tao Yao, Kun Yuan
Abstract: The substantial memory demands of pre-training and fine-tuning large language models (LLMs) require memory-efficient optimization algorithms. One promising approach is layer-wise optimization, which treats each transformer block as a single layer and optimizes it sequentially, while freezing the other layers to save optimizer states and activations. Although effective, these methods ignore the varying importance of the modules within each layer, leading to suboptimal performance. Moreover, layer-wise sampling provides only limited memory savings, as at least one full layer must remain active during optimization. To overcome these limitations, we propose Module-wise Importance SAmpling (MISA), a novel method that divides each layer into smaller modules and assigns importance scores to each module. MISA uses a weighted random sampling mechanism to activate modules, provably reducing gradient variance compared to layer-wise sampling. Additionally, we establish an \(\mathcal{O}(1/\sqrt{K})\) convergence rate under non-convex and stochastic conditions, where $K$ is the total number of block updates, and provide a detailed memory analysis showcasing MISA's superiority over existing baseline methods. Experiments on diverse learning tasks validate the effectiveness of MISA. Source code is available at https://github.com/pkumelon/MISA.
Authors: Aditya Singh, Zihang Wen, Srujananjali Medicherla, Adam Karvonen, Can Rager
Abstract: OthelloGPT, a transformer trained to predict valid moves in Othello, provides an ideal testbed for interpretability research. The model is complex enough to exhibit rich computational patterns, yet grounded in rule-based game logic that enables meaningful reverse-engineering. We present an automated approach based on decision trees to identify and interpret MLP neurons that encode rule-based game logic. Our method trains regression decision trees to map board states to neuron activations, then extracts decision paths where neurons are highly active to convert them into human-readable logical forms. These descriptions reveal highly interpretable patterns; for instance, neurons that specifically detect when diagonal moves become legal. Our findings suggest that roughly half of the neurons in layer 5 can be accurately described by compact, rule-based decision trees ($R^2 > 0.7$ for 913 of 2,048 neurons), while the remainder likely participate in more distributed or non-rule-based computations. We verify the causal relevance of patterns identified by our decision trees through targeted interventions. For a specific square, for specific game patterns, we ablate neurons corresponding to those patterns and find an approximately 5-10 fold stronger degradation in the model's ability to predict legal moves along those patterns compared to control patterns. To facilitate future work, we provide a Python tool that maps rule-based game behaviors to their implementing neurons, serving as a resource for researchers to test whether their interpretability methods recover meaningful computational structures.
Authors: Randolph Wiredu-Aidoo
Abstract: Clustering algorithms often rely on restrictive assumptions: K-Means and Gaussian Mixtures presuppose convex, Gaussian-like clusters, while DBSCAN and HDBSCAN capture non-convexity but can be highly sensitive. I introduce EVINGCA (Evolving Variance-Informed Nonparametric Graph Construction Algorithm), a density-variance based clustering algorithm that treats cluster formation as an adaptive, evolving process on a nearest-neighbor graph. EVINGCA expands rooted graphs via breadth-first search, guided by continuously updated local distance and shape statistics, replacing fixed density thresholds with local statistical feedback. With spatial indexing, EVINGCA features log-linear complexity in the average case and exhibits competitive performance against baselines across a variety of synthetic, real-world, low-d, and high-d datasets.
Authors: Kateryna Shapovalenko, Quentin Auster
Abstract: When we hear the word "house", we don't just process sound, we imagine walls, doors, memories. The brain builds meaning through layers, moving from raw acoustics to rich, multimodal associations. Inspired by this, we build on recent work from Meta that aligned EEG signals with averaged wav2vec2 speech embeddings, and ask a deeper question: which layers of pre-trained models best reflect this layered processing in the brain? We compare embeddings from two models: wav2vec2, which encodes sound into language, and CLIP, which maps words to images. Using EEG recorded during natural speech perception, we evaluate how these embeddings align with brain activity using ridge regression and contrastive decoding. We test three strategies: individual layers, progressive concatenation, and progressive summation. The findings suggest that combining multimodal, layer-aware representations may bring us closer to decoding how the brain understands language, not just as sound, but as experience.
Authors: Tue Le, Nghi D. Q. Bui, Linh Ngo Van, Trung Le
Abstract: Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful approach for strengthening the reasoning capabilities of large language models (LLMs). Among existing algorithms, Group Relative Policy Optimization (GRPO) has demonstrated strong performance, yet it suffers from a critical issue: low-probability tokens disproportionately dominate gradient updates due to their inherently large gradient magnitudes. This imbalance leads to unstable training and suppresses the contribution of high-probability tokens that are more reliable for learning. In this work, we introduce Token-Regulated Group Relative Policy Optimization (TR-GRPO), a simple yet effective extension of GRPO that assigns token-level weights positively correlated with the model's predicted probability. By downweighting low-probability tokens and emphasizing high-probability ones, TR-GRPO mitigates gradient over-amplification while preserving informative learning signals. Extensive experiments demonstrate that TR-GRPO consistently outperforms GRPO across RLVR tasks, including logic, math, and agentic reasoning, highlighting the importance of regulating token contributions during RL training and establishing TR-GRPO as a robust framework for enhancing LLM reasoning.
Authors: Zhixing Li, Arsham Gholamzadeh Khoee, Yinan Yu
Abstract: The objective of domain generalization (DG) is to enable models to be robust against domain shift. DG is crucial for deploying vision-language models (VLMs) in real-world applications, yet most existing methods rely on domain labels that may not be available and often ambiguous. We instead study the DG setting where models must generalize well without access to explicit domain labels. Our key idea is to represent an unseen target domain as a combination of latent domains automatically discovered from training data, enabling the model to adaptively transfer knowledge across domains. To realize this, we perform latent domain clustering on image features and fuse domain-specific text features based on the similarity between the input image and each latent domain. Experiments on four benchmarks show that this strategy yields consistent gains over VLM-based baselines and provides new insights into improving robustness under domain shift.
Authors: Muhammad Bilal Awan, Abdul Razzaq, Abdul Shahid
Abstract: This paper investigates the performance of Large Language Models (LLMs) as generative optimizers for solving constrained multi-objective regression tasks, specifically within the challenging domain of inverse design (property-to-structure mapping). This problem, critical to materials informatics, demands finding complex, feasible input vectors that lie on the Pareto optimal front. While LLMs have demonstrated universal effectiveness across generative and reasoning tasks, their utility in constrained, continuous, high-dimensional numerical spaces tasks they weren't explicitly architected for remains an open research question. We conducted a rigorous comparative study between established Bayesian Optimization (BO) frameworks and a suite of fine-tuned LLMs and BERT models. For BO, we benchmarked the foundational BoTorch Ax implementation against the state-of-the-art q-Expected Hypervolume Improvement (qEHVI, BoTorchM). The generative approach involved fine-tuning models via Parameter-Efficient Fine-Tuning (PEFT), framing the challenge as a regression problem with a custom output head. Our results show that BoTorch qEHVI achieved perfect convergence (GD=0.0), setting the performance ceiling. Crucially, the best-performing LLM (WizardMath-7B) achieved a Generational Distance (GD) of 1.21, significantly outperforming the traditional BoTorch Ax baseline (GD=15.03). We conclude that specialized BO frameworks remain the performance leader for guaranteed convergence, but fine-tuned LLMs are validated as a promising, computationally fast alternative, contributing essential comparative metrics to the field of AI-driven optimization. The findings have direct industrial applications in optimizing formulation design for resins, polymers, and paints, where multi-objective trade-offs between mechanical, rheological, and chemical properties are critical to innovation and production efficiency.
Authors: Ertugrul Mutlu
Abstract: This paper explores a deliberately over-engineered approach to the classical problem of parity detection -- determining whether a number is odd or even -- by combining wavelet-based feature extraction with unsupervised clustering. Instead of relying on modular arithmetic, integers are transformed into wavelet-domain representations, from which multi-scale statistical features are extracted and clustered using the k-means algorithm. The resulting feature space reveals meaningful structural differences between odd and even numbers, achieving a classification accuracy of approximately 69.67% without any label supervision. These results suggest that classical signal-processing techniques, originally designed for continuous data, can uncover latent structure even in purely discrete symbolic domains. Beyond parity detection, the study provides an illustrative perspective on how feature engineering and clustering may be repurposed for unconventional machine learning problems, potentially bridging symbolic reasoning and feature-based learning.
Authors: Zihao Wan, Pau Tong Lin Xu, Fuwen Luo, Ziyue Wang, Peng Li, Yang Liu
Abstract: While Vision-language Models (VLMs) have demonstrated strong semantic capabilities, their ability to interpret the underlying geometric structure of visual information is less explored. Pictographic characters, which combine visual form with symbolic structure, provide an ideal test case for this capability. We formulate this visual recognition challenge in the mathematical domain, where each character is represented by an executable program of geometric primitives. This is framed as a program synthesis task, training a VLM to decompile raster images into programs composed of B\'ezier curves. Our model, acting as a "visual decompiler", demonstrates performance superior to strong zero-shot baselines, including GPT-4o. The most significant finding is that when trained solely on modern Chinese characters, the model is able to reconstruct ancient Oracle Bone Script in a zero-shot context. This generalization provides strong evidence that the model acquires an abstract and transferable geometric grammar, moving beyond pixel-level pattern recognition to a more structured form of visual understanding.
Authors: Maximilian Willer, Peter Ruckdeschel
Abstract: flowengineR is an R package designed to provide a modular and extensible framework for building reproducible algorithmic workflows for general-purpose machine learning pipelines. It is motivated by the rapidly evolving field of algorithmic fairness where new metrics, mitigation strategies, and machine learning methods continuously emerge. A central challenge in fairness, but also far beyond, is that existing toolkits either focus narrowly on single interventions or treat reproducibility and extensibility as secondary considerations rather than core design principles. flowengineR addresses this by introducing a unified architecture of standardized engines for data splitting, execution, preprocessing, training, inprocessing, postprocessing, evaluation, and reporting. Each engine encapsulates one methodological task yet communicates via a lightweight interface, ensuring workflows remain transparent, auditable, and easily extensible. Although implemented in R, flowengineR builds on ideas from workflow languages (CWL, YAWL), graph-oriented visual programming languages (KNIME), and R frameworks (BatchJobs, batchtools). Its emphasis, however, is less on orchestrating engines for resilient parallel execution but rather on the straightforward setup and management of distinct engines as data structures. This orthogonalization enables distributed responsibilities, independent development, and streamlined integration. In fairness context, by structuring fairness methods as interchangeable engines, flowengineR lets researchers integrate, compare, and evaluate interventions across the modeling pipeline. At the same time, the architecture generalizes to explainability, robustness, and compliance metrics without core modifications. While motivated by fairness, it ultimately provides a general infrastructure for any workflow context where reproducibility, transparency, and extensibility are essential.
Authors: Shakib Khan, A. Ben Hamza, Amr Youssef
Abstract: Adversarial attacks present a significant risk to the integrity and performance of graph neural networks, particularly in tasks where graph structure and node features are vulnerable to manipulation. In this paper, we present a novel model, called fixed-point iterative graph convolutional network (Fix-GCN), which achieves robustness against adversarial perturbations by effectively capturing higher-order node neighborhood information in the graph without additional memory or computational complexity. Specifically, we introduce a versatile spectral modulation filter and derive the feature propagation rule of our model using fixed-point iteration. Unlike traditional defense mechanisms that rely on additional design elements to counteract attacks, the proposed graph filter provides a flexible-pass filtering approach, allowing it to selectively attenuate high-frequency components while preserving low-frequency structural information in the graph signal. By iteratively updating node representations, our model offers a flexible and efficient framework for preserving essential graph information while mitigating the impact of adversarial manipulation. We demonstrate the effectiveness of the proposed model through extensive experiments on various benchmark graph datasets, showcasing its resilience against adversarial attacks.
Authors: Jolanta \'Sliwa
Abstract: In recent years, the pen and paper RPG market has experienced significant growth. As a result, companies are increasingly exploring the integration of AI technologies to enhance player experience and gain a competitive edge. One of the key challenges faced by publishers is designing new opponents and estimating their challenge level. Currently, there are no automated methods for determining a monster's level; the only approaches used are based on manual testing and expert evaluation. Although these manual methods can provide reasonably accurate estimates, they are time-consuming and resource-intensive. Level prediction can be approached using ordinal regression techniques. This thesis presents an overview and evaluation of state-of-the-art methods for this task. It also details the construction of a dedicated dataset for level estimation. Furthermore, a human-inspired model was developed to serve as a benchmark, allowing comparison between machine learning algorithms and the approach typically employed by pen and paper RPG publishers. In addition, a specialized evaluation procedure, grounded in domain knowledge, was designed to assess model performance and facilitate meaningful comparisons.
Authors: Peilin Tan, Chuanqi Shi, Dian Tu, Liang Xie
Abstract: Stock trend prediction is crucial for profitable trading strategies and portfolio management yet remains challenging due to market volatility, complex temporal dynamics and multifaceted inter-stock relationships. Existing methods struggle to effectively capture temporal dependencies and dynamic inter-stock interactions, often neglecting cross-sectional market influences, relying on static correlations, employing uniform treatments of nodes and edges, and conflating diverse relationships. This work introduces MaGNet, a novel Mamba dual-hyperGraph Network for stock prediction, integrating three key innovations: (1) a MAGE block, which leverages bidirectional Mamba with adaptive gating mechanisms for contextual temporal modeling and integrates a sparse Mixture-of-Experts layer to enable dynamic adaptation to diverse market conditions, alongside multi-head attention for capturing global dependencies; (2) Feature-wise and Stock-wise 2D Spatiotemporal Attention modules enable precise fusion of multivariate features and cross-stock dependencies, effectively enhancing informativeness while preserving intrinsic data structures, bridging temporal modeling with relational reasoning; and (3) a dual hypergraph framework consisting of the Temporal-Causal Hypergraph (TCH) that captures fine-grained causal dependencies with temporal constraints, and Global Probabilistic Hypergraph (GPH) that models market-wide patterns through soft hyperedge assignments and Jensen-Shannon Divergence weighting mechanism, jointly disentangling localized temporal influences from instantaneous global structures for multi-scale relational learning. Extensive experiments on six major stock indices demonstrate MaGNet outperforms state-of-the-art methods in both superior predictive performance and exceptional investment returns with robust risk management capabilities. Codes available at: https://github.com/PeilinTime/MaGNet.
Authors: Fali Wang, Jihai Chen, Shuhua Yang, Runxue Bao, Tianxiang Zhao, Zhiwei Zhang, Xianfeng Tang, Hui Liu, Qi He, Suhang Wang
Abstract: Test-Time Scaling (TTS) improves large language models (LLMs) by allocating additional computation during inference, typically through parallel, sequential, or hybrid scaling. However, prior studies often assume fixed collaboration architectures (e.g., topologies) and single-model usage, overlooking that optimal architectures and model combinations can vary across tasks. Therefore, we study the novel problem of searching for compute-optimal model combinations and architectures in TTS under a fixed budget. We formalize it as a multi-LLM collaboration graph, where nodes encode roles and LLM model assignments, and edges capture information flow. This problem is challenging because (i) the combinatorial search space is prohibitively large, and (ii) task-specific requirements demand tailored designs. To address these, we reformulate the problem as probabilistic graph optimization and, through pilot experiments, derive three empirical insights into TTS collaboration graphs. Guided by these insights, we propose Agent-REINFORCE, an LLM-agent-augmented framework that mirrors the REINFORCE pipeline by mapping sampling-gradient-update to sampling-feedback-update, where feedback serves as a textual gradient to update the probabilistic graph and efficiently search for optimal multi-LLM collaboration graphs. Experiments show that Agent-REINFORCE outperforms both traditional and LLM-based baselines in sample efficiency and search performance, and effectively identifies optimal graphs under joint objectives of accuracy and inference latency.
Authors: Zihao Guo, Qingyun Sun, Ziwei Zhang, Haonan Yuan, Huiping Zhuang, Xingcheng Fu, Jianxin Li
Abstract: Graph incremental learning (GIL), which continuously updates graph models by sequential knowledge acquisition, has garnered significant interest recently. However, existing GIL approaches focus on task-incremental and class-incremental scenarios within a single domain. Graph domain-incremental learning (Domain-IL), aiming at updating models across multiple graph domains, has become critical with the development of graph foundation models (GFMs), but remains unexplored in the literature. In this paper, we propose Graph Domain-Incremental Learning via Knowledge Dientanglement and Preservation (GraphKeeper), to address catastrophic forgetting in Domain-IL scenario from the perspectives of embedding shifts and decision boundary deviations. Specifically, to prevent embedding shifts and confusion across incremental graph domains, we first propose the domain-specific parameter-efficient fine-tuning together with intra- and inter-domain disentanglement objectives. Consequently, to maintain a stable decision boundary, we introduce deviation-free knowledge preservation to continuously fit incremental domains. Additionally, for graphs with unobservable domains, we perform domain-aware distribution discrimination to obtain precise embeddings. Extensive experiments demonstrate the proposed GraphKeeper achieves state-of-the-art results with 6.5%~16.6% improvement over the runner-up with negligible forgetting. Moreover, we show GraphKeeper can be seamlessly integrated with various representative GFMs, highlighting its broad applicative potential.
Authors: Marios Impraimakis, Evangelia Nektaria Palkanoglou
Abstract: The optimization-based damage detection and damage state digital twinning capabilities are examined here of a novel conditional-labeled generative adversarial network methodology. The framework outperforms current approaches for fault anomaly detection as no prior information is required for the health state of the system: a topic of high significance for real-world applications. Specifically, current artificial intelligence-based digital twinning approaches suffer from the uncertainty related to obtaining poor predictions when a low number of measurements is available, physics knowledge is missing, or when the damage state is unknown. To this end, an unsupervised framework is examined and validated rigorously on the benchmark structural health monitoring measurements of Z24 Bridge: a post-tensioned concrete highway bridge in Switzerland. In implementing the approach, firstly, different same damage-level measurements are used as inputs, while the model is forced to converge conditionally to two different damage states. Secondly, the process is repeated for a different group of measurements. Finally, the convergence scores are compared to identify which one belongs to a different damage state. The process for both healthy-to-healthy and damage-to-healthy input data creates, simultaneously, measurements for digital twinning purposes at different damage states, capable of pattern recognition and machine learning data generation. Further to this process, a support vector machine classifier and a principal component analysis procedure is developed to assess the generated and real measurements of each damage category, serving as a secondary new dynamics learning indicator in damage scenarios. Importantly, the approach is shown to capture accurately damage over healthy measurements, providing a powerful tool for vibration-based system-level monitoring and scalable infrastructure resilience.
Authors: Marios Impraimakis
Abstract: The dynamic structural load identification capabilities of the gated recurrent unit, long short-term memory, and convolutional neural networks are examined herein. The examination is on realistic small dataset training conditions and on a comparative view to the physics-based residual Kalman filter (RKF). The dynamic load identification suffers from the uncertainty related to obtaining poor predictions when in civil engineering applications only a low number of tests are performed or are available, or when the structural model is unidentifiable. In considering the methods, first, a simulated structure is investigated under a shaker excitation at the top floor. Second, a building in California is investigated under seismic base excitation, which results in loading for all degrees of freedom. Finally, the International Association for Structural Control-American Society of Civil Engineers (IASC-ASCE) structural health monitoring benchmark problem is examined for impact and instant loading conditions. Importantly, the methods are shown to outperform each other on different loading scenarios, while the RKF is shown to outperform the networks in physically parametrized identifiable cases.
Authors: Yuchen Zhang, Hanyue Du, Chun Cao, Jingwei Xu
Abstract: Low-Rank Adaptation (LoRA) has become a widely adopted parameter-efficient fine-tuning (PEFT) technique for adapting large language models (LLMs) to downstream tasks. While prior work has explored strategies for integrating LLM training and serving, there still remains a gap in unifying fine-tuning and inference for LoRA-based models. We present Loquetier, a virtualized multi-LoRA framework that seamlessly integrates LoRA fine-tuning and serving within a single runtime. Loquetier introduces two key components: (1) a Virtualized Module that isolates PEFT-based modifications and supports multiple adapters on a shared base model, and (2) an optimized computation flow with a kernel design that merges fine-tuning and inference paths in forward propagation, enabling efficient batching and minimizing kernel invocation overhead. Extensive experiments across three task settings show that Loquetier consistently outperforms existing baselines in both performance and flexibility, achieving up to $3.0\times$ the throughput of the state-of-the-art co-serving system on inference-only tasks and $46.4\times$ higher SLO attainment than PEFT on unified fine-tuning and inference tasks. The implementation of Loquetier is publicly available at https://github.com/NJUDeepEngine/Loquetier.
Authors: Vivan Doshi
Abstract: The discovery of conservation laws is a cornerstone of scientific progress. However, identifying these invariants from observational data remains a significant challenge. We propose a hybrid framework to automate the discovery of conserved quantities from noisy trajectory data. Our approach integrates three components: (1) a Neural Ordinary Differential Equation (Neural ODE) that learns a continuous model of the system's dynamics, (2) a Transformer that generates symbolic candidate invariants conditioned on the learned vector field, and (3) a symbolic-numeric verifier that provides a strong numerical certificate for the validity of these candidates. We test our framework on canonical physical systems and show that it significantly outperforms baselines that operate directly on trajectory data. This work demonstrates the robustness of a decoupled learn-then-search approach for discovering mathematical principles from imperfect data.
Authors: Yi Zhang, Che Liu, Xiancong Ren, Hanchu Ni, Shuai Zhang, Zeyuan Ding, Jiayu Hu, Hanzhe Shan, Zhenwei Niu, Zhaoyang Liu, Yue Zhao, Junbo Qi, Qinfan Zhang, Dengjie Li, Yidong Wang, Jiachen Luo, Yong Dai, Jian Tang, Xiaozhu Ju
Abstract: This report presents Pelican-VL 1.0, a new family of open-source embodied brain models with parameter scales ranging from 7 billion to 72 billion. Our explicit mission is clearly stated as: To embed powerful intelligence into various embodiments. Pelican-VL 1.0 is currently the largest-scale open-source embodied multimodal brain model. Its core advantage lies in the in-depth integration of data power and intelligent adaptive learning mechanisms. Specifically, metaloop distilled a high-quality dataset from a raw dataset containing 4+ billion tokens. Pelican-VL 1.0 is trained on a large-scale cluster of 1000+ A800 GPUs, consuming over 50k+ A800 GPU-hours per checkpoint. This translates to a 20.3% performance uplift from its base model and outperforms 100B-level open-source counterparts by 10.6%, placing it on par with leading proprietary systems on well-known embodied benchmarks. We establish a novel framework, DPPO (Deliberate Practice Policy Optimization), inspired by human metacognition to train Pelican-VL 1.0. We operationalize this as a metaloop that teaches the AI to practice deliberately, which is a RL-Refine-Diagnose-SFT loop.
Authors: Huseyin Goksu
Abstract: Spectral Graph Neural Networks (GNNs) have achieved state-of-the-art results by defining graph convolutions in the spectral domain. A common approach, popularized by ChebyNet, is to use polynomial filters based on continuous orthogonal polynomials (e.g., Chebyshev). This creates a theoretical disconnect, as these continuous-domain filters are applied to inherently discrete graph structures. We hypothesize this mismatch can lead to suboptimal performance and fragility to hyperparameter settings. In this paper, we introduce MeixnerNet, a novel spectral GNN architecture that employs discrete orthogonal polynomials -- specifically, the Meixner polynomials $M_k(x; \beta, c)$. Our model makes the two key shape parameters of the polynomial, beta and c, learnable, allowing the filter to adapt its polynomial basis to the specific spectral properties of a given graph. We overcome the significant numerical instability of these polynomials by introducing a novel stabilization technique that combines Laplacian scaling with per-basis LayerNorm. We demonstrate experimentally that MeixnerNet achieves competitive-to-superior performance against the strong ChebyNet baseline at the optimal K = 2 setting (winning on 2 out of 3 benchmarks). More critically, we show that MeixnerNet is exceptionally robust to variations in the polynomial degree K, a hyperparameter to which ChebyNet proves to be highly fragile, collapsing in performance where MeixnerNet remains stable.
Authors: Avisek Naug, Antonio Guillen, Vineet Kumar, Scott Greenwood, Wesley Brewer, Sahand Ghorbanpour, Ashwin Ramesh Babu, Vineet Gundecha, Ricardo Luna Gutierrez, Soumyendu Sarkar
Abstract: Liquid cooling is critical for thermal management in high-density data centers with the rising AI workloads. However, machine learning-based controllers are essential to unlock greater energy efficiency and reliability, promoting sustainability. We present LC-Opt, a Sustainable Liquid Cooling (LC) benchmark environment, for reinforcement learning (RL) control strategies in energy-efficient liquid cooling of high-performance computing (HPC) systems. Built on the baseline of a high-fidelity digital twin of Oak Ridge National Lab's Frontier Supercomputer cooling system, LC-Opt provides detailed Modelica-based end-to-end models spanning site-level cooling towers to data center cabinets and server blade groups. RL agents optimize critical thermal controls like liquid supply temperature, flow rate, and granular valve actuation at the IT cabinet level, as well as cooling tower (CT) setpoints through a Gymnasium interface, with dynamic changes in workloads. This environment creates a multi-objective real-time optimization challenge balancing local thermal regulation and global energy efficiency, and also supports additional components like a heat recovery unit (HRU). We benchmark centralized and decentralized multi-agent RL approaches, demonstrate policy distillation into decision and regression trees for interpretable control, and explore LLM-based methods that explain control actions in natural language through an agentic mesh architecture designed to foster user trust and simplify system management. LC-Opt democratizes access to detailed, customizable liquid cooling models, enabling the ML community, operators, and vendors to develop sustainable data center liquid cooling control solutions.
Authors: Antonio Guillen-Perez, Avisek Naug, Vineet Gundecha, Sahand Ghorbanpour, Ricardo Luna Gutierrez, Ashwin Ramesh Babu, Munther Salim, Shubhanker Banerjee, Eoin H. Oude Essink, Damien Fay, Soumyendu Sarkar
Abstract: The increasing energy demands and carbon footprint of large-scale AI require intelligent workload management in globally distributed data centers. Yet progress is limited by the absence of benchmarks that realistically capture the interplay of time-varying environmental factors (grid carbon intensity, electricity prices, weather), detailed data center physics (CPUs, GPUs, memory, HVAC energy), and geo-distributed network dynamics (latency and transmission costs). To bridge this gap, we present DCcluster-Opt: an open-source, high-fidelity simulation benchmark for sustainable, geo-temporal task scheduling. DCcluster-Opt combines curated real-world datasets, including AI workload traces, grid carbon intensity, electricity markets, weather across 20 global regions, cloud transmission costs, and empirical network delay parameters with physics-informed models of data center operations, enabling rigorous and reproducible research in sustainable computing. It presents a challenging scheduling problem where a top-level coordinating agent must dynamically reassign or defer tasks that arrive with resource and service-level agreement requirements across a configurable cluster of data centers to optimize multiple objectives. The environment also models advanced components such as heat recovery. A modular reward system enables an explicit study of trade-offs among carbon emissions, energy costs, service level agreements, and water use. It provides a Gymnasium API with baseline controllers, including reinforcement learning and rule-based strategies, to support reproducible ML research and a fair comparison of diverse algorithms. By offering a realistic, configurable, and accessible testbed, DCcluster-Opt accelerates the development and validation of next-generation sustainable computing solutions for geo-distributed data centers.
Authors: Shoma Yagi, Jun Ichikawa, Genki Ichinose
Abstract: In football, attacking teams attempt to break through the opponent's defensive line to create scoring opportunities. This action, known as a Line Break, is a critical indicator of offensive effectiveness and tactical performance, yet previous studies have mainly focused on shots or goal opportunities rather than on how teams break the defensive line. In this study, we develop a machine learning model to predict Line Breaks using event and tracking data from the 2023 J1 League season. The model incorporates 189 features, including player positions, velocities, and spatial configurations, and employs an XGBoost classifier to estimate the probability of Line Breaks. The proposed model achieved high predictive accuracy, with an AUC of 0.982 and a Brier score of 0.015. Furthermore, SHAP analysis revealed that factors such as offensive player speed, gaps in the defensive line, and offensive players' spatial distributions significantly contribute to the occurrence of Line Breaks. Finally, we found a moderate positive correlation between the predicted probability of being Line-Broken and the number of shots and crosses conceded at the team level. These results suggest that Line Breaks are closely linked to the creation of scoring opportunities and provide a quantitative framework for understanding tactical dynamics in football.
Authors: Sai Niranjan Ramachandran, Manish Krishan Lal, Suvrit Sra
Abstract: We analyse how the sampling dynamics of distributions evolve in score-based diffusion models using cross-fluctuations, a centered-moment statistic from statistical physics. Specifically, we show that starting from an unbiased isotropic normal distribution, samples undergo sharp, discrete transitions, eventually forming distinct events of a desired distribution while progressively revealing finer structure. As this process is reversible, these transitions also occur in reverse, where intermediate states progressively merge, tracing a path back to the initial distribution. We demonstrate that these transitions can be detected as discontinuities in $n^{\text{th}}$-order cross-fluctuations. For variance-preserving SDEs, we derive a closed-form for these cross-fluctuations that is efficiently computable for the reverse trajectory. We find that detecting these transitions directly boosts sampling efficiency, accelerates class-conditional and rare-class generation, and improves two zero-shot tasks--image classification and style transfer--without expensive grid search or retraining. We also show that this viewpoint unifies classical coupling and mixing from finite Markov chains with continuous dynamics while extending to stochastic SDEs and non Markovian samplers. Our framework therefore bridges discrete Markov chain theory, phase analysis, and modern generative modeling.
Authors: Lu Bowen
Abstract: Recent deep trajectory predictors (e.g., Jiang et al., 2023; Zhou et al., 2022) have achieved strong average accuracy but remain unreliable in complex long-tail driving scenarios. These limitations reveal the weakness of the prevailing "one-model-fits-all" paradigm, particularly in safety-critical urban contexts where simpler physics-based models can occasionally outperform advanced networks (Kalman, 1960). To bridge this gap, we propose a dynamic multi-expert gating framework that adaptively selects the most reliable trajectory predictor among a physics-informed LSTM, a Transformer, and a fine-tuned GameFormer on a per-sample basis. Our method leverages internal model signals (meta-features) such as stability and uncertainty (Gal and Ghahramani, 2016), which we demonstrate to be substantially more informative than geometric scene descriptors. To the best of our knowledge, this is the first work to formulate trajectory expert selection as a pairwise-ranking problem over internal model signals (Burges et al., 2005), directly optimizing decision quality without requiring post-hoc calibration. Evaluated on the nuPlan-mini dataset (Caesar et al., 2021) with 1,287 samples, our LLM-enhanced tri-expert gate achieves a Final Displacement Error (FDE) of 2.567 m, representing a 9.5 percent reduction over GameFormer (2.835 m), and realizes 57.8 percent of the oracle performance bound. In open-loop simulations, after trajectory horizon alignment, the same configuration reduces FDE on left-turn scenarios by approximately 10 percent, demonstrating consistent improvements across both offline validation and open-loop evaluation. These results indicate that adaptive hybrid systems enhance trajectory reliability in safety-critical autonomous driving, providing a practical pathway beyond static single-model paradigms.
Authors: Siyu Xiao, Xindi Zhao, Tianhao Mao, Yiwei Wang, Yuqiao Chen, Hongyun Zhang, Jian Wang, Junjie Wang, Shuang Liu, Tupei Chen, Yang Liu
Abstract: Accurate downhole depth measurement is essential for oil and gas well operations, directly influencing reservoir contact, production efficiency, and operational safety. Collar correlation using a casing collar locator (CCL) is fundamental for precise depth calibration. While neural network-based CCL signal recognition has achieved significant progress in collar identification, preprocessing methods for such applications remain underdeveloped. Moreover, the limited availability of real well data poses substantial challenges for training neural network models that require extensive datasets. This paper presents a system integrated into downhole tools for CCL signal acquisition to facilitate dataset construction. We propose comprehensive preprocessing methods for data augmentation and evaluate their effectiveness using our AlexNet-based neural network models. Through systematic experimentation across various configuration combinations, we analyze the contribution of each augmentation method. Results demonstrate that standardization, label distribution smoothing (LDS), and random cropping are fundamental requirements for model training, while label smoothing regularization (LSR), time scaling, and multiple sampling significantly enhance model generalization capability. The F1 scores of our two benchmark models trained with the proposed augmentation methods maximumly improve from 0.937 and 0.952 to 1.0 and 1.0, respectively. Performance validation on real CCL waveforms confirms the effectiveness and practical applicability of our approach. This work addresses the gaps in data augmentation methodologies for training casing collar recognition models in CCL data-limited environments.
Authors: Bernd Bohnet, Rumen Dangovski, Kevin Swersky, Sherry Moore, Arslan Chaudhry, Kathleen Kenealy, Noah Fiedel
Abstract: The remarkable capabilities of Large Language Models (LLMs) often need to be tailored for specific applications, requiring the integration of new knowledge or the acquisition of new skills. While full fine-tuning is a powerful adaptation method, it is computationally expensive and can lead to a degradation of general reasoning abilities, a phenomenon known as catastrophic forgetting. A range of alternative techniques exists, each with its own trade-offs. In-Context Learning (ICL) is fast but limited by context length, while Parameter-Efficient Fine-Tuning (PEFT) methods like Low-Rank Adaptation (LoRA) offer a middle ground by minimizing parameter changes. However, the challenge of catastrophic forgetting persists, raising questions about the best adaptation strategy for a given task. This paper presents a comparative analysis of Supervised Finetuning (SFT), LoRA, and ICL in data-scarce scenarios. We find that LoRA provides the most effective balance, successfully instilling new skills with minimal impact on the base model's general knowledge. In contrast, while SFT excels at skill acquisition, it is highly susceptible to catastrophic forgetting. ICL is effective for incorporating factual knowledge but struggles with complex skills. Our findings offer a practical framework for selecting an LLM adaptation strategy. We highlight the critical distinction between skill acquisition and knowledge integration, clarify the trade-offs between task-specific performance and the preservation of general capabilities.
Authors: Kowshik Balasubramanian, Andre Williams, Ismail Butun
Abstract: This paper introduces a novel framework for enhancing Random Forest classifiers by integrating probabilistic feature sampling and hyperparameter tuning via Simulated Annealing. The proposed framework exhibits substantial advancements in predictive accuracy and generalization, adeptly tackling the multifaceted challenges of robust classification across diverse domains, including credit risk evaluation, anomaly detection in IoT ecosystems, early-stage medical diagnostics, and high-dimensional biological data analysis. To overcome the limitations of conventional Random Forests, we present an approach that places stronger emphasis on capturing the most relevant signals from data while enabling adaptive hyperparameter configuration. The model is guided towards features that contribute more meaningfully to classification and optimizing this with dynamic parameter tuning. The results demonstrate consistent accuracy improvements and meaningful insights into feature relevance, showcasing the efficacy of combining importance aware sampling and metaheuristic optimization.
Authors: Angana Borah, Adrija Datta, Ashish S. Kumar, Raviraj Dave, Udit Bhatia
Abstract: Efforts to green cities for cooling are succeeding unevenly because the same vegetation that cools surfaces can also intensify how hot the air feels. Previous studies have identified humid heat as a growing urban hazard, yet how physiologically active vegetation governs this trade-off between cooling and moisture accumulation remains poorly understood, leaving mitigation policy and design largely unguided. Here we quantify how vegetation structure and function influence the Heat Index (HI), a combined measure of temperature and humidity in 138 Indian cities spanning tropical savanna, semi-arid steppe, and humid subtropical climates, and across dense urban cores and semi-urban rings. Using an extreme-aware, one kilometre reconstruction of HI and an interpretable machine-learning framework that integrates SHapley Additive Explanations (SHAP) and Accumulated Local Effects (ALE), we isolate vegetation-climate interactions. Cooling generally strengthens for EVI >= 0.4 and LAI >= 0.05, but joint-high regimes begin to reverse toward warming when EVI >= 0.5, LAI >= 0.2, and fPAR >= 0.5,with an earlier onset for fPAR >= 0.25 in humid, dense cores. In such environments, highly physiologically active vegetation elevates near-surface humidity faster than it removes heat, reversing its cooling effect and amplifying perceived heat stress. These findings establish the climatic limits of vegetation-driven cooling and provide quantitative thresholds for climate-specific greening strategies that promote equitable and heat-resilient cities.
Authors: Qing Guo, Xinhang Li, Junyu Chen, Zheng Guo, Xiaocong Li, Lin Zhang, Lei Li
Abstract: Leveraging large language models (LLMs) in traffic signal control (TSC) improves optimization efficiency and interpretability compared to traditional reinforcement learning (RL) methods. However, existing LLM-based approaches are limited by fixed time signal durations and are prone to hallucination errors, while RL methods lack robustness in signal timing decisions and suffer from poor generalization. To address these challenges, this paper proposes HeraldLight, a dual LLMs architecture enhanced by Herald guided prompts. The Herald Module extracts contextual information and forecasts queue lengths for each traffic phase based on real-time conditions. The first LLM, LLM-Agent, uses these forecasts to make fine grained traffic signal control, while the second LLM, LLM-Critic, refines LLM-Agent's outputs, correcting errors and hallucinations. These refined outputs are used for score-based fine-tuning to improve accuracy and robustness. Simulation experiments using CityFlow on real world datasets covering 224 intersections in Jinan (12), Hangzhou (16), and New York (196) demonstrate that HeraldLight outperforms state of the art baselines, achieving a 20.03% reduction in average travel time across all scenarios and a 10.74% reduction in average queue length on the Jinan and Hangzhou scenarios. The source code is available on GitHub: https://github.com/BUPT-ANTlab/HeraldLight.
Authors: Shiman Zhang, Jinghan Zhou, Zhoufan Yu, Ningai Leng
Abstract: To improve decision-making and planning efficiency in back-end centralized redundant supply chains, this paper proposes a decision model integrating deep learning with intelligent particle swarm optimization. A distributed node deployment model and optimal planning path are constructed for the supply chain network. Deep learning such as convolutional neural networks extracts features from historical data, and linear programming captures high-order statistical features. The model is optimized using fuzzy association rule scheduling and deep reinforcement learning, while neural networks fit dynamic changes. A hybrid mechanism of "deep learning feature extraction - intelligent particle swarm optimization" guides global optimization and selects optimal decisions for adaptive control. Simulations show reduced resource consumption, enhanced spatial planning, and in dynamic environments improved real-time decision adjustment, distribution path optimization, and robust intelligent control.
Authors: Hiba Ahsan, Byron C. Wallace
Abstract: LLMs are increasingly being used in healthcare. This promises to free physicians from drudgery, enabling better care to be delivered at scale. But the use of LLMs in this space also brings risks; for example, such models may worsen existing biases. How can we spot when LLMs are (spuriously) relying on patient race to inform predictions? In this work we assess the degree to which Sparse Autoencoders (SAEs) can reveal (and control) associations the model has made between race and stigmatizing concepts. We first identify SAE latents in Gemma-2 models which appear to correlate with Black individuals. We find that this latent activates on reasonable input sequences (e.g., "African American") but also problematic words like "incarceration". We then show that we can use this latent to steer models to generate outputs about Black patients, and further that this can induce problematic associations in model outputs as a result. For example, activating the Black latent increases the risk assigned to the probability that a patient will become "belligerent". We evaluate the degree to which such steering via latents might be useful for mitigating bias. We find that this offers improvements in simple settings, but is less successful for more realistic and complex clinical tasks. Overall, our results suggest that: SAEs may offer a useful tool in clinical applications of LLMs to identify problematic reliance on demographics but mitigating bias via SAE steering appears to be of marginal utility for realistic tasks.
Authors: Shaghayegh Fazliani, Madeleine Udell
Abstract: Current LLM-driven approaches using test-time computing to generate PDE solvers execute a large number of solver samples to identify high-accuracy solvers. These paradigms are especially costly for complex PDEs requiring substantial computational resources for numerical evaluation. We introduce PDE-SHARP, a framework to reduce computational costs by replacing expensive scientific computation by cheaper LLM inference that achieves superior solver accuracy with 60-75% fewer computational evaluations. PDE-SHARP employs three stages: (1) Analysis: mathematical chain-of-thought analysis including PDE classification, solution type detection, and stability analysis; (2) Genesis: solver generation based on mathematical insights from the previous stage; and (3) Synthesis: collaborative selection-hybridization tournaments in which LLM judges iteratively refine implementations through flexible performance feedback. To generate high-quality solvers, PDE-SHARP requires fewer than 13 solver evaluations on average compared to 30+ for baseline methods, improving accuracy uniformly across tested PDEs by $4\times$ on average, and demonstrates robust performance across LLM architectures, from general-purpose to specialized reasoning models.
Authors: Ali Satvaty, Suzan Verberne, Fatih Turkmen
Abstract: Membership inference attacks (MIA) aim to infer whether a particular data point is part of the training dataset of a model. In this paper, we propose a new task in the context of LLM privacy: entity-level discovery of membership risk focused on sensitive information (PII, credit card numbers, etc). Existing methods for MIA can detect the presence of entire prompts or documents in the LLM training data, but they fail to capture risks at a finer granularity. We propose the ``EL-MIA'' framework for auditing entity-level membership risks in LLMs. We construct a benchmark dataset for the evaluation of MIA methods on this task. Using this benchmark, we conduct a systematic comparison of existing MIA techniques as well as two newly proposed methods. We provide a comprehensive analysis of the results, trying to explain the relation of the entity level MIA susceptability with the model scale, training epochs, and other surface level factors. Our findings reveal that existing MIA methods are limited when it comes to entity-level membership inference of the sensitive attributes, while this susceptibility can be outlined with relatively straightforward methods, highlighting the need for stronger adversaries to stress test the provided threat model.
Authors: David L\"udke, Tom Wollschl\"ager, Paul Ungermann, Stephan G\"unnemann, Leo Schwinn
Abstract: We introduce a novel framework that transforms the resource-intensive (adversarial) prompt optimization problem into an \emph{efficient, amortized inference task}. Our core insight is that pretrained, non-autoregressive generative LLMs, such as Diffusion LLMs, which model the joint distribution over prompt-response pairs, can serve as powerful surrogates for prompt search. This approach enables the direct conditional generation of prompts, effectively replacing costly, per-instance discrete optimization with a small number of parallelizable samples. We provide a probabilistic analysis demonstrating that under mild fidelity assumptions, only a few conditional samples are required to recover high-reward (harmful) prompts. Empirically, we find that the generated prompts are low-perplexity, diverse jailbreaks that exhibit strong transferability to a wide range of black-box target models, including robustly trained and proprietary LLMs. Beyond adversarial prompting, our framework opens new directions for red teaming, automated prompt optimization, and leveraging emerging Flow- and Diffusion-based LLMs.
Authors: Yiquan Wang, Yahui Ma, Yuhan Chang, Jiayao Yan, Jialin Zhang, Minnuo Cai, Kai Wei
Abstract: Diffusion models have emerged as a leading framework in generative modeling, showing significant potential to accelerate and transform the traditionally slow and costly process of drug discovery. This review provides a systematic comparison of their application in designing two principal therapeutic modalities: small molecules and therapeutic peptides. We analyze how a unified framework of iterative denoising is adapted to the distinct molecular representations, chemical spaces, and design objectives of each modality. For small molecules, these models excel at structure-based design, generating novel, pocket-fitting ligands with desired physicochemical properties, yet face the critical hurdle of ensuring chemical synthesizability. Conversely, for therapeutic peptides, the focus shifts to generating functional sequences and designing de novo structures, where the primary challenges are achieving biological stability against proteolysis, ensuring proper folding, and minimizing immunogenicity. Despite these distinct challenges, both domains face shared hurdles: the need for more accurate scoring functions, the scarcity of high-quality experimental data, and the crucial requirement for experimental validation. We conclude that the full potential of diffusion models will be unlocked by bridging these modality-specific gaps and integrating them into automated, closed-loop Design-Build-Test-Learn (DBTL) platforms, thereby shifting the paradigm from chemical exploration to the targeted creation of novel therapeutics.
Authors: Pouya M. Ghari, Simone Sciabola, Ye Wang
Abstract: Fine-tuning foundation models has emerged as a powerful approach for generating objects with specific desired properties. Reinforcement learning (RL) provides an effective framework for this purpose, enabling models to generate outputs that maximize a given reward function. However, in many applications such as text generation and drug discovery, it can be suboptimal to optimize using a single reward signal, as multiple evaluation criteria are often necessary. This paper proposes a novel reinforcement learning-based method for fine-tuning foundation models using multiple reward signals. By employing an iterative fine-tuning strategy across these rewards, our approach generalizes state-of-the-art RL-based methods. We further provide a theoretical analysis that offers insights into the performance of multi-reward RL fine-tuning. Experimental results across diverse domains including text, biological sequence, and small molecule generation, demonstrate the effectiveness of the proposed algorithm compared to state-of-the-art baselines.
Authors: Wadduwage Shanika Perera, ABM Islam, Van Vung Pham, Min Kyung An
Abstract: Melanoma is one of the most aggressive and deadliest skin cancers, leading to mortality if not detected and treated in the early stages. Artificial intelligence techniques have recently been developed to help dermatologists in the early detection of melanoma, and systems based on deep learning (DL) have been able to detect these lesions with high accuracy. However, the entire community must overcome the explainability limit to get the maximum benefit from DL for diagnostics in the healthcare domain. Because of the black box operation's shortcomings in DL models' decisions, there is a lack of reliability and trust in the outcomes. However, Explainable Artificial Intelligence (XAI) can solve this problem by interpreting the predictions of AI systems. This paper proposes a machine learning model using ensemble learning of three state-of-the-art deep transfer Learning networks, along with an approach to ensure the reliability of the predictions by utilizing XAI techniques to explain the basis of the predictions.
Authors: Zachary Chase, Shinji Ito, Idan Mehalel
Abstract: We determine the minimax optimal expected regret in the classic non-stochastic multi-armed bandit with expert advice problem, by proving a lower bound that matches the upper bound of Kale (2014). The two bounds determine the minimax optimal expected regret to be $\Theta\left( \sqrt{T K \log (N/K) } \right)$, where $K$ is the number of arms, $N$ is the number of experts, and $T$ is the time horizon.
Authors: Aanchal Rajesh Chugh, Marion Neumeier, Sebastian Dorn
Abstract: Recent advancements in Recurrent Neural Network (RNN) architectures, particularly the Extended Long Short Term Memory (xLSTM), have addressed the limitations of traditional Long Short Term Memory (LSTM) networks by introducing exponential gating and enhanced memory structures. These improvements make xLSTM suitable for time-series prediction tasks as they exhibit the ability to model long-term temporal dependencies better than LSTMs. Despite their potential, these xLSTM-based models remain largely unexplored in the context of vehicle trajectory prediction. Therefore, this paper introduces a novel xLSTM-based vehicle trajectory prediction framework, X-TRAJ, and its physics-aware variant, X-TRACK (eXtended LSTM for TRAjectory prediction Constraint by Kinematics), which explicitly integrates vehicle motion kinematics into the model learning process. By introducing physical constraints, the proposed model generates realistic and feasible trajectories. A comprehensive evaluation on the highD and NGSIM datasets demonstrates that X-TRACK outperforms state-of-the-art baselines.
Authors: Michiel Straat, Thorben Markmann, Sebastian Peitz, Barbara Hammer
Abstract: Chaotic convective flows arise in many real-world systems, such as microfluidic devices and chemical reactors. Stabilizing these flows is highly desirable but remains challenging, particularly in chaotic regimes where conventional control methods often fail. Reinforcement Learning (RL) has shown promise for control in laminar flow settings, but its ability to generalize and remain robust under chaotic and turbulent dynamics is not well explored, despite being critical for real-world deployment. In this work, we improve the practical feasibility of RL-based control of such flows focusing on Rayleigh-B\'enard Convection (RBC), a canonical model for convective heat transport. To enhance generalization and sample efficiency, we introduce domain-informed RL agents that are trained using Proximal Policy Optimization across diverse initial conditions and flow regimes. We incorporate domain knowledge in the reward function via a term that encourages B\'enard cell merging, as an example of a desirable macroscopic property. In laminar flow regimes, the domain-informed RL agents reduce convective heat transport by up to 33%, and in chaotic flow regimes, they still achieve a 10% reduction, which is significantly better than the conventional controllers used in practice. We compare the domain-informed to uninformed agents: Our results show that the domain-informed reward design results in steady flows, faster convergence during training, and generalization across flow regimes without retraining. Our work demonstrates that elegant domain-informed priors can greatly enhance the robustness of RL-based control of chaotic flows, bringing real-world deployment closer.
Authors: Abhinav Joshi, Areeb Ahmad, Ashutosh Modi
Abstract: Large Language Models (LLMs) have demonstrated inherent calibration capabilities, where predicted probabilities align well with correctness, despite prior findings that deep neural networks are often overconfident. Recent studies have linked this behavior to specific components in the final layer, such as entropy neurons and the unembedding matrix null space. In this work, we provide a complementary perspective by investigating how calibration evolves throughout the network depth. Analyzing multiple open-weight models on the MMLU benchmark, we uncover a distinct confidence correction phase in the upper/later layers, where model confidence is actively recalibrated after decision certainty has been reached. Furthermore, we identify a low-dimensional calibration direction in the residual stream whose perturbation significantly improves calibration metrics (ECE and MCE) without harming accuracy. Our findings suggest that calibration is a distributed phenomenon, shaped throughout the network forward pass, not just in its final projection, providing new insights into how confidence-regulating mechanisms operate within LLMs.
Authors: Ciaran Bench, Oskar Pfeffer, Vivek Desai, Mohammad Moulaeifard, Lo\"ic Coquelin, Peter H. Charlton, Nils Strodthoff, Nando Hegemann, Philip J. Aston, Andrew Thompson
Abstract: In principle, deep learning models trained on medical time-series, including wearable photoplethysmography (PPG) sensor data, can provide a means to continuously monitor physiological parameters outside of clinical settings. However, there is considerable risk of poor performance when deployed in practical measurement scenarios leading to negative patient outcomes. Reliable uncertainties accompanying predictions can provide guidance to clinicians in their interpretation of the trustworthiness of model outputs. It is therefore of interest to compare the effectiveness of different approaches. Here we implement an unprecedented set of eight uncertainty quantification (UQ) techniques to models trained on two clinically relevant prediction tasks: Atrial Fibrillation (AF) detection (classification), and two variants of blood pressure regression. We formulate a comprehensive evaluation procedure to enable a rigorous comparison of these approaches. We observe a complex picture of uncertainty reliability across the different techniques, where the most optimal for a given task depends on the chosen expression of uncertainty, evaluation metric, and scale of reliability assessed. We find that assessing local calibration and adaptivity provides practically relevant insights about model behaviour that otherwise cannot be acquired using more commonly implemented global reliability metrics. We emphasise that criteria for evaluating UQ techniques should cater to the model's practical use case, where the use of a small number of measurements per patient places a premium on achieving small-scale reliability for the chosen expression of uncertainty, while preserving as much predictive performance as possible.
Authors: Dana Kim, Yichen Xu, Tiffany Lin
Abstract: Large Language Models (LLMs) offer a flexible means to generate synthetic tabular data, yet existing approaches often fail to preserve key causal parameters such as the average treatment effect (ATE). In this technical exploration, we first demonstrate that state-of-the-art synthetic data generators, both GAN- and LLM-based, can achieve high predictive fidelity while substantially misestimating causal effects. To address this gap, we propose a hybrid generation framework that combines model-based covariate synthesis (monitored via distance-to-closest-record filtering) with separately learned propensity and outcome models, thereby ensuring that (W, A, Y) triplets retain their underlying causal structure. We further introduce a synthetic pairing strategy to mitigate positivity violations and a realistic evaluation protocol that leverages unlimited synthetic samples to benchmark traditional estimators (IPTW, AIPW, substitution) under complex covariate distributions. This work lays the groundwork for LLM-powered data pipelines that support robust causal analysis. Our code is available at https://github.com/Xyc-arch/llm-synthetic-for-causal-inference.git.
URLs: https://github.com/Xyc-arch/llm-synthetic-for-causal-inference.git.
Authors: Amir Ziashahabi, Yavuz Faruk Bakman, Duygu Nur Yaldiz, Mostafa El-Khamy, Sai Praneeth Karimireddy, Salman Avestimehr
Abstract: Speculative Decoding (SD) ensures that the output matches the target model's distribution exactly. However, we argue that this distribution matching requirement is too stringent and results in unnecessarily low acceptance rates, limiting potential speedups. Instead, we advocate a reformulation of the decoding objective: the proposed decoding strategy should match the expected utility, i.e., the task-specific performance, of the target model. This perspective also aligns better with real-world use cases of LLMs, where utility (e.g., code correctness, factual accuracy) is often more important than sampling distribution. Based on this reformulation, we propose a novel decoding strategy: Pivot-Aware Speculative Decoding, which rejects only those tokens that would lead to a utility drop in the final output. We refer to these critical tokens as pivot tokens. We propose a method for labeling tokens as pivotal or non-pivotal and train a lightweight classifier to detect them. This method can be viewed as a relaxed version of standard SD, which offers much higher acceptance while preserving utility. We evaluate our method across various datasets, demonstrating that we can achieve up to $2.5\times$ speedup with comparable utility. Source code is available at https://github.com/amir-zsh/PAD.
Authors: Zhecheng Sheng, Jiawei Zhang, Enmao Diao
Abstract: Ensuring algorithmic fairness remains a significant challenge in machine learning, particularly as models are increasingly applied across diverse domains. While numerous fairness criteria exist, they often lack generalizability across different machine learning problems. This paper examines the connections and differences among various sparsity measures in promoting fairness and proposes a unified sparsity-based framework for evaluating algorithmic fairness. The framework aligns with existing fairness criteria and demonstrates broad applicability to a wide range of machine learning tasks. We demonstrate the effectiveness of the proposed framework as an evaluation metric through extensive experiments on a variety of datasets and bias mitigation methods. This work provides a novel perspective to algorithmic fairness by framing it through the lens of sparsity and social equity, offering potential for broader impact on fairness research and applications.
Authors: Farjana Aktar, Mohd Ruhul Ameen, Akif Islam, Md Ekramul Hamid
Abstract: Achieving both accurate and interpretable classification of motor imagery EEG remains a key challenge in brain computer interface (BCI) research. This paper compares a transparent fuzzy reasoning approach (ANFIS-FBCSP-PSO) with a deep learning benchmark (EEGNet) using the BCI Competition IV-2a dataset. The ANFIS pipeline combines filter bank common spatial pattern feature extraction with fuzzy IF-THEN rules optimized via particle swarm optimization, while EEGNet learns hierarchical spatial temporal representations directly from raw EEG data. In within-subject experiments, the fuzzy neural model performed better (68.58 percent +/- 13.76 percent accuracy, kappa = 58.04 percent +/- 18.43), while in cross-subject (LOSO) tests, the deep model exhibited stronger generalization (68.20 percent +/- 12.13 percent accuracy, kappa = 57.33 percent +/- 16.22). The study provides practical guidance for selecting MI-BCI systems according to design goals: interpretability or robustness across users. Future investigations into transformer based and hybrid neuro symbolic frameworks are expected to advance transparent EEG decoding.
Authors: Xin Wang, Yunhao Xiao, Rui Qiao
Abstract: We introduce PolyRecommender, a multimodal discovery framework that integrates chemical language representations from PolyBERT with molecular graph-based representations from a graph encoder. The system first retrieves candidate polymers using language-based similarity and then ranks them using fused multimodal embeddings according to multiple target properties. By leveraging the complementary knowledge encoded in both modalities, PolyRecommender enables efficient retrieval and robust ranking across related polymer properties. Our work establishes a generalizable multimodal paradigm, advancing AI-guided design for the discovery of next-generation polymers.
Authors: Zhibin Lan, Liqiang Niu, Fandong Meng, Jie Zhou, Jinsong Su
Abstract: The remarkable success of multimodal large language models (MLLMs) has driven advances in multimodal embeddings, yet existing models remain inherently discriminative, limiting their ability to benefit from reasoning-driven generation paradigm. In this work, we pioneer the exploration of generative embeddings, unifying embedding tasks within a generative paradigm. We propose UME-R1, a universal multimodal embedding framework consisting of a two-stage training strategy: a cold-start supervised fine-tuning equips the model with reasoning capabilities and enables it to generate both discriminative and generative embeddings; a subsequent reinforcement learning enhances reasoning and further optimizes generative embedding quality. This pioneering work reveals four key insights: 1) generative embeddings unlock substantial performance gains over conventional discriminative embeddings by leveraging the powerful generative reasoning capabilities of MLLMs; 2) discriminative and generative embeddings are complementary, whose combined oracle performance far exceeding that of either alone; 3) RL can effectively enhance generative embeddings, establishing a scalable optimization paradigm.; 4) repeated sampling at inference boosts downstream task coverage (pass@k), highlighting the inference-time scalability potential of generative embeddings. Evaluated on the MMEB-V2 benchmark across 78 tasks spanning video, image, and visual documents, UME-R1 significantly outperforms conventional discriminative embedding models and offers a foundation for more interpretable, reasoning-driven generative multimodal embeddings. Our code, models, and datasets will be publicly available at https://github.com/XMUDeepLIT/UME-R1.
Authors: Zenghao Niu, Weicheng Xie, Siyang Song, Zitong Yu, Feng Liu, Linlin Shen
Abstract: Adversarial attacks present a critical challenge to deep neural networks' robustness, particularly in transfer scenarios across different model architectures. However, the transferability of adversarial attacks faces a fundamental dilemma between Exploitation (maximizing attack potency) and Exploration (enhancing cross-model generalization). Traditional momentum-based methods over-prioritize Exploitation, i.e., higher loss maxima for attack potency but weakened generalization (narrow loss surface). Conversely, recent methods with inner-iteration sampling over-prioritize Exploration, i.e., flatter loss surfaces for cross-model generalization but weakened attack potency (suboptimal local maxima). To resolve this dilemma, we propose a simple yet effective Gradient-Guided Sampling (GGS), which harmonizes both objectives through guiding sampling along the gradient ascent direction to improve both sampling efficiency and stability. Specifically, based on MI-FGSM, GGS introduces inner-iteration random sampling and guides the sampling direction using the gradient from the previous inner-iteration (the sampling's magnitude is determined by a random distribution). This mechanism encourages adversarial examples to reside in balanced regions with both flatness for cross-model generalization and higher local maxima for strong attack potency. Comprehensive experiments across multiple DNN architectures and multimodal large language models (MLLMs) demonstrate the superiority of our method over state-of-the-art transfer attacks. Code is made available at https://github.com/anuin-cat/GGS.
Authors: Shaojie Wang, Jinghui Wang, Yinghan Cui, Xuxing Chen, Chao Wang, Liang Huang, Xiaojiang Zhang, Junyi Peng, Li Wan, Haotian Zhang, Bin Chen
Abstract: In agentic LLM scenarios, an agent's interaction process during a single rollout often exhibits branching behaviors. Due to memory retrieval and concurrent tool executions at certain decision points, the token trajectory of one task evolves into a tree-like structure rather than a linear sequence. However, current training pipelines decompose such tree-structured trajectories into separate linear segments, treating each branch as an independent sequence. As a result, shared prefixes across these branches are repeatedly recomputed during both forward and backward passes. To address this inefficiency, we propose Tree Training, a paradigm that computes each shared prefix only once and reuses its intermediate results across related branches during both forward and backward passes, substantially improving computation efficiency in large-scale agentic training. This is achieved via (i) Tree Packing, which efficiently reuses shared computations across trajectories, and (ii) Gradient Restoration, which ensures correct gradient propagation across reused prefixes. Experiments on multiple open-source models demonstrate up to 3.9x reduction in total training time, enabling more efficient agentic LLM SFT and RL training.
Authors: Victory Obieke, Emmanuel Oguadimma
Abstract: Physics-Informed Neural Networks (PINNs) offer a flexible framework for solving nonlinear partial differential equations (PDEs), yet conventional implementations often fail to preserve key physical invariants during long-term integration. This paper introduces a \emph{structure-preserving PINN} framework for the nonlinear Korteweg--de Vries (KdV) equation, a prototypical model for nonlinear and dispersive wave propagation. The proposed method embeds the conservation of mass and Hamiltonian energy directly into the loss function, ensuring physically consistent and energy-stable evolution throughout training and prediction. Unlike standard \texttt{tanh}-based PINNs~\cite{raissi2019pinn,wang2022modifiedpinn}, our approach employs sinusoidal activation functions that enhance spectral expressiveness and accurately capture the oscillatory and dispersive nature of KdV solitons. Through representative case studies -- including single-soliton propagation (shape-preserving translation), two-soliton interaction (elastic collision with phase shift), and cosine-pulse initialization (nonlinear dispersive breakup) -- the model successfully reproduces hallmark behaviors of KdV dynamics while maintaining conserved invariants. Ablation studies demonstrate that combining invariant-constrained optimization with sinusoidal feature mappings accelerates convergence, improves long-term stability, and mitigates drift without multi-stage pretraining. These results highlight that computationally efficient, invariant-aware regularization coupled with sinusoidal representations yields robust, energy-consistent PINNs for Hamiltonian partial differential equations such as the KdV equation.
Authors: Guojian Zhan, Likun Wang, Xiangteng Zhang, Jiaxin Gao, Masayoshi Tomizuka, Shengbo Eben Li
Abstract: Online planning has proven effective in reinforcement learning (RL) for improving sample efficiency and final performance. However, using planning for environment interaction inevitably introduces a divergence between the collected data and the policy's actual behaviors, degrading both model learning and policy improvement. To address this, we propose BOOM (Bootstrap Off-policy with WOrld Model), a framework that tightly integrates planning and off-policy learning through a bootstrap loop: the policy initializes the planner, and the planner refines actions to bootstrap the policy through behavior alignment. This loop is supported by a jointly learned world model, which enables the planner to simulate future trajectories and provides value targets to facilitate policy improvement. The core of BOOM is a likelihood-free alignment loss that bootstraps the policy using the planner's non-parametric action distribution, combined with a soft value-weighted mechanism that prioritizes high-return behaviors and mitigates variability in the planner's action quality within the replay buffer. Experiments on the high-dimensional DeepMind Control Suite and Humanoid-Bench show that BOOM achieves state-of-the-art results in both training stability and final performance. The code is accessible at https://github.com/molumitu/BOOM_MBRL.
Authors: Ruthwik Reddy Doodipala, Pankaj Pandey, Carolina Torres Rojas, Manob Jyoti Saikia, Ranganatha Sitaram
Abstract: The emergence of foundation models in neuroimaging is driven by the increasing availability of large-scale and heterogeneous brain imaging datasets. Recent advances in self-supervised learning, particularly reconstruction-based objectives, have demonstrated strong potential for pretraining models that generalize effectively across diverse downstream functional MRI (fMRI) tasks. In this study, we explore region-aware reconstruction strategies for a foundation model in resting-state fMRI, moving beyond approaches that rely on random region masking. Specifically, we introduce an ROI-guided masking strategy using the Automated Anatomical Labelling Atlas (AAL3), applied directly to full 4D fMRI volumes to selectively mask semantically coherent brain regions during self-supervised pretraining. Using the ADHD-200 dataset comprising 973 subjects with resting-state fMRI scans, we show that our method achieves a 4.23% improvement in classification accuracy for distinguishing healthy controls from individuals diagnosed with ADHD, compared to conventional random masking. Region-level attribution analysis reveals that brain volumes within the limbic region and cerebellum contribute most significantly to reconstruction fidelity and model representation. Our results demonstrate that masking anatomical regions during model pretraining not only enhances interpretability but also yields more robust and discriminative representations. In future work, we plan to extend this approach by evaluating it on additional neuroimaging datasets, and developing new loss functions explicitly derived from region-aware reconstruction objectives. These directions aim to further improve the robustness and interpretability of foundation models for functional neuroimaging.
Authors: Xin Chen, Saili Uday Gadgil, Kangning Gao, Yi Hu, Cong Nie
Abstract: An anomaly detection method based on deep autoencoders is proposed to address anomalies that often occur in enterprise-level ETL data streams. The study first analyzes multiple types of anomalies in ETL processes, including delays, missing values, duplicate loading, and sudden abnormal changes, and applies data standardization and feature modeling to ensure stable and usable inputs. In the method design, the encoder-decoder structure compresses high-dimensional inputs into latent representations and reconstructs them, while reconstruction error is used to measure anomaly levels. Regularization constraints are introduced in the latent space to enhance feature sparsity and distribution learning, thereby improving robustness in complex data streams. Systematic analyses under different hyperparameter settings, environmental changes, and data characteristics show that the proposed method achieves superior performance in AUC, ACC, Precision, and Recall. The results demonstrate that the deep autoencoder-based detection mechanism can effectively capture latent distribution patterns in enterprise-level ETL data streams and accurately identify diverse anomalies, providing reliable support for enterprise data processing and intelligent analysis.
Authors: Zhongxiang Lei, Qi Yang, Ping Qiu, Gang Zhang, Yuanchi Ma, Jinyan Liu
Abstract: Federated optimization is a constrained form of distributed optimization that enables training a global model without directly sharing client data. Although existing algorithms can guarantee convergence in theory and often achieve stable training in practice, the reasons behind performance degradation under data heterogeneity remain unclear. To address this gap, the main contribution of this paper is to provide a theoretical perspective that explains why such degradation occurs. We introduce the assumption that heterogeneous client data lead to distinct local optima, and show that this assumption implies two key consequences: 1) the distance among clients' local optima raises the lower bound of the global objective, making perfect fitting of all client data impossible; and 2) in the final training stage, the global model oscillates within a region instead of converging to a single optimum, limiting its ability to fully fit the data. These results provide a principled explanation for performance degradation in non-iid settings, which we further validate through experiments across multiple tasks and neural network architectures. The framework used in this paper is open-sourced at: https://github.com/NPCLEI/fedtorch.
Authors: Travis Barrett, Amit Kumar Mishra, Joyce Mwangama
Abstract: In this paper we present a new implementation of a Variational Autoencoder (VAE) for the calibration of sensors. We propose that the VAE can be used to calibrate sensor data by training the latent space as a calibration output. We discuss this new approach and show a proof-of-concept using an existing multi-sensor gas dataset. We show the performance of the proposed calibration VAE and found that it was capable of performing as calibration model while performing as an autoencoder simultaneously. Additionally, these models have shown that they are capable of creating statistically similar outputs from both the calibration output as well as the reconstruction output to their respective truth data. We then discuss the methods of future testing and planned expansion of this work.
Authors: Bao Nguyen, Hieu Trung Nguyen, Ruifeng She, Xiaojin Fu, Viet Anh Nguyen
Abstract: Selecting an appropriate reasoning method for a given query remains a key challenge in language model generation. Existing approaches typically generate multiple candidate responses and use an aggregation strategy to select the output answer, often assuming that more candidate answers yield higher accuracy. We revisit this assumption through a rigorous theoretical analysis, deriving accuracy bounds for standard aggregation methods under fixed generation distributions and candidate sizes. Building on these insights, we introduce EPIC, an Ensemble Planning with Contrastive learning framework to learn a shared representation space that captures both model reasoning abilities and query-method compatibility. EPIC incorporates our probability bounds as a regularizer in a utility-driven optimization that balances accuracy and computational cost. Experiments on diverse mathematical reasoning tasks show that EPIC consistently selects optimal reasoning methods, improving accuracy while reducing computational overhead. Our code can be found at https://github.com/nguyenngocbaocmt02/EPIC.
Authors: Drago\c{s}-Andrei \c{S}erban, R\u{a}zvan-Alexandru Sm\u{a}du, Dumitru-Clementin Cercel
Abstract: Air pollution, especially the particulate matter 2.5 (PM2.5), has become a growing concern in recent years, primarily in urban areas. Being exposed to air pollution is linked to developing numerous health problems, like the aggravation of respiratory diseases, cardiovascular disorders, lung function impairment, and even cancer or early death. Forecasting future levels of PM2.5 has become increasingly important over the past few years, as it can provide early warnings and help prevent diseases. This paper aims to design, fine-tune, test, and evaluate machine learning models for predicting future levels of PM2.5 over various time horizons. Our primary objective is to assess and compare the performance of multiple models, ranging from linear regression algorithms and ensemble-based methods to deep learning models, such as advanced recurrent neural networks and transformers, as well as large language models, on this forecasting task.
Authors: Yunchuan Guan, Yu Liu, Ke Zhou, Hui Li, Sen Jia, Zhiqi Shen, Ziyang Wang, Xinglin Zhang, Tao Chen, Jenq-Neng Hwang, Lei Li
Abstract: Recent advances in generative modeling enable neural networks to generate weights without relying on gradient-based optimization. However, current methods are limited by issues of over-coupling and long-horizon. The former tightly binds weight generation with task-specific objectives, thereby limiting the flexibility of the learned optimizer. The latter leads to inefficiency and low accuracy during inference, caused by the lack of local constraints. In this paper, we propose Lo-Hp, a decoupled two-stage weight generation framework that enhances flexibility through learning various optimization policies. It adopts a hybrid-policy sub-trajectory balance objective, which integrates on-policy and off-policy learning to capture local optimization policies. Theoretically, we demonstrate that learning solely local optimization policies can address the long-horizon issue while enhancing the generation of global optimal weights. In addition, we validate Lo-Hp's superior accuracy and inference efficiency in tasks that require frequent weight updates, such as transfer learning, few-shot learning, domain generalization, and large language model adaptation.
Authors: Qiang Li, Jin Niu, Lina Yu
Abstract: Traffic congestion, primarily driven by intersection queuing, significantly impacts urban living standards, safety, environmental quality, and economic efficiency. While Traffic Signal Control (TSC) systems hold potential for congestion mitigation, traditional optimization models often fail to capture real-world traffic complexity and dynamics. This study introduces a novel single-agent reinforcement learning (RL) framework for regional adaptive TSC, circumventing the coordination complexities inherent in multi-agent systems through a centralized decision-making paradigm. The model employs an adjacency matrix to unify the encoding of road network topology, real-time queue states derived from probe vehicle data, and current signal timing parameters. Leveraging the efficient learning capabilities of the DreamerV3 world model, the agent learns control policies where actions sequentially select intersections and adjust their signal phase splits to regulate traffic inflow/outflow, analogous to a feedback control system. Reward design prioritizes queue dissipation, directly linking congestion metrics (queue length) to control actions. Simulation experiments conducted in SUMO demonstrate the model's effectiveness: under inference scenarios with multi-level (10%, 20%, 30%) Origin-Destination (OD) demand fluctuations, the framework exhibits robust anti-fluctuation capability and significantly reduces queue lengths. This work establishes a new paradigm for intelligent traffic control compatible with probe vehicle technology. Future research will focus on enhancing practical applicability by incorporating stochastic OD demand fluctuations during training and exploring regional optimization mechanisms for contingency events.
Authors: Santhi Bharath Punati, Sandeep Kanta, Udaya Bhasker Cheerala, Madhusudan G Lanjewar, Praveen Damacharla
Abstract: Accurate multi-horizon retail forecasts are critical for inventory and promotions. We present a novel study of weekly Walmart sales (45 stores, 2010--2012) using a Temporal Fusion Transformer (TFT) that fuses static store identifiers with time-varying exogenous signals (holidays, CPI, fuel price, temperature). The pipeline produces 1--5-week-ahead probabilistic forecasts via Quantile Loss, yielding calibrated 90\% prediction intervals and interpretability through variable-selection networks, static enrichment, and temporal attention. On a fixed 2012 hold-out dataset, TFT achieves an RMSE of \$57.9k USD per store-week and an $R^2$ of 0.9875. Across a 5-fold chronological cross-validation, the averages are RMSE = \$64.6k USD and $R^2$ = 0.9844, outperforming the XGB, CNN, LSTM, and CNN-LSTM baseline models. These results demonstrate practical value for inventory planning and holiday-period optimization, while maintaining model transparency.
Authors: Phil Blandfort, Robert Graham
Abstract: Activation probes are attractive monitors for AI systems due to low cost and latency, but their real-world robustness remains underexplored. We ask: What failure modes arise under realistic, black-box adversarial pressure, and how can we surface them with minimal effort? We present a lightweight black-box red-teaming procedure that wraps an off-the-shelf LLM with iterative feedback and in-context learning (ICL), and requires no fine-tuning, gradients, or architectural access. Running a case study with probes for high-stakes interactions, we show that our approach can help discover valuable insights about a SOTA probe. Our analysis uncovers interpretable brittleness patterns (e.g., legalese-induced FPs; bland procedural tone FNs) and reduced but persistent vulnerabilities under scenario-constraint attacks. These results suggest that simple prompted red-teaming scaffolding can anticipate failure patterns before deployment and might yield promising, actionable insights to harden future probes.
Authors: Varun Teja Chirukiri, Udaya Bhasker Cheerala, Sandeep Kanta, Abdul Karim, Praveen Damacharla
Abstract: Accurate prediction of the remaining useful life (RUL) of industrial machinery is essential for reducing downtime and optimizing maintenance schedules. Existing approaches, such as long short-term memory (LSTM) networks and convolutional neural networks (CNNs), often struggle to model both global temporal dependencies and fine-grained degradation trends in multivariate sensor data. We propose a hybrid model, FTT-GRU, which combines a Fast Temporal Transformer (FTT) -- a lightweight Transformer variant using linearized attention via fast Fourier transform (FFT) -- with a gated recurrent unit (GRU) layer for sequential modeling. To the best of our knowledge, this is the first application of an FTT with a GRU for RUL prediction on NASA CMAPSS, enabling simultaneous capture of global and local degradation patterns in a compact architecture. On CMAPSS FD001, FTT-GRU attains RMSE 30.76, MAE 18.97, and $R^2=0.45$, with 1.12 ms CPU latency at batch=1. Relative to the best published deep baseline (TCN--Attention), it improves RMSE by 1.16\% and MAE by 4.00\%. Training curves averaged over $k=3$ runs show smooth convergence with narrow 95\% confidence bands, and ablations (GRU-only, FTT-only) support the contribution of both components. These results demonstrate that a compact Transformer-RNN hybrid delivers accurate and efficient RUL predictions on CMAPSS, making it suitable for real-time industrial prognostics.
Authors: Yinghuan Zhang, Yufei Zhang, Parisa Kordjamshidi, Zijun Cui
Abstract: Understanding probabilistic relationships among variables is crucial for analyzing complex systems. Traditional structure learning methods often require extensive observational data and incur high computational costs. Recent studies have explored using large language models (LLMs) for structure learning, but most treat LLMs as auxiliary tools for pre-processing or post-processing, leaving the core learning process data-driven. In this work, we propose a unified framework for Bayesian network structure discovery that places LLMs at the center, supporting both data-free and data-aware settings. In the data-free case, we introduce \textbf{PromptBN} to query LLMs with metadata and efficiently uncover valid probabilistic relationships. When observational data are available, we introduce \textbf{ReActBN}, which integrates the ReAct reasoning paradigm with structure scores such as the Bayesian Information Criterion (BIC) for iterative refinement. Unlike prior methods that offload refinement to external algorithms, our framework maintains the LLM actively in the loop throughout the discovery process. Experiments demonstrate that our method significantly outperforms both existing LLM-based approaches and traditional data-driven algorithms, particularly in the low- or no-data scenario. Code is publicly available at {\texttt{\textcolor{magenta}{https://github.com/sherryzyh/prompt2bn}}}.
Authors: G. Pillonetto, A. Giaretta, A. Aravkin, M. Bisiacco, T. Elston
Abstract: Data-driven discovery of model equations is a powerful approach for understanding the behavior of dynamical systems in many scientific fields. In particular, the ability to learn mathematical models from data would benefit systems biology, where the complex nature of these systems often makes a bottom up approach to modeling unfeasible. In recent years, sparse estimation techniques have gained prominence in system identification, primarily using parametric paradigms to efficiently capture system dynamics with minimal model complexity. In particular, the Sindy algorithm has successfully used sparsity to estimate nonlinear systems by extracting from a library of functions only a few key terms needed to capture the dynamics of these systems. However, parametric models often fall short in accurately representing certain nonlinearities inherent in complex systems. To address this limitation, we introduce a novel framework that integrates sparse parametric estimation with nonparametric techniques. It captures nonlinearities that Sindy cannot describe without requiring a priori information about their functional form. That is, without expanding the library of functions to include the one that is trying to be discovered. We illustrate our approach on several examples related to estimation of complex biological phenomena.
Authors: Dong Chen, Yanzhe Wei, Zonglin He, Guan-Ming Kuang, Canhua Ye, Meiru An, Huili Peng, Yong Hu, Huiren Tao, Kenneth MC Cheung
Abstract: Large language models (LLMs) offer transformative potential for clinical decision support in spine surgery but pose significant risks through hallucinations, which are factually inconsistent or contextually misaligned outputs that may compromise patient safety. This study introduces a clinician-centered framework to quantify hallucination risks by evaluating diagnostic precision, recommendation quality, reasoning robustness, output coherence, and knowledge alignment. We assessed six leading LLMs across 30 expert-validated spinal cases. DeepSeek-R1 demonstrated superior overall performance (total score: 86.03 $\pm$ 2.08), particularly in high-stakes domains such as trauma and infection. A critical finding reveals that reasoning-enhanced model variants did not uniformly outperform standard counterparts: Claude-3.7-Sonnet's extended thinking mode underperformed relative to its standard version (80.79 $\pm$ 1.83 vs. 81.56 $\pm$ 1.92), indicating extended chain-of-thought reasoning alone is insufficient for clinical reliability. Multidimensional stress-testing exposed model-specific vulnerabilities, with recommendation quality degrading by 7.4% under amplified complexity. This decline contrasted with marginal improvements in rationality (+2.0%), readability (+1.7%) and diagnosis (+4.7%), highlighting a concerning divergence between perceived coherence and actionable guidance. Our findings advocate integrating interpretability mechanisms (e.g., reasoning chain visualization) into clinical workflows and establish a safety-aware validation framework for surgical LLM deployment.
Authors: Daniel Griffiths, Piper Moskow
Abstract: We present a unified, data-driven framework for quantifying and enhancing offensive momentum and scoring likelihood (expected goals, xG) in professional hockey. Leveraging a Sportlogiq dataset of 541,000 NHL event records, our end-to-end pipeline comprises five stages: (1) interpretable momentum weighting of micro-events via logistic regression; (2) nonlinear xG estimation using gradient-boosted decision trees; (3) temporal sequence modeling with Long Short-Term Memory (LSTM) networks; (4) spatial formation discovery through principal component analysis (PCA) followed by K-Means clustering on standardized player coordinates; and (5) use of an X-Learner causal inference estimator to quantify the average treatment effect (ATE) of adopting the identified "optimal" event sequences and formations. We observe an ATE of 0.12 (95% CI: 0.05-0.17, p < 1e-50), corresponding to a 15% relative gain in scoring potential. These results demonstrate that strategically structured sequences and compact formations causally elevate offensive performance. Our framework delivers real-time, actionable insights for coaches and analysts, advancing hockey analytics toward principled, causally grounded tactical optimization.
Authors: Eric Bigelow, Daniel Wurgaft, YingQiao Wang, Noah Goodman, Tomer Ullman, Hidenori Tanaka, Ekdeep Singh Lubana
Abstract: Large language models (LLMs) can be controlled at inference time through prompts (in-context learning) and internal activations (activation steering). Different accounts have been proposed to explain these methods, yet their common goal of controlling model behavior raises the question of whether these seemingly disparate methodologies can be seen as specific instances of a broader framework. Motivated by this, we develop a unifying, predictive account of LLM control from a Bayesian perspective. Specifically, we posit that both context- and activation-based interventions impact model behavior by altering its belief in latent concepts: steering operates by changing concept priors, while in-context learning leads to an accumulation of evidence. This results in a closed-form Bayesian model that is highly predictive of LLM behavior across context- and activation-based interventions in a set of domains inspired by prior work on many-shot in-context learning. This model helps us explain prior empirical phenomena - e.g., sigmoidal learning curves as in-context evidence accumulates - while predicting novel ones - e.g., additivity of both interventions in log-belief space, which results in distinct phases such that sudden and dramatic behavioral shifts can be induced by slightly changing intervention controls. Taken together, this work offers a unified account of prompt-based and activation-based control of LLM behavior, and a methodology for empirically predicting the effects of these interventions.
Authors: Emmeran Johnson, Alberto Rumi, Ciara Pike-Burke, Patrick Rebeschini
Abstract: We study the adversarial Stochastic Shortest Path (SSP) problem with sparse costs under full-information feedback. In the known transition setting, existing bounds based on Online Mirror Descent (OMD) with negative-entropy regularization scale with $\sqrt{\log S A}$, where $SA$ is the size of the state-action space. While we show that this is optimal in the worst-case, this bound fails to capture the benefits of sparsity when only a small number $M \ll SA$ of state-action pairs incur cost. In fact, we also show that the negative-entropy is inherently non-adaptive to sparsity: it provably incurs regret scaling with $\sqrt{\log S}$ on sparse problems. Instead, we propose a family of $\ell_r$-norm regularizers ($r \in (1,2)$) that adapts to the sparsity and achieves regret scaling with $\sqrt{\log M}$ instead of $\sqrt{\log SA}$. We show this is optimal via a matching lower bound, highlighting that $M$ captures the effective dimension of the problem instead of $SA$. Finally, in the unknown transition setting the benefits of sparsity are limited: we prove that even on sparse problems, the minimax regret for any learner scales polynomially with $SA$.
Authors: C. D\'iaz-Faloh, R. Mulet
Abstract: Recent advances in artificial intelligence have relied heavily on increasingly large neural networks, raising concerns about their computational and environmental costs. This paper investigates whether simpler, sparser networks can maintain strong performance by studying Restricted Boltzmann Machines (RBMs) under extreme pruning conditions. Inspired by the Lottery Ticket Hypothesis, we demonstrate that RBMs can achieve high-quality generative performance even when up to 80% of the connections are pruned before training, confirming that they contain viable sub-networks. However, our experiments reveal crucial limitations: trained networks cannot fully recover lost performance through retraining once additional pruning is applied. We identify a sharp transition above which the generative quality degrades abruptly when pruning disrupts a minimal core of essential connections. Moreover, re-trained networks remain constrained by the parameters originally learned performing worse than networks trained from scratch at equivalent sparsity levels. These results suggest that for sparse networks to work effectively, pruning should be implemented early in training rather than attempted afterwards. Our findings provide practical insights for the development of efficient neural architectures and highlight the persistent influence of initial conditions on network capabilities.
Authors: Baris Askin, Holger R. Roth, Zhenyu Sun, Carlee Joe-Wong, Gauri Joshi, Ziyue Xu
Abstract: Federated Learning (FL) enables collaborative model training across distributed clients without sharing raw data, yet its scalability is limited by synchronization overhead. Asynchronous Federated Learning (AFL) alleviates this issue by allowing clients to communicate independently, thereby improving wall-clock efficiency in large-scale, heterogeneous environments. However, this asynchrony introduces stale updates (client updates computed on outdated global models) that can destabilize optimization and hinder convergence. We propose FedRevive, an asynchronous FL framework that revives stale updates through data-free knowledge distillation (DFKD). FedRevive integrates parameter-space aggregation with a lightweight, server-side DFKD process that transfers knowledge from stale client models to the current global model without access to real or public data. A meta-learned generator synthesizes pseudo-samples, which enables multi-teacher distillation. A hybrid aggregation scheme that combines raw updates with DFKD updates effectively mitigates staleness while retaining the scalability of AFL. Experiments on various vision and text benchmarks show that FedRevive achieves faster training up to 32.1% and higher final accuracy up to 21.5% compared to asynchronous baselines.
Authors: Alex Dobra, Jakiw Pidstrigach, Tim Reichelt, Paolo Fraccaro, Johannes Jakubik, Anne Jones, Christian Schroeder de Witt, Philip Stier, Philip Torr
Abstract: Sensitivity analysis is a cornerstone of climate science, essential for understanding phenomena ranging from storm intensity to long-term climate feedbacks. However, computing these sensitivities using traditional physical models is often prohibitively expensive in terms of both computation and development time. While modern AI-based generative models are orders of magnitude faster to evaluate, computing sensitivities with them remains a significant bottleneck. This work addresses this challenge by applying the adjoint state method for calculating gradients in generative flow models, with diffusion models as a special case. We apply this method to the cBottle generative model, an emulator of ERA5 data, to perform sensitivity analysis with respect to sea surface temperatures. Furthermore, we propose a novel gradient self-consistency check to quantitatively validate the computed sensitivities against the model's own outputs. Our results provide initial evidence that this approach can produce reliable gradients, reducing the computational cost of sensitivity analysis from weeks on a supercomputer with a physical model to hours on a GPU, thereby simplifying a critical workflow in climate science.
Authors: Sophie Li (Columbia University), Nicholas Huang (University of British Columbia), Nayan Saxena (Algoverse AI Research), Nina Luo (Harvey Mudd College), Vincent Lin (University of Florida), Kevin Zhu (Algoverse AI Research), Sunishchal Dev (Algoverse AI Research)
Abstract: Large language models (LLMs) improve reasoning accuracy when generating multiple candidate solutions at test time, but standard methods like Best-of-N (BoN) incur high computational cost by fully generating all branches. Self-Truncation Best-of-N (ST-BoN) mitigates this by truncating unpromising paths early, but its reliance on consistency-based heuristics is a limitation as it does not directly evaluate branch quality. We present KL-Adjusted Pruned Path Algorithm (KAPPA), an inference-time method that combines Kullback-Leibler divergence, confidence, and entropy into a principled scoring function to guide progressive pruning. By promoting diversity during exploration and selectively eliminating low-scoring branches, KAPPA maintains accuracy while substantially reducing memory and token usage. Experiments on GSM8K and MATH500 with DeepSeek-R1-Distill-Qwen-1.5B and Qwen2.5-7B-Instruct demonstrate that KAPPA stabilizes performance in smaller models and achieves up to ~60% reduction in peak memory and ~90% reduction in total token generation relative to BoN, with minimal impact on accuracy.
Authors: Penghang Liu, Haibei Zhu, Eleonora Kreacic, Svitlana Vyetrenko
Abstract: Sharing sensitive time series data in domains such as finance, healthcare, and energy consumption, such as patient records or investment accounts, is often restricted due to privacy concerns. Privacy-aware synthetic time series generation addresses this challenge by enforcing noise during training, inherently introducing a trade-off between privacy and utility. In many cases, sensitive sequences is correlated with publicly available, non-sensitive contextual metadata (e.g., household electricity consumption may be influenced by weather conditions and electricity prices). However, existing privacy-aware data generation methods often overlook this opportunity, resulting in suboptimal privacy-utility trade-offs. In this paper, we present Pub2Priv, a novel framework for generating private time series data by leveraging heterogeneous public knowledge. Our model employs a self-attention mechanism to encode public data into temporal and feature embeddings, which serve as conditional inputs for a diffusion model to generate synthetic private sequences. Additionally, we introduce a practical metric to assess privacy by evaluating the identifiability of the synthetic data. Experimental results show that Pub2Priv consistently outperforms state-of-the-art benchmarks in improving the privacy-utility trade-off across finance, energy, and commodity trading domains.
Authors: Morgan Lee, Artem Frenk, Eamon Worden, Karish Gupta, Thinh Pham, Ethan Croteau, Neil Heffernan
Abstract: Knowledge Tracing (KT) has been an established problem in the educational data mining field for decades, and it is commonly assumed that the underlying learning process be- ing modeled remains static. Given the ever-changing land- scape of online learning platforms (OLPs), we investigate how concept drift and changing student populations can im- pact student behavior within an OLP through testing model performance both within a single academic year and across multiple academic years. Four well-studied KT models were applied to five academic years of data to assess how suscep- tible KT models are to concept drift. Through our analysis, we find that all four families of KT models can exhibit de- graded performance, Bayesian Knowledge Tracing (BKT) remains the most stable KT model when applied to newer data, while more complex, attention based models lose pre- dictive power significantly faster. To foster more longitu- dinal evaluations of KT models, the data used to conduct our analysis is available at https://osf.io/hvfn9/?view_ only=b936c63dfdae4b0b987a2f0d4038f72a
Authors: Nardeep Kumar, Arun Kanwar
Abstract: TRISKELION-1 is a unified descriptive-predictive-generative architecture that integrates statistical, mechanistic, and generative reasoning within a single encoder-decoder framework. The model demonstrates how descriptive representation learning, predictive inference, and generative synthesis can be jointly optimized using variational objectives. Experiments on MNIST validate that descriptive reconstruction, predictive classification, and generative sampling can coexist stably within one model. The framework provides a blueprint toward universal intelligence architectures that connect interpretability, accuracy, and creativity.
Authors: Rama Kassoumeh, David R\"ugamer, Henning Oppel
Abstract: The increasing frequency of heavy rainfall events, which are a major cause of urban flooding, underscores the urgent need for accurate precipitation forecasting - particularly in urban areas where localized events often go undetected by ground-based sensors. In Germany, only 17.3% of hourly heavy rain events between 2001 and 2018 were recorded by rain gauges, highlighting the limitations of traditional monitoring systems. Radar data are another source that effectively tracks ongoing precipitation; however, forecasting the development of heavy rain using radar alone remains challenging due to the brief and unpredictable nature of such events. Our focus is on evaluating the effectiveness of fusing satellite and radar data for nowcasting. We develop a multimodal nowcasting model that combines both radar and satellite imagery for predicting precipitation at lead times of 5, 15, and 30 minutes. We demonstrate that this multimodal strategy significantly outperforms radar-only approaches. Experimental results show that integrating satellite data improves prediction accuracy, particularly for intense precipitation. The proposed model increases the Critical Success Index for heavy rain by 4% and for violent rain by 3% at a 5-minute lead time. Moreover, it maintains higher predictive skill at longer lead times, where radar-only performance declines. A qualitative analysis of the severe flooding event in the state of North Rhine-Westphalia, Germany in 2021 further illustrates the superior performance of the multimodal model. Unlike the radar-only model, which captures general precipitation patterns, the multimodal model yields more detailed and accurate forecasts for regions affected by heavy rain. This improved precision enables timely, reliable, life-saving warnings. Implementation available at https://github.com/RamaKassoumeh/Multimodal_heavy_rain
URLs: https://github.com/RamaKassoumeh/Multimodal_heavy_rain
Authors: Zixuan Ma, Chenfeng Huang
Abstract: Time series generation focuses on modeling the underlying data distribution and resampling to produce authentic time series data. Key components, such as trend and seasonality, drive temporal fluctuations, yet many existing approaches fail to employ interpretative decomposition methods, limiting their ability to synthesize meaningful trend and seasonal patterns. To address this gap, we introduce Seasonal-Trend Diffusion (STDiffusion), a novel framework for multivariate time series generation that integrates diffusion probabilistic models with advanced learnable series decomposition techniques, enhancing the interpretability of the generation process. Our approach separates the trend and seasonal learning into distinct blocks: a Multi-Layer Perceptron (MLP) structure captures the trend, while adaptive wavelet distillation facilitates effective multi-resolution learning of seasonal components. This decomposition improves the interpretability of the model on multiple scales. In addition, we designed a comprehensive correction mechanism aimed at ensuring that the generated components exhibit a high degree of internal consistency and preserve meaningful interrelationships with one another. Our empirical studies on eight real-world datasets demonstrate that STDiffusion achieves state-of-the-art performance in time series generation tasks. Furthermore, we extend the model's application to multi-window long-sequence time series generation, which delivered reliable results and highlighted its robustness and versatility.
Authors: Akshay Sai Banderwaar, Abhishek Gupta
Abstract: Eigenvalue problems have a distinctive forward-inverse structure and are fundamental to characterizing a system's thermal response, stability, and natural modes. Physics-Informed Neural Networks (PINNs) offer a mesh-free alternative for solving such problems but are often orders of magnitude slower than classical numerical schemes. In this paper, we introduce a reformulated PINN approach that casts the search for eigenpairs as a biconvex optimization problem, enabling fast and provably convergent alternating convex search (ACS) over eigenvalues and eigenfunctions using analytically optimal updates. Numerical experiments show that PINN-ACS attains high accuracy with convergence speeds up to 500$\times$ faster than gradient-based PINN training. We release our codes at https://github.com/NeurIPS-ML4PS-2025/PINN_ACS_CODES.
Authors: Yan Sun, Jia Guo, Stanley Kok, Zihao Wang, Zujie Wen, Zhiqiang Zhang
Abstract: Reinforcement learning with verifiable rewards (RLVR) has improved the reasoning ability of large language models, yet training remains costly because many rollouts contribute little to optimization, considering the amount of computation required. This study investigates how simply leveraging intrinsic data properties, almost free benefit during training, can improve data efficiency for RLVR. We propose PREPO with two complementary components. First, we adopt prompt perplexity as an indicator of model adaptability in learning, enabling the model to progress from well-understood contexts to more challenging ones. Second, we amplify the discrepancy among the rollouts by differentiating their relative entropy, and prioritize sequences that exhibit a higher degree of exploration. Together, these mechanisms reduce rollout demand while preserving competitive performance. On the Qwen and Llama models, PREPO achieves effective results on mathematical reasoning benchmarks with up to 3 times fewer rollouts than the baselines. Beyond empirical gains, we provide theoretical and in-depth analyses explaining the underlying rationale of our method to improve the data efficiency of RLVR.
Authors: Wang Zixian
Abstract: Pre-trained Transformers often exhibit over-confidence in source patterns and difficulty in forming new target-domain patterns during fine-tuning. We formalize the mechanism of output saturation leading to gradient suppression through standard cross-entropy and softmax analysis, showing that gradient suppression at inflection layers confines adaptation to high-level recombination of existing features while preventing low-level reconstruction. We introduce a set of layer-wise diagnostic metrics -- attention entropy (saturation proxy), activation gradient norm, parameter gradient norm, and Delta-CKA under a shared PCA basis -- to identify inflection layers characterized by both low attention entropy and steep gradient decay. Building on these findings, we propose a diagnose-first, inject-light fine-tuning strategy: selectively inserting LoRA adapters at inflection layers to restore suppressed backward signals with minimal parameter overhead. Experiments on BERT-base transfer from SST-2 to Rotten Tomatoes under under-trained and over-trained source regimes reveal that over-trained initialization benefits from inflection-layer LoRA injection, while under-trained initialization suffers performance degradation. When base features are strong, unblocking inflection layers facilitates high-level compositional adaptation; when base features are weak, full-pathway unblocking is required for low-level reconstruction, as supported by joint analysis of layer-wise activation gradients and Delta-CKA dynamics.
Authors: Abhiram Kusumba, Maitreya Patel, Kyle Min, Changhoon Kim, Chitta Baral, Yezhou Yang
Abstract: Erasing harmful or proprietary concepts from powerful text to image generators is an emerging safety requirement, yet current "concept erasure" techniques either collapse image quality, rely on brittle adversarial losses, or demand prohibitive retraining cycles. We trace these limitations to a myopic view of the denoising trajectories that govern diffusion based generation. We introduce EraseFlow, the first framework that casts concept unlearning as exploration in the space of denoising paths and optimizes it with GFlowNets equipped with the trajectory balance objective. By sampling entire trajectories rather than single end states, EraseFlow learns a stochastic policy that steers generation away from target concepts while preserving the model's prior. EraseFlow eliminates the need for carefully crafted reward models and by doing this, it generalizes effectively to unseen concepts and avoids hackable rewards while improving the performance. Extensive empirical results demonstrate that EraseFlow outperforms existing baselines and achieves an optimal trade off between performance and prior preservation.
Authors: Guangxi Wan, Peng Zeng, Xiaoting Dong, Chunhe Song, Shijie Cui, Dong Li, Qingwei Dong, Yiyang Liu, Hongfei Bai
Abstract: Cyber-physical systems (CPS) require the joint optimization of discrete cyber actions and continuous physical parameters under stringent safety logic constraints. However, existing hierarchical approaches often compromise global optimality, whereas reinforcement learning (RL) in hybrid action spaces often relies on brittle reward penalties, masking, or shielding and struggles to guarantee constraint satisfaction. We present logic-informed reinforcement learning (LIRL), which equips standard policy-gradient algorithms with projection that maps a low-dimensional latent action onto the admissible hybrid manifold defined on-the-fly by first-order logic. This guarantees feasibility of every exploratory step without penalty tuning. Experimental evaluations have been conducted across multiple scenarios, including industrial manufacturing, electric vehicle charging stations, and traffic signal control, in all of which the proposed method outperforms existing hierarchical optimization approaches. Taking a robotic reducer assembly system in industrial manufacturing as an example, LIRL achieves a 36.47\% to 44.33\% reduction at most in the combined makespan-energy objective compared to conventional industrial hierarchical scheduling methods. Meanwhile, it consistently maintains zero constraint violations and significantly surpasses state-of-the-art hybrid-action reinforcement learning baselines. Thanks to its declarative logic-based constraint formulation, the framework can be seamlessly transferred to other domains such as smart transportation and smart grid, thereby paving the way for safe and real-time optimization in large-scale CPS.
Authors: Runyu Lu, Peng Zhang, Ruochuan Shi, Yuanheng Zhu, Dongbin Zhao, Yang Liu, Dong Wang, Cesare Alippi
Abstract: Equilibrium learning in adversarial games is an important topic widely examined in the fields of game theory and reinforcement learning (RL). Pursuit-evasion game (PEG), as an important class of real-world games from the fields of robotics and security, requires exponential time to be accurately solved. When the underlying graph structure varies, even the state-of-the-art RL methods require recomputation or at least fine-tuning, which can be time-consuming and impair real-time applicability. This paper proposes an Equilibrium Policy Generalization (EPG) framework to effectively learn a generalized policy with robust cross-graph zero-shot performance. In the context of PEGs, our framework is generally applicable to both pursuer and evader sides in both no-exit and multi-exit scenarios. These two generalizability properties, to our knowledge, are the first to appear in this domain. The core idea of the EPG framework is to train an RL policy across different graph structures against the equilibrium policy for each single graph. To construct an equilibrium oracle for single-graph policies, we present a dynamic programming (DP) algorithm that provably generates pure-strategy Nash equilibrium with near-optimal time complexity. To guarantee scalability with respect to pursuer number, we further extend DP and RL by designing a grouping mechanism and a sequence model for joint policy decomposition, respectively. Experimental results show that, using equilibrium guidance and a distance feature proposed for cross-graph PEG training, the EPG framework guarantees desirable zero-shot performance in various unseen real-world graphs. Besides, when trained under an equilibrium heuristic proposed for the graphs with exits, our generalized pursuer policy can even match the performance of the fine-tuned policies from the state-of-the-art PEG methods.
Authors: Shashank Nag, Alan T. L. Bacellar, Zachary Susskind, Anshul Jha, Logan Liberty, Aishwarya Sivakumar, Eugene B. John, Krishnan Kailas, Priscila M. V. Lima, Neeraja J. Yadwadkar, Felipe M. G. Franca, Lizy K. John
Abstract: Vision Transformers have been tremendously successful in computer vision tasks. However, their large computational, memory, and energy demands are a challenge for edge inference on FPGAs -- a field that has seen a recent surge in demand. We recognize the benefits of recent works on logic and Look Up Table (LUT) based networks, such as LogicNets, NeuraLUT, DWN, among others, in offering models that simultaneously reduce both the memory and compute footprints. However, these models natively do not perform well on common vision tasks, such as CIFAR-10/100. In this work, we propose LL-ViT, a novel edge optimized vision transformer design that integrates layers of LUT neurons within the transformer architecture. Based on our characterization that reveals that a majority of model weights and computations are from the channel mixer (MLP layer), we design an alternate LUT-based channel mixer, and simultaneously develop an FPGA-based accelerator for LL-ViT. Contrary to some attempts to replace each multiplication with a table lookup, our architecture utilizes a neural learning approach which natively learns the LUT functions. This approach allows for reduced model sizes, and a computational and energy-efficient inference solution for vision transformer models. Evaluating on edge-suitable workloads, we achieve accuracies of 95.5% on CIFAR-10, 78.8% on CIFAR-100, and 60.9% on Tiny-ImageNet datasets, comparable to the baseline transformer. LL-ViT eliminates over 60% of the model weights and 50% of the multiplications in the model, and achieves 1.9x energy efficiency and 1.3x lower latency over an integer quantized ViT accelerator, while also offering superior throughput against prior works at a 10.9W power budget.
Authors: Abhishek Patange, Sharat Chidambaran, Prabhat Shankar, Manjunath G. B., Anindya Chatterjee
Abstract: Slug formation in oil and gas pipelines poses significant challenges to operational safety and efficiency, yet existing detection approaches are often offline, require domain expertise, and lack real-time interpretability. We present an interactive application that enables end-to-end data-driven slug detection through a compact and user-friendly interface. The system integrates data exploration and labeling, configurable model training and evaluation with multiple classifiers, visualization of classification results with time-series overlays, and a real-time inference module that generates persistence-based alerts when slug events are detected. The demo supports seamless workflows from labeled CSV uploads to live inference on unseen datasets, making it lightweight, portable, and easily deployable. By combining domain-relevant analytics with novel UI/UX features such as snapshot persistence, visual labeling, and real-time alerting, our tool adds significant dissemination value as both a research prototype and a practical industrial application. The demo showcases how interactive human-in-the-loop ML systems can bridge the gap between data science methods and real-world decision-making in critical process industries, with broader applicability to time-series fault diagnosis tasks beyond oil and gas.
Authors: Nazmul Takbir, Hamidreza Alikhani, Nikil Dutt, Sangeetha Abdu Jyothi
Abstract: Large Language Model (LLM) serving is increasingly constrained by the growing size of the key-value (KV) cache, which scales with both context length and generation length. Prior work shows that attention is dominated by a small subset of critical tokens, yet existing systems struggle to exploit this efficiently without degrading accuracy, especially in long generation. We make a key observation: the temporal stability of these critical tokens varies significantly across KV heads: some heads consistently focus on the same tokens, while others shift frequently. Building on this insight, we introduce FlexiCache, a hierarchical KV-cache management system that leverages the temporal stability of KV heads to reduce GPU memory usage and computation overhead, while preserving model accuracy. FlexiCache classifies KV heads as stable or unstable: it retains all KV-cache pages from unstable heads in GPU memory, whereas for stable heads, it keeps only the top-K pages on the GPU and offloads the rest to host memory. By exploiting temporal stability, FlexiCache performs periodic reranking for stable heads to fetch newly promoted top pages. Implemented atop vLLM, FlexiCache reduces GPU memory footprint for long-context requests by up to 70%, improves offline serving throughput by 1.38-1.55x, and lowers online token latency by 1.6-2.1x, all while maintaining accuracy in long-context, long-generation scenarios.
Authors: Taowen Liu, Marta Andronic, Deniz G\"und\"uz, George A. Constantinides
Abstract: LLM training is resource-intensive. Quantized training improves computational and memory efficiency but introduces quantization noise, which can hinder convergence and degrade model accuracy. Stochastic Rounding (SR) has emerged as a theoretically attractive alternative to deterministic rounding, offering unbiased gradient estimates. However, its interaction with other training factors -- especially batch size -- remains under explored. In this paper, we present a theoretical and empirical study of mini-batch stochastic gradient descent (SGD) with SR, showing that increased batch sizes can compensate for reduced precision during back-propagation. Furthermore, we show that quantizing weights and activations impacts gradient variance in distinct ways. Our experiments validate these theoretical insights.
Authors: Joonyoung Lim, Younghwan Yoo
Abstract: We propose KFCPO, a novel Safe Reinforcement Learning (Safe RL) algorithm that combines scalable Kronecker-Factored Approximate Curvature (K-FAC) based second-order policy optimization with safety-aware gradient manipulation. KFCPO leverages K-FAC to perform efficient and stable natural gradient updates by approximating the Fisher Information Matrix (FIM) in a layerwise, closed form manner, avoiding iterative approximation overheads. To address the tradeoff between reward maximization and constraint satisfaction, we introduce a margin aware gradient manipulation mechanism that adaptively adjusts the influence of reward and cost gradients based on the agent's proximity to safety boundaries. This method blends gradients using a direction sensitive projection, eliminating harmful interference and avoiding abrupt changes caused by fixed hard thresholds. Additionally, a minibatch level KL rollback strategy is adopted to ensure trust region compliance and to prevent destabilizing policy shifts. Experiments on Safety Gymnasium using OmniSafe show that KFCPO achieves 10.3% to 50.2% higher average return across environments compared to the best baseline that respected the safety constraint, demonstrating superior balance of safety and performance.
Authors: Tal Argov, Tal Wagner
Abstract: Explainable clustering by axis-aligned decision trees was introduced by Moshkovitz et al. (2020) and has gained considerable interest. Prior work has focused on minimizing the price of explainability for specific clustering objectives, lacking a general method to fit an explanation tree to any given clustering, without restrictions. In this work, we propose a new and generic approach to explainable clustering, based on spectral graph partitioning. With it, we design an explainable clustering algorithm that can fit an explanation tree to any given non-explainable clustering, or directly to the dataset itself. Moreover, we show that prior algorithms can also be interpreted as graph partitioning, through a generalized framework due to Trevisan (2013) wherein cuts are optimized in two graphs simultaneously. Our experiments show the favorable performance of our method compared to baselines on a range of datasets.
Authors: Yoshihiro Maruyama
Abstract: Human activity recognition is challenging because sensor signals shift with context, motion, and environment; effective models must therefore remain stable as the world around them changes. We introduce a categorical symmetry-aware learning framework that captures how signals vary over time, scale, and sensor hierarchy. We build these factors into the structure of feature representations, yielding models that automatically preserve the relationships between sensors and remain stable under realistic distortions such as time shifts, amplitude drift, and device orientation changes. On the UCI Human Activity Recognition benchmark, this categorical symmetry-driven design improves out-of-distribution accuracy by approx. 46 percentage points (approx. 3.6x over the baseline), demonstrating that abstract symmetry principles can translate into concrete performance gains in everyday sensing tasks via category-equivariant representation theory.
Authors: Ernesto Araya, Massimiliano Datres, Gitta Kutyniok
Abstract: Spiking neural networks (SNNs) are a promising paradigm for energy-efficient computation, yet their theoretical foundations-especially regarding stability and robustness-remain limited compared to artificial neural networks. In this work, we study discrete-time leaky integrate-and-fire (LIF) SNNs through the lens of Boolean function analysis. We focus on noise sensitivity and stability in classification tasks, quantifying how input perturbations affect outputs. Our main result shows that wide LIF-SNN classifiers are stable on average, a property explained by the concentration of their Fourier spectrum on low-frequency components. Motivated by this, we introduce the notion of spectral simplicity, which formalizes simplicity in terms of Fourier spectrum concentration and connects our analysis to the simplicity bias observed in deep networks. Within this framework, we show that random LIF-SNNs are biased toward simple functions. Experiments on trained networks confirm that these stability properties persist in practice. Together, these results provide new insights into the stability and robustness properties of SNNs.
Authors: Ruifeng Ren, Sheng Ouyang, Huayi Tang, Yong Liu
Abstract: Transformers have demonstrated strong adaptability across a wide range of tasks and have become the backbone of modern Large Language Models (LLMs). However, their underlying mechanisms remain open for further exploration. The energy-based perspective has long provided a valuable principle for understanding neural computation. In this paper, we revisit the principle of energy as a lens to understand attention-based Transformer models. We present a unified energy-based framework which is composed of three key components: the global energy $F^*$, the energy function $E_i$ and the employed gradient descent (GD) form. Within this framework, standard softmax attention can be viewed as a special case of minimizing the Helmholtz free energy as $F^*$ using standard GD when $E_i$ takes the form of elastic potential energy, with residual connections ensuring that this optimization proceeds in an incremental manner. In addition, linear attentions can also be naturally incorporated into this framework by adjusting the corresponding energy forms. We also extend the above analysis to the multi-head setting, where the energy is defined across multiple low-dimensional subspaces. Building on this framework, we propose energy-based modifications of attention structures. Inspired by classical GD algorithms, we extend the original attention formulation based on standard GD to the momentum-based GD, Nesterov Accelerated Gradient (NAG), and Newton's method variants, each inducing a corresponding new attention structure. Our experiments provide preliminary support for the potential of the energy-based framework for designing attention mechanisms.
Authors: Yangyang Zhao, Matti Kaisti, Olli Lahdenoja, Tero Koivisto
Abstract: Atrial fibrillation (AF) is a leading cause of stroke and mortality, particularly in elderly patients. Wrist-worn photoplethysmography (PPG) enables non-invasive, continuous rhythm monitoring, yet suffers from significant vulnerability to motion artifacts and physiological noise. Many existing approaches rely solely on single-channel PPG and are limited to binary AF detection, often failing to capture the broader range of arrhythmias encountered in clinical settings. We introduce RhythmiNet, a residual neural network enhanced with temporal and channel attention modules that jointly leverage PPG and accelerometer (ACC) signals. The model performs three-class rhythm classification: AF, sinus rhythm (SR), and Other. To assess robustness across varying movement conditions, test data are stratified by accelerometer-based motion intensity percentiles without excluding any segments. RhythmiNet achieved a 4.3% improvement in macro-AUC over the PPG-only baseline. In addition, performance surpassed a logistic regression model based on handcrafted HRV features by 12%, highlighting the benefit of multimodal fusion and attention-based learning in noisy, real-world clinical data.
Authors: Khoat Than
Abstract: Normalization methods are fundamental components of modern deep neural networks (DNNs). Empirically, they are known to stabilize optimization dynamics and improve generalization. However, the underlying theoretical mechanism by which normalization contributes to both optimization and generalization remains largely unexplained, especially when using many normalization layers in a DNN architecture. In this work, we develop a theoretical framework that elucidates the role of normalization through the lens of capacity control. We prove that an unnormalized DNN can exhibit exponentially large Lipschitz constants with respect to either its parameters or inputs, implying excessive functional capacity and potential overfitting. Such bad DNNs are uncountably many. In contrast, the insertion of normalization layers provably can reduce the Lipschitz constant at an exponential rate in the number of normalization operations. This exponential reduction yields two fundamental consequences: (1) it smooths the loss landscape at an exponential rate, facilitating faster and more stable optimization; and (2) it constrains the effective capacity of the network, thereby enhancing generalization guarantees on unseen data. Our results thus offer a principled explanation for the empirical success of normalization methods in deep learning.
Authors: Hai Hoang Thanh, Duy-Tung Nguyen, Hung The Tran, Khoat Than
Abstract: Accurately evaluating model performance is crucial for deploying machine learning systems in real-world applications. Traditional methods often require a sufficiently large labeled test set to ensure a reliable evaluation. However, in many contexts, a large labeled dataset is costly and labor-intensive. Therefore, we sometimes have to do evaluation by a few labeled samples, which is theoretically challenging. Recent advances in generative models offer a promising alternative by enabling the synthesis of high-quality data. In this work, we make a systematic investigation about the use of synthetic data to estimate the test error of a trained model under limited labeled data conditions. To this end, we develop novel generalization bounds that take synthetic data into account. Those bounds suggest novel ways to optimize synthetic samples for evaluation and theoretically reveal the significant role of the generator's quality. Inspired by those bounds, we propose a theoretically grounded method to generate optimized synthetic data for model evaluation. Experimental results on simulation and tabular datasets demonstrate that, compared to existing baselines, our method achieves accurate and more reliable estimates of the test error.
Authors: Kristiyan Sakalyan, Alessandro Palma, Filippo Guerranti, Fabian J. Theis, Stephan G\"unnemann
Abstract: Understanding the evolution of cellular microenvironments in spatiotemporal data is essential for deciphering tissue development and disease progression. While experimental techniques like spatial transcriptomics now enable high-resolution mapping of tissue organization across space and time, current methods that model cellular evolution operate at the single-cell level, overlooking the coordinated development of cellular states in a tissue. We introduce NicheFlow, a flow-based generative model that infers the temporal trajectory of cellular microenvironments across sequential spatial slides. By representing local cell neighborhoods as point clouds, NicheFlow jointly models the evolution of cell states and spatial coordinates using optimal transport and Variational Flow Matching. Our approach successfully recovers both global spatial architecture and local microenvironment composition across diverse spatiotemporal datasets, from embryonic to brain development.
Authors: Rongrong Xie, Guido Sanguinetti
Abstract: Multimodal learning has increasingly become a focal point in research, primarily due to its ability to integrate complementary information from diverse modalities. Nevertheless, modality imbalance, stemming from factors such as insufficient data acquisition and disparities in data quality, has often been inadequately addressed. This issue is particularly prominent in biological data analysis, where datasets are frequently limited, costly to acquire, and inherently heterogeneous in quality. Conventional multimodal methodologies typically fall short in concurrently harnessing intermodal synergies and effectively resolving modality conflicts. In this study, we propose a novel unified framework explicitly designed to address modality imbalance by utilizing mutual information to quantify interactions between modalities. Our approach adopts a balanced multimodal learning strategy comprising two key stages: cross-modal knowledge distillation (KD) and a multitask-like training paradigm. During the cross-modal KD pretraining phase, stronger modalities are leveraged to enhance the predictive capabilities of weaker modalities. Subsequently, our primary training phase employs a multitask-like learning mechanism, dynamically calibrating gradient contributions based on modality-specific performance metrics and intermodal mutual information. This approach effectively alleviates modality imbalance, thereby significantly improving overall multimodal model performance.
Authors: Asal Meskin, Alireza Mirrokni, Ali Najar, Ali Behrouz
Abstract: In recent years, effectively modeling multivariate time series has gained significant popularity, mainly due to its wide range of applications, ranging from healthcare to financial markets and energy management. Transformers, MLPs, and linear models as the de facto backbones of modern time series models have shown promising results in single-variant and/or short-term forecasting. These models, however: (1) are permutation equivariant and so lack temporal inductive bias, being less expressive to capture the temporal dynamics; (2) are naturally designed for univariate setup, missing the inter-dependencies of temporal and variate dimensions; and/or (3) are inefficient for Long-term time series modeling. To overcome training and inference efficiency as well as the lack of temporal inductive bias, recently, linear Recurrent Neural Networks (RNNs) have gained attention as an alternative to Transformer-based models. These models, however, are inherently limited to a single sequence, missing inter-variate dependencies, and can propagate errors due to their additive nature. In this paper, we present Hydra, a by-design two-headed meta in-context memory module that learns how to memorize patterns at test time by prioritizing time series patterns that are more informative about the data. Hydra uses a 2-dimensional recurrence across both time and variate at each step, which is more powerful than mixing methods. Although the 2-dimensional nature of the model makes its training recurrent and non-parallelizable, we present a new 2D-chunk-wise training algorithm that approximates the actual recurrence with $\times 10$ efficiency improvement, while maintaining the effectiveness. Our experimental results on a diverse set of tasks and datasets, including time series forecasting, classification, and anomaly detection show the superior performance of Hydra compared to state-of-the-art baselines.
Authors: Diantong Li, Kyunghyun Cho, Chong Liu
Abstract: Bayesian Optimization (BO) is an efficient tool for optimizing black-box functions, but its theoretical guarantees typically hold in the asymptotic regime. In many critical real-world applications such as drug discovery or materials design, where each evaluation can be very costly and time-consuming, BO becomes impractical for many evaluations. In this paper, we introduce the Procedure-inFormed BO (ProfBO) algorithm, which solves black-box optimization with remarkably few function evaluations. At the heart of our algorithmic design are Markov Decision Process (MDP) priors that model optimization trajectories from related source tasks, thereby capturing procedural knowledge on efficient optimization. We embed these MDP priors into a prior-fitted neural network and employ model-agnostic meta-learning for fast adaptation to new target tasks. Experiments on real-world Covid and Cancer benchmarks and hyperparameter tuning tasks demonstrate that ProfBO consistently outperforms state-of-the-art methods by achieving high-quality solutions with significantly fewer evaluations, making it ready for practical deployment.
Authors: Fabricio Olivetti de Franca, Gabriel Kronberger
Abstract: In Symbolic Regression (SR), Genetic Programming (GP) is a popular search algorithm that delivers state-of-the-art results in term of accuracy. Its success relies on the concept of neutrality, which induces large plateaus that the search can safely navigate to more promising regions. Navigating these plateaus, while necessary, requires the computation of redundant expressions, up to 60% of the total number of evaluation, as noted in a recent study. The equality graph (e-graph) structure can compactly store and group equivalent expressions enabling us to verify if a given expression and their variations were already visited by the search, thus enabling us to avoid unnecessary computation. We propose a new search algorithm for symbolic regression called SymRegg that revolves around the e-graph structure following simple steps: perturb solutions sampled from a selection of expressions stored in the e-graph, if it generates an unvisited expression, insert it into the e-graph and generates its equivalent forms. We show that SymRegg is capable of improving the efficiency of the search, maintaining consistently accurate results across different datasets while requiring a choice of a minimalist set of hyperparameters.
Authors: Nabeel Seedat, Jiashuo Liu, Mihaela van der Schaar
Abstract: The recent surge in AI agents that autonomously communicate, collaborate with humans and use diverse tools has unlocked promising opportunities in various real-world settings. However, a vital aspect remains underexplored: how agents handle data. Scalable autonomy demands agents that continuously acquire, process, and evolve their data. In this paper, we argue that data-savvy capabilities should be a top priority in the design of agentic systems to ensure reliable real-world deployment. Specifically, we propose four key capabilities to realize this vision: (1) Proactive data acquisition: enabling agents to autonomously gather task-critical knowledge or solicit human input to address data gaps; (2) Sophisticated data processing: requiring context-aware and flexible handling of diverse data challenges and inputs; (3) Interactive test data synthesis: shifting from static benchmarks to dynamically generated interactive test data for agent evaluation; and (4) Continual adaptation: empowering agents to iteratively refine their data and background knowledge to adapt to shifting environments. While current agent research predominantly emphasizes reasoning, we hope to inspire a reflection on the role of data-savvy agents as the next frontier in data-centric AI.
Authors: Haoran Ye, Qiuzhuang Sun, Yang Yang
Abstract: This study develops a SARIMAX-based prediction system for short-term power outage forecasting during extreme weather events. Using hourly data from Michigan counties with outage counts and comprehensive weather features, we implement a systematic two-stage feature engineering pipeline: data cleaning to remove zero-variance and unknown features, followed by correlation-based filtering to eliminate highly correlated predictors. The selected features are augmented with temporal embeddings, multi-scale lag features, and weather variables with their corresponding lags as exogenous inputs to the SARIMAX model. To address data irregularity and numerical instability, we apply standardization and implement a hierarchical fitting strategy with sequential optimization methods, automatic downgrading to ARIMA when convergence fails, and historical mean-based fallback predictions as a final safeguard. The model is optimized separately for short-term (24 hours) and medium-term (48 hours) forecast horizons using RMSE as the evaluation metric. Our approach achieves an RMSE of 177.2, representing an 8.4\% improvement over the baseline method (RMSE = 193.4), thereby validating the effectiveness of our feature engineering and robust optimization strategy for extreme weather-related outage prediction.
Authors: Sama Salarian, Yue Zhang, Swati Padhee, Srinivasan Parthasarathy
Abstract: Synthetic healthcare data generation presents a viable approach to enhance data accessibility and support research by overcoming limitations associated with real-world medical datasets. However, ensuring fairness across protected attributes in synthetic data is critical to avoid biased or misleading results in clinical research and decision-making. In this study, we assess the fairness of synthetic data generated by multiple generative adversarial network (GAN)-based models using the MIMIC-III dataset, with a focus on representativeness across protected demographic attributes. We measure subgroup representation using the logarithmic disparity metric and observe significant imbalances, with many subgroups either underrepresented or overrepresented in the synthetic data, compared to the real data. To mitigate these disparities, we introduce MedEqualizer, a model-agnostic augmentation framework that enriches the underrepresented subgroups prior to synthetic data generation. Our results show that MedEqualizer significantly improves demographic balance in the resulting synthetic datasets, offering a viable path towards more equitable and representative healthcare data synthesis.
Authors: Andrew Hallam, R G Gayathri, Glory Lee, Atul Sajjanhar
Abstract: Cognitive workload is a topic of increasing interest across various fields such as health, psychology, and defense applications. In this research, we focus on classifying cognitive workload using the COLET dataset, employing a window-based approach for feature generation and machine/deep learning techniques for classification. We apply window-based temporal partitioning to enhance features used in existing research, followed by machine learning and deep learning models to classify different levels of cognitive workload. The results demonstrate that deep learning models, particularly tabular architectures, outperformed traditional machine learning methods in precision, F1-score, accuracy, and classification precision. This study highlights the effectiveness of window-based temporal feature extraction and the potential of deep learning techniques for real-time cognitive workload assessment in complex and dynamic tasks.
Authors: Przemys{\l}aw Spyra, Witold Dzwinel
Abstract: The long-held assumption that backpropagation (BP) is essential for state-of-the-art performance is challenged by this work. We present rigorous, hardware-validated evidence that the Mono-Forward (MF) algorithm, a backpropagation-free method, consistently surpasses an optimally tuned BP baseline in classification accuracy on its native Multi-Layer Perceptron (MLP) architectures. This superior generalization is achieved with profound efficiency gains, including up to 41% less energy consumption and up to 34% faster training. Our analysis, which charts an evolutionary path from Geoffrey Hinton's Forward-Forward (FF) to the Cascaded Forward (CaFo) and finally to MF, is grounded in a fair comparative framework using identical architectures and universal hyperparameter optimization. We further provide a critical re-evaluation of memory efficiency in BP-free methods, empirically demonstrating that practical overhead can offset theoretical gains. Ultimately, this work establishes MF as a practical, high-performance, and sustainable alternative to BP for MLPs.
Authors: Georg Pichler, Marco Romanelli, Pablo Piantanida
Abstract: In this paper, we propose a novel fairness framework grounded in the concept of happi- ness, a measure of the utility each group gains fromdecisionoutcomes. Bycapturingfairness through this intuitive lens, we not only offer a more human-centered approach, but also one that is mathematically rigorous: In order to compute the optimal, fair post-processing strategy, only a linear program needs to be solved. This makes our method both efficient and scalable with existing optimization tools. Furthermore, it unifies and extends several well-known fairness definitions, and our em- pirical results highlight its practical strengths across diverse scenarios.
Authors: David McCoy, Yulun Wu, Zachary Butzin-Dozier
Abstract: This position paper challenges the "scaling fundamentalism" dominating AI research, where unbounded growth in model size and computation has led to unsustainable environmental impacts and widening resource inequality. We argue that LLM development should be fundamentally reoriented toward capability-per-resource rather than capability alone. We present a theoretical framework demonstrating that resource-allocation decisions guided by gradient influence patterns can dramatically improve efficiency throughout the AI lifecycle. Our analysis shows that in transformer-based models, where a small fraction of parameters exert outsized influence (following heavy-tailed distributions), three critical insights emerge: (1) updating only high-influence parameters strictly outperforms full-parameter tuning on a performance-per-resource basis; (2) simple gradient norms provide computationally efficient proxies for identifying these high-influence components; and (3) coordinated parameter and data selection yields multiplicative efficiency gains, potentially reducing resource requirements by orders of magnitude. Building on these theoretical foundations, we propose a two stage paradigm marginal-return pretraining for foundation developers and influence guided adaptation for downstream users bridged by gradient blueprints, metadata describing which parameters matter most for various tasks. This capability-per-resource perspective transforms what were once considered pragmatic hardware workarounds into theoretically optimal strategies, democratizing access to cutting-edge AI capabilities while significantly reducing environmental impact. By embedding resource consciousness into how we develop, adapt, and evaluate models, we can reshape AI progress toward a more sustainable and equitable future.
Authors: Aman Jaglan, Jarrod Barnes
Abstract: Continual Learning (CL) methods have traditionally focused on mitigating catastrophic forgetting through gradient-based retraining, an approach ill-suited for deployed agents that must adapt in real time. We introduce our Adaptive Teaching and Learning System (ATLAS), a dual-agent architecture that decouples reasoning (Teacher) from execution (Student) and incorporates a persistent learning memory that stores distilled guidance from experience. This informs the orchestration layer, enabling the system to dynamically adjust its operational strategies, such as supervision level or initial plan selection, at inference time. In doing so, ATLAS achieves gradient-free continual learning, shifting the locus of adaptation from model parameters to system-level orchestration. We formulate this as a system-centric paradigm for continual learning, where the objective is adaptive efficiency: maximizing task success while minimizing computational cost through inference-time orchestration rather than parameter updates. Evaluated on Microsoft's ExCyTIn-Bench, an open-source benchmark simulating complex cyberthreat investigation, ATLAS achieves 54.1% success with GPT-5-mini as its Student, outperforming the larger GPT-5 (High) by 13% while reducing cost by 86%. Cross-incident validation demonstrates generalization: frozen pamphlets from Incident #5 improve accuracy from 28% to 41% with zero retraining, while shifting output composition from verbose exploration to structured reasoning. Together, these findings establish gradient-free continual learning as a viable path toward adaptive, deployable AI systems and provide causally annotated traces valuable for training explicit world models.
Authors: Takashi Furuya, Anastasis Kratsios, Dylan Possama\"i, Bogdan Raoni\'c
Abstract: We introduce a mild generative variant of the classical neural operator model, which leverages Kolmogorov--Arnold networks to solve infinite families of second-order backward stochastic differential equations ($2$BSDEs) on regular bounded Euclidean domains with random terminal time. Our first main result shows that the solution operator associated with a broad range of $2$BSDE families is approximable by appropriate neural operator models. We then identify a structured subclass of (infinite) families of $2$BSDEs whose neural operator approximation requires only a polynomial number of parameters in the reciprocal approximation rate, as opposed to the exponential requirement in general worst-case neural operator guarantees.
Authors: Parvin Nazari, Bojian Hou, Davoud Ataee Tarzanagh, Li Shen, George Michailidis
Abstract: Online bilevel optimization (OBO) is a powerful framework for machine learning problems where both outer and inner objectives evolve over time, requiring dynamic updates. Current OBO approaches rely on deterministic \textit{window-smoothed} regret minimization, which may not accurately reflect system performance when functions change rapidly. In this work, we introduce a novel search direction and show that both first- and zeroth-order (ZO) stochastic OBO algorithms leveraging this direction achieve sublinear {stochastic bilevel regret without window smoothing}. Beyond these guarantees, our framework enhances efficiency by: (i) reducing oracle dependence in hypergradient estimation, (ii) updating inner and outer variables alongside the linear system solution, and (iii) employing ZO-based estimation of Hessians, Jacobians, and gradients. Experiments on online parametric loss tuning and black-box adversarial attacks validate our approach.
Authors: Kathryn Lindsey, Govind Menon
Abstract: We use geometric invariant theory (GIT) to study the deep linear network (DLN). The Kempf-Ness theorem is used to establish that the $L^2$ regularizer is minimized on the balanced manifold. This allows us to decompose the training dynamics into two distinct gradient flows: a regularizing flow on fibers and a learning flow on the balanced manifold. We show that the regularizing flow is exactly solvable using the moment map. This approach provides a common mathematical framework for balancedness in deep learning and linear systems theory. We use this framework to interpret balancedness in terms of model reduction and Bayesian principles.
Authors: Ali Owfi, Amirmohammad Bamdad, Tolunay Seyfi, Fatemeh Afghah
Abstract: Deep learning has emerged as a leading approach for Automatic Modulation Classification (AMC), demonstrating superior performance over traditional methods. However, vulnerability to adversarial attacks and susceptibility to data distribution shifts hinder their practical deployment in real-world, dynamic environments. To address these threats, we propose a novel, unified framework that integrates meta-learning with domain adaptation, making AMC systems resistant to both adversarial attacks and environmental changes. Our framework utilizes a two-phase strategy. First, in an offline phase, we employ a meta-learning approach to train the model on clean and adversarially perturbed samples from a single source domain. This method enables the model to generalize its defense, making it resistant to a combination of previously unseen attacks. Subsequently, in the online phase, we apply domain adaptation to align the model's features with a new target domain, allowing it to adapt without requiring substantial labeled data. As a result, our framework achieves a significant improvement in modulation classification accuracy against these combined threats, offering a critical solution to the deployment and operational challenges of modern AMC systems.
Authors: Ruyue Zhang, Xiaopeng Ke, Ming Liu, Fangzhou Shi, Chang Men, Zhengdan Zhu
Abstract: Uplift modeling has emerged as a crucial technique for individualized treatment effect estimation, particularly in fields such as marketing and healthcare. Modeling uplift effects in multi-treatment scenarios plays a key role in real-world applications. Current techniques for modeling multi-treatment uplift are typically adapted from binary-treatment works. In this paper, we investigate and categorize all current model adaptations into two types: Structure Adaptation and Feature Adaptation. Through our empirical experiments, we find that these two adaptation types cannot maintain effectiveness under various data characteristics (noisy data, mixed with observational data, etc.). To enhance estimation ability and robustness, we propose Orthogonal Function Adaptation (OFA) based on the function approximation theorem. We conduct comprehensive experiments with multiple data characteristics to study the effectiveness and robustness of all model adaptation techniques. Our experimental results demonstrate that our proposed OFA can significantly improve uplift model performance compared to other vanilla adaptation methods and exhibits the highest robustness.
Authors: Lijia Yu, Xiao-Shan Gao, Lijun Zhang
Abstract: It has been shown that the chain of thought (CoT) can enhance the power of large language models (LLMs) to solve certain mathematical reasoning problems. However, the capacity of CoT is still not fully explored. As an important instance, the following basic question has not yet been answered: Does CoT expand the capability of transformers across all reasoning tasks? We demonstrate that reasoning with transformers is essentially a memorization problem for reasoning datasets. Thus, examining the power of CoT across all reasoning tasks amounts to analyzing the memorization capabilities of CoT transformers. In this paper, we give a complete description of the memorization capabilities of fixed-precision transformers with or without CoT and give a negative answer to the above-mentioned question. Precisely, we first give necessary and sufficient conditions for fixed-precision transformers with and without CoT to memorize a finite reasoning dataset and show that these two conditions do not imply each other. Then, we give lower and upper bounds for the number of parameters needed for transformers with or without CoT to memorize a finite reasoning dataset with $N$ elements, which are $\overline{\Theta}(N)$ in all cases. This implies that there exist reasoning tasks for which CoT does not enhance the reasoning power of transformers, leading to a negative answer to the above-mentioned question. Finally, we give the first results on memorizing infinite reasoning datasets by CoT transformers and show that some simple infinite datasets cannot be memorized by transformers with or without CoT.
Authors: Tariq Abdul-Quddoos, Tasnia Sharmin, Xiangfang Li, Lijun Qian
Abstract: As spectrum sharing becomes increasingly vital to meet rising wireless demands in the future, spectrum monitoring and transmitter identification are indispensable for enforcing spectrum usage policy, efficient spectrum utilization, and net- work security. This study proposed a robust framework for transmitter identification and protocol categorization via multi- task RF signal classification in shared spectrum environments, where the spectrum monitor will classify transmission protocols (e.g., 4G LTE, 5G-NR, IEEE 802.11a) operating within the same frequency bands, and identify different transmitting base stations, as well as their combinations. A Convolutional Neural Network (CNN) is designed to tackle critical challenges such as overlapping signal characteristics and environmental variability. The proposed method employs a multi-channel input strategy to extract meaningful signal features, achieving remarkable accuracy: 90% for protocol classification, 100% for transmitting base station classification, and 92% for joint classification tasks, utilizing RF data from the POWDER platform. These results highlight the significant potential of the proposed method to enhance spectrum monitoring, management, and security in modern wireless networks.
Authors: Pavel Rumiantsev, Soumyasundar Pal, Yingxue Zhang, Mark Coates
Abstract: The performance of Large Language Models (LLMs) and the associated dollar costs of API calls can fluctuate over time, potentially invalidating conclusions drawn in prior research. To address this, we propose a Fair Evaluation protocol for Test-Time Compute (FEval-TTC), designed to ensure consistent assessment of test-time compute (TTC) methods, regardless of such fluctuations. FEval-TTC focuses on the evaluation of TTC methods that utilize underlying Chains-of-Thought (CoT). It supports evaluations across multiple LLMs on a diverse set of mathematical and commonsense reasoning datasets. The few-shot prompting and answer extraction processes are standardized across datasets, reducing both time and monetary overhead for researchers. Furthermore, we provide a cost modelling procedure that estimates both the token and dollar cost per query, facilitating equitable comparisons of prevalent TTC methods. We open-source FEval-TTC for public use at https://github.com/networkslab/feval_ttc .
Authors: Minh-Duc Nguyen, Dung D. Le, Phi Long Nguyen
Abstract: The rapid growth of electric vehicles (EVs) necessitates the strategic placement of charging stations to optimize resource utilization and minimize user inconvenience. Reinforcement learning (RL) offers an innovative approach to identifying optimal charging station locations; however, existing methods face challenges due to their deterministic reward systems, which limit efficiency. Because real-world conditions are dynamic and uncertain, a deterministic reward structure cannot fully capture the complexities of charging station placement. As a result, evaluation becomes costly and time-consuming, and less reflective of real-world scenarios. To address this challenge, we propose a novel framework that integrates deep RL with agent-based simulations to model EV movement and estimate charging demand in real time. Our approach employs a hybrid RL agent with dual Q-networks to select optimal locations and configure charging ports, guided by a hybrid reward function that combines deterministic factors with simulation-derived feedback. Case studies in Hanoi, Vietnam, show that our method reduces average waiting times by 53.28% compared to the initial state, outperforming static baseline methods. This scalable and adaptive solution enhances EV infrastructure planning, effectively addressing real-world complexities and improving user experience.
Authors: Themistoklis Vargiemezis, Charilaos Kanatsoulis, Catherine Gorl\'e
Abstract: Accurate prediction of wind loading on buildings is crucial for structural safety and sustainable design, yet conventional approaches such as wind tunnel testing and large-eddy simulation (LES) are prohibitively expensive for large-scale exploration. Each LES case typically requires at least 24 hours of computation, making comprehensive parametric studies infeasible. We introduce WindMiL, a new machine learning framework that combines systematic dataset generation with symmetry-aware graph neural networks (GNNs). First, we introduce a large-scale dataset of wind loads on low-rise buildings by applying signed distance function interpolation to roof geometries and simulating 462 cases with LES across varying shapes and wind directions. Second, we develop a reflection-equivariant GNN that guarantees physically consistent predictions under mirrored geometries. Across interpolation and extrapolation evaluations, WindMiL achieves high accuracy for both the mean and the standard deviation of surface pressure coefficients (e.g., RMSE $\leq 0.02$ for mean $C_p$) and remains accurate under reflected-test evaluation, maintaining hit rates above $96\%$ where the non-equivariant baseline model drops by more than $10\%$. By pairing a systematic dataset with an equivariant surrogate, WindMiL enables efficient, scalable, and accurate predictions of wind loads on buildings.
Authors: Min Gan, Guang-Yong Chen, Yang Yi, Lin Yang
Abstract: The proliferation of saddle points, rather than poor local minima, is increasingly understood to be a primary obstacle in large-scale non-convex optimization for machine learning. Variable elimination algorithms, like Variable Projection (VarPro), have long been observed to exhibit superior convergence and robustness in practice, yet a principled understanding of why they so effectively navigate these complex energy landscapes has remained elusive. In this work, we provide a rigorous geometric explanation by comparing the optimization landscapes of the original and reduced formulations. Through a rigorous analysis based on Hessian inertia and the Schur complement, we prove that variable elimination fundamentally reshapes the critical point structure of the objective function, revealing that local maxima in the reduced landscape are created from, and correspond directly to, saddle points in the original formulation. Our findings are illustrated on the canonical problem of non-convex matrix factorization, visualized directly on two-parameter neural networks, and finally validated in training deep Residual Networks, where our approach yields dramatic improvements in stability and convergence to superior minima. This work goes beyond explaining an existing method; it establishes landscape simplification via saddle point transformation as a powerful principle that can guide the design of a new generation of more robust and efficient optimization algorithms.
Authors: Kun-Wei Lin, Yu-Chen Kuo, Hsin-Yao Wang, Yi-Ju Tseng
Abstract: Clinical risk prediction using electronic health records (EHRs) is vital to facilitate timely interventions and clinical decision support. However, modeling heterogeneous and irregular temporal EHR data presents significant challenges. We propose \textbf{KAT-GNN} (Knowledge-Augmented Temporal Graph Neural Network), a graph-based framework that integrates clinical knowledge and temporal dynamics for risk prediction. KAT-GNN first constructs modality-specific patient graphs from EHRs. These graphs are then augmented using two knowledge sources: (1) ontology-driven edges derived from SNOMED CT and (2) co-occurrence priors extracted from EHRs. Subsequently, a time-aware transformer is employed to capture longitudinal dynamics from the graph-encoded patient representations. KAT-GNN is evaluated on three distinct datasets and tasks: coronary artery disease (CAD) prediction using the Chang Gung Research Database (CGRD) and in-hospital mortality prediction using the MIMIC-III and MIMIC-IV datasets. KAT-GNN achieves state-of-the-art performance in CAD prediction (AUROC: 0.9269 $\pm$ 0.0029) and demonstrated strong results in mortality prediction in MIMIC-III (AUROC: 0.9230 $\pm$ 0.0070) and MIMIC-IV (AUROC: 0.8849 $\pm$ 0.0089), consistently outperforming established baselines such as GRASP and RETAIN. Ablation studies confirm that both knowledge-based augmentation and the temporal modeling component are significant contributors to performance gains. These findings demonstrate that the integration of clinical knowledge into graph representations, coupled with a time-aware attention mechanism, provides an effective and generalizable approach for risk prediction across diverse clinical tasks and datasets.
Authors: Yiyang Yang, Xiejian Chi, Shanxing Gao, Kaidong Wang, Yao Wang
Abstract: Data quality is critical to Intelligent Transportation Systems (ITS), as complete and accurate traffic data underpin reliable decision-making in traffic control and management. Recent advances in low-rank tensor recovery algorithms have shown strong potential in capturing the inherent structure of high-dimensional traffic data and restoring degraded observations. However, traditional batch-based methods demand substantial computational and storage resources, which limits their scalability in the face of continuously expanding traffic data volumes. Moreover, recent online tensor recovery methods often suffer from severe performance degradation in complex real-world scenarios due to their insufficient exploitation of the intrinsic structural properties of traffic data. To address these challenges, we reformulate the traffic data recovery problem within a streaming framework, and propose a novel online robust tensor recovery algorithm that simultaneously leverages both the global spatio-temporal correlations and local consistency of traffic data, achieving high recovery accuracy and significantly improved computational efficiency in large-scale scenarios. Our method is capable of simultaneously handling missing and anomalous values in traffic data, and demonstrates strong adaptability across diverse missing patterns. Experimental results on three real-world traffic datasets demonstrate that the proposed approach achieves high recovery accuracy while significantly improving computational efficiency by up to three orders of magnitude compared to state-of-the-art batch-based methods. These findings highlight the potential of the proposed approach as a scalable and effective solution for traffic data quality enhancement in ITS.
Authors: Zan Li, Kyongmin Yeo, Wesley Gifford, Lara Marcuse, Madeline Fields, B\"ulent Yener
Abstract: Forecasting epileptic seizures from multivariate EEG signals represents a critical challenge in healthcare time series prediction, requiring high sensitivity, low false alarm rates, and subject-specific adaptability. We present STAN, an Adversarial Spatio-Temporal Attention Network that jointly models spatial brain connectivity and temporal neural dynamics through cascaded attention blocks with alternating spatial and temporal modules. Unlike existing approaches that assume fixed preictal durations or separately process spatial and temporal features, STAN captures bidirectional dependencies between spatial and temporal patterns through a unified cascaded architecture. Adversarial training with gradient penalty enables robust discrimination between interictal and preictal states learned from clearly defined 15-minute preictal windows. Continuous 90-minute pre-seizure monitoring reveals that the learned spatio-temporal attention patterns enable early detection: reliable alarms trigger at subject-specific times (typically 15-45 minutes before onset), reflecting the model's capacity to capture subtle preictal dynamics without requiring individualized training. Experiments on two benchmark EEG datasets (CHB-MIT scalp: 8 subjects, 46 events; MSSM intracranial: 4 subjects, 14 events) demonstrate state-of-the-art performance: 96.6% sensitivity with 0.011 false detections per hour and 94.2% sensitivity with 0.063 false detections per hour, respectively, while maintaining computational efficiency (2.3M parameters, 45 ms latency, 180 MB memory) for real-time edge deployment. Beyond epilepsy, the proposed framework provides a general paradigm for spatio-temporal forecasting in healthcare and other time series domains where individual heterogeneity and interpretability are crucial.
Authors: Annabelle Martin, Daphne Kontogiorgos-Heintz, Jeff Nivala
Abstract: Nanopore protein sequencing produces long, noisy ionic current traces in which key molecular phases, such as protein capture and translocation, are embedded. Capture phases mark the successful entry of a protein into the pore and serve as both a checkpoint and a signal that a channel merits further analysis. However, manual identification of capture phases is time-intensive, often requiring several days for expert reviewers to annotate the data due to the need for domain-specific interpretation of complex signal patterns. To address this, a lightweight one-dimensional convolutional neural network (1D CNN) was developed and trained to detect capture phases in down-sampled signal windows. Evaluated against CNN-LSTM (Long Short-Term Memory) hybrids, histogram-based classifiers, and other CNN variants using run-level data splits, our best model, CaptureNet-Deep, achieved an F1 score of 0.94 and precision of 93.39% on held-out test data. The model supports low-latency inference and is integrated into a dashboard for Oxford Nanopore experiments, reducing the total analysis time from several days to under thirty minutes. These results show that efficient, real-time capture detection is possible using simple, interpretable architectures and suggest a broader role for lightweight ML models in sequencing workflows.
Authors: Yupu Lu, Shijie Lin, Hao Xu, Zeqing Zhang, Jia Pan
Abstract: Finding a control Lyapunov function (CLF) in a dynamical system with a controller is an effective way to guarantee stability, which is a crucial issue in safety-concerned applications. Recently, deep learning models representing CLFs have been applied into a learner-verifier framework to identify satisfiable candidates. However, the learner treats Lyapunov conditions as complex constraints for optimisation, which is hard to achieve global convergence. It is also too complicated to implement these Lyapunov conditions for verification. To improve this framework, we treat Lyapunov conditions as inductive biases and design a neural CLF and a CLF-based controller guided by this knowledge. This design enables a stable optimisation process with limited constraints, and allows end-to-end learning of both the CLF and the controller. Our approach achieves a higher convergence rate and larger region of attraction (ROA) in learning the CLF compared to existing methods among abundant experiment cases. We also thoroughly reveal why the success rate decreases with previous methods during learning.
Authors: Sivaram Krishnan, Jinho Choi, Jihong Park, Gregory Sherman, Benjamin Campbell
Abstract: The application of machine learning (ML) to communication systems is expected to play a pivotal role in future artificial intelligence (AI)-based next-generation wireless networks. While most existing works focus on ML techniques for static wireless environments, they often face limitations when applied to highly dynamic environments, such as flying ad hoc networks (FANETs). This paper explores the use of data-driven Koopman approaches to address these challenges. Specifically, we investigate how these approaches can model UAV trajectory dynamics within FANETs, enabling more accurate predictions and improved network performance. By leveraging Koopman operator theory, we propose two possible approaches -- centralized and distributed -- to efficiently address the challenges posed by the constantly changing topology of FANETs. To demonstrate this, we consider a FANET performing surveillance with UAVs following pre-determined trajectories and predict signal-to-interference-plus-noise ratios (SINRs) to ensure reliable communication between UAVs. Our results show that these approaches can accurately predict connectivity and isolation events that lead to modelled communication outages. This capability could help UAVs schedule their transmissions based on these predictions.
Authors: Guanjie Cheng, Mengzhen Yang, Xinkui Zhao, Shuyi Yu, Tianyu Du, Yangyang Wu, Mengying Zhu, Shuiguang Deng
Abstract: Federated learning (FL) enables collaborative model training across distributed nodes without exposing raw data, but its decentralized nature makes it vulnerable in trust-deficient environments. Inference attacks may recover sensitive information from gradient updates, while poisoning attacks can degrade model performance or induce malicious behaviors. Existing defenses often suffer from high communication and computation costs, or limited detection precision. To address these issues, we propose LSHFed, a robust and communication-efficient FL framework that simultaneously enhances aggregation robustness and privacy preservation. At its core, LSHFed incorporates LSHGM, a novel gradient verification mechanism that projects high-dimensional gradients into compact binary representations via multi-hyperplane locally-sensitive hashing. This enables accurate detection and filtering of malicious gradients using only their irreversible hash forms, thus mitigating privacy leakage risks and substantially reducing transmission overhead. Extensive experiments demonstrate that LSHFed maintains high model performance even when up to 50% of participants are collusive adversaries while achieving up to a 1000x reduction in gradient verification communication compared to full-gradient methods.
Authors: \'Alvaro V\'azquez Rodr\'iguez, Manuel Fern\'andez-Veiga, Carlos Giraldo-Rodr\'iguez
Abstract: The placement of Cloud-Native Network Functions (CNFs) across the Cloud-Continuum represents a core challenge in the orchestration of current 5G and future 6G networks. The process involves the placement of interdependent computing tasks, structured as Service Function Chains, over distributed cloud infrastructures. This is achieved while satisfying strict resource, bandwidth and latency constraints. It is acknowledged that classical approaches, including mixed-integer nonlinear programming, heuristics and reinforcement learning are limited in terms of scalability, constraint handling and generalisation capacity. In the present study, a novel theoretical framework is proposed, which is based on Denoising Diffusion Probabilistic Models (DDPM) for CNF placement. The present approach proposes a reconceptualisation of placement as a generative graph to assignment task, where the placement problem is encoded as a heterogeneous graph, and a Graph Neural Network denoiser is trained to iteratively refine noisy CNF-to-cloud assignment matrices. The model incorporates constraint-specific losses directly into the loss function, thereby allowing it to learn feasible solution spaces. The integration of the DDPM formulation with structured combinatorial constraints is achieved through a rigorous and systematic approach. Extensive evaluations across diverse topologies have been conducted, which have confirmed that the model consistently produces feasible solutions with orders of magnitude faster inference than MINLP solvers. The results obtained demonstrate the potential of diffusion-based generative modelling for constrained network embedding problems, making an impact towards the practical, scalable orchestration of distributed Cloud-Native Network Functions.
Authors: Lucie Flek, Oliver Janik, Philipp Alexander Jung, Akbar Karimi, Timo Saala, Alexander Schmidt, Matthias Schott, Philipp Soldin, Matthias Thiesmeyer, Christopher Wiebusch, Ulrich Willemsen
Abstract: In this paper, we present a new algorithm, MiniFool, that implements physics-inspired adversarial attacks for testing neural network-based classification tasks in particle and astroparticle physics. While we initially developed the algorithm for the search for astrophysical tau neutrinos with the IceCube Neutrino Observatory, we apply it to further data from other science domains, thus demonstrating its general applicability. Here, we apply the algorithm to the well-known MNIST data set and furthermore, to Open Data data from the CMS experiment at the Large Hadron Collider. The algorithm is based on minimizing a cost function that combines a $\chi^2$ based test-statistic with the deviation from the desired target score. The test statistic quantifies the probability of the perturbations applied to the data based on the experimental uncertainties. For our studied use cases, we find that the likelihood of a flipped classification differs for both the initially correctly and incorrectly classified events. When testing changes of the classifications as a function of an attack parameter that scales the experimental uncertainties, the robustness of the network decision can be quantified. Furthermore, this allows testing the robustness of the classification of unlabeled experimental data.
Authors: Rana Alaa, Dar\'io Gonz\'alez-Ferreiro, Carlos Beis-Penedo, Manuel Fern\'andez-Veiga, Rebeca P. D\'iaz-Redondo, Ana Fern\'andez-Vilas
Abstract: Split learning is an approach to collaborative learning in which a deep neural network is divided into two parts: client-side and server-side at a cut layer. The client side executes its model using its raw input data and sends the intermediate activation to the server side. This configuration architecture is very useful for enabling collaborative training when data or resources are separated between devices. However, split learning lacks the ability to verify the correctness and honesty of the computations that are performed and exchanged between the parties. To this purpose, this paper proposes a verifiable split learning framework that integrates a zk-SNARK proof to ensure correctness and verifiability. The zk-SNARK proof and verification are generated for both sides in forward propagation and backward propagation on the server side, guaranteeing verifiability on both sides. The verifiable split learning architecture is compared to a blockchain-enabled system for the same deep learning network, one that records updates but without generating the zero-knowledge proof. From the comparison, it can be deduced that applying the zk-SNARK test achieves verifiability and correctness, while blockchains are lightweight but unverifiable.
Authors: Ziqi Wang, Jiashun Liu, Ling Pan
Abstract: Traditional continuous deep reinforcement learning (RL) algorithms employ deterministic or unimodal Gaussian actors, which cannot express complex multimodal decision distributions. This limitation can hinder their performance in diversity-critical scenarios. There have been some attempts to design online multimodal RL algorithms based on diffusion or amortized actors. However, these actors are intractable, making existing methods struggle with balancing performance, decision diversity, and efficiency simultaneously. To overcome this challenge, we first reformulate existing intractable multimodal actors within a unified framework, and prove that they can be directly optimized by policy gradient via reparameterization. Then, we propose a distance-based diversity regularization that does not explicitly require decision probabilities. We identify two diversity-critical domains, namely multi-goal achieving and generative RL, to demonstrate the advantages of multimodal policies and our method, particularly in terms of few-shot robustness. In conventional MuJoCo benchmarks, our algorithm also shows competitive performance. Moreover, our experiments highlight that the amortized actor is a promising policy model class with strong multimodal expressivity and high performance. Our code is available at https://github.com/PneuC/DrAC
Authors: Amir Hossein Khorasani, Ali Jahanian, Maryam Rastgarpour
Abstract: Machine learning is a powerful tool for building predictive models. However, it is vulnerable to adversarial attacks. Fast Gradient Sign Method (FGSM) attacks are a common type of adversarial attack that adds small perturbations to input data to trick a model into misclassifying it. In response to these attacks, researchers have developed methods for "unlearning" these attacks, which involves retraining a model on the original data without the added perturbations. Machine unlearning is a technique that tries to "forget" specific data points from the training dataset, to improve the robustness of a machine learning model against adversarial attacks like FGSM. In this paper, we focus on applying unlearning techniques to the LeNet neural network, a popular architecture for image classification. We evaluate the efficacy of unlearning FGSM attacks on the LeNet network and find that it can significantly improve its robustness against these types of attacks.
Authors: Xinyu Ding, Bangtian Liu, Siyu Liao, Zhongfeng Wang
Abstract: Fast Fourier Transforms (FFT) are widely used to reduce memory and computational costs in deep learning. However, existing implementations, including standard FFT and real FFT (rFFT), cannot achieve true in-place computation. In particular, rFFT maps an input of size n to a complex output of size n/2+1, causing dimensional mismatch and requiring additional memory allocation. We propose the first real-domain, fully in-place FFT framework (rdFFT) that preserves input-output memory space consistency. By leveraging butterfly operation symmetry and conjugate properties in the frequency domain, we design an implicit complex encoding scheme that eliminates intermediate cache usage entirely. Experiments on multiple natural language understanding tasks demonstrate the method effectiveness in reducing training memory cost, offering a promising direction for frequency-domain lightweight adaptation.
Authors: Markus B. Pettersson, Adel Daoud
Abstract: Accurate, fine-grained poverty maps remain scarce across much of the Global South. While Demographic and Health Surveys (DHS) provide high-quality socioeconomic data, their spatial coverage is limited and reported coordinates are randomly displaced for privacy, further reducing their quality. We propose a graph-based approach leveraging low-dimensional AlphaEarth satellite embeddings to predict cluster-level wealth indices across Sub-Saharan Africa. By modeling spatial relations between surveyed and unlabeled locations, and by introducing a probabilistic "fuzzy label" loss to account for coordinate displacement, we improve the generalization of wealth predictions beyond existing surveys. Our experiments on 37 DHS datasets (2017-2023) show that incorporating graph structure slightly improves accuracy compared to "image-only" baselines, demonstrating the potential of compact EO embeddings for large-scale socioeconomic mapping.
Authors: Seunghun Yu, Youngjoon Lee, Jinu Gong, Joonhyuk Kang
Abstract: Federated learning (FL), widely used in privacy-critical applications, suffers from limited interpretability, whereas Kolmogorov-Arnold Networks (KAN) address this limitation via learnable spline functions. However, existing FL studies applying KAN overlook the communication overhead introduced by grid extension, which is essential for modeling complex functions. In this letter, we propose CG-FKAN, which compresses extended grids by sparsifying and transmitting only essential coefficients under a communication budget. Experiments show that CG-FKAN achieves up to 13.6% lower RMSE than fixed-grid KAN in communication-constrained settings. In addition, we derive a theoretical upper bound on its approximation error.
Authors: Jacob Poschl
Abstract: Curvature influences generalization, robustness, and how reliably neural networks respond to small input perturbations. Existing sharpness metrics are typically defined in parameter space (e.g., Hessian eigenvalues) and can be expensive, sensitive to reparameterization, and difficult to interpret in functional terms. We introduce a scalar curvature measure defined directly in input space: the curvature rate {\lambda}, given by the exponential growth rate of higher-order input derivatives. Empirically, {\lambda} is estimated as the slope of log ||D^n f|| versus n for small n. This growth-rate perspective unifies classical analytic quantities: for analytic functions, {\lambda} corresponds to the inverse radius of convergence, and for bandlimited signals, it reflects the spectral cutoff. The same principle extends to neural networks, where {\lambda} tracks the emergence of high-frequency structure in the decision boundary. Experiments on analytic functions and neural networks (Two Moons and MNIST) show that {\lambda} evolves predictably during training and can be directly shaped using a simple derivative-based regularizer, Curvature Rate Regularization (CRR). Compared to Sharpness-Aware Minimization (SAM), CRR achieves similar accuracy while yielding flatter input-space geometry and improved confidence calibration. By grounding curvature in differentiation dynamics, {\lambda} provides a compact, interpretable, and parameterization-invariant descriptor of functional smoothness in learned models.
Authors: Chaoqun Fei, Tinglve Zhou, Tianyong Hao, Yangyang Li
Abstract: Graph curvature provides geometric priors for Graph Neural Networks (GNNs), enhancing their ability to model complex graph structures, particularly in terms of structural awareness, robustness, and theoretical interpretability. Among existing methods, Ollivier-Ricci curvature has been extensively studied due to its strong geometric interpretability, effectively characterizing the local geometric distribution between nodes. However, its prohibitively high computational complexity limits its applicability to large-scale graph datasets. To address this challenge, we propose a novel graph curvature measure--Effective Resistance Curvature--which quantifies the ease of message passing along graph edges using the effective resistance between node pairs, instead of the optimal transport distance. This method significantly outperforms Ollivier-Ricci curvature in computational efficiency while preserving comparable geometric expressiveness. Theoretically, we prove the low computational complexity of effective resistance curvature and establish its substitutability for Ollivier-Ricci curvature. Furthermore, extensive experiments on diverse GNN tasks demonstrate that our method achieves competitive performance with Ollivier-Ricci curvature while drastically reducing computational overhead.
Authors: Hao Wang, Zixuan Weng, Jindong Han, Wei Fan, Hao Liu
Abstract: Data Assimilation is a cornerstone of atmospheric system modeling, tasked with reconstructing system states by integrating sparse, noisy observations with prior estimation. While traditional approaches like variational and ensemble Kalman filtering have proven effective, recent advances in deep learning offer more scalable, efficient, and flexible alternatives better suited for complex, real-world data assimilation involving large-scale and multi-modal observations. However, existing deep learning-based DA research suffers from two critical limitations: (1) reliance on oversimplified scenarios with synthetically perturbed observations, and (2) the absence of standardized benchmarks for fair model comparison. To address these gaps, in this work, we introduce DAMBench, the first large-scale multi-modal benchmark designed to evaluate data-driven DA models under realistic atmospheric conditions. DAMBench integrates high-quality background states from state-of-the-art forecasting systems and real-world multi-modal observations (i.e., real-world weather stations and satellite imagery). All data are resampled to a common grid and temporally aligned to support systematic training, validation, and testing. We provide unified evaluation protocols and benchmark representative data assimilation approaches, including latent generative models and neural process frameworks. Additionally, we propose a lightweight multi-modal plugin to demonstrate how integrating realistic observations can enhance even simple baselines. Through comprehensive experiments, DAMBench establishes a rigorous foundation for future research, promoting reproducibility, fair comparison, and extensibility to real-world multi-modal scenarios. Our dataset and code are publicly available at https://github.com/figerhaowang/DAMBench.
Authors: Elvin Hajizada, Danielle Rager, Timothy Shea, Leobardo Campos-Macias, Andreas Wild, Eyke H\"ullermeier, Yulia Sandamirskaya, Mike Davies
Abstract: AI systems on edge devices face a critical challenge in open-world environments: adapting when data distributions shift and novel classes emerge. While offline training dominates current paradigms, online continual learning (OCL)--where models learn incrementally from non-stationary streams without catastrophic forgetting--remains challenging in power-constrained settings. We present a neuromorphic solution called CLP-SNN: a spiking neural network architecture for Continually Learning Prototypes and its implementation on Intel's Loihi 2 chip. Our approach introduces three innovations: (1) event-driven and spatiotemporally sparse local learning, (2) a self-normalizing three-factor learning rule maintaining weight normalization, and (3) integrated neurogenesis and metaplasticity for capacity expansion and forgetting mitigation. On OpenLORIS few-shot learning experiments, CLP-SNN achieves accuracy competitive with replay methods while being rehearsal-free. CLP-SNN delivers transformative efficiency gains: 70\times faster (0.33ms vs 23.2ms), and 5,600\times more energy efficient (0.05mJ vs 281mJ) than the best alternative OCL on edge GPU. This demonstrates that co-designed brain-inspired algorithms and neuromorphic hardware can break traditional accuracy-efficiency trade-offs for future edge AI systems.
Authors: Xiaosha Xue, Peibo Duan, Zhipeng Liu, Qi Chu, Changsheng Zhang, Bin zhang
Abstract: Accurately predicting stock market movements remains a formidable challenge due to the inherent volatility and complex interdependencies among stocks. Although multi-scale Graph Neural Networks (GNNs) hold potential for modeling these relationships, they frequently neglect two key points: the subtle intra-attribute patterns within each stock affecting inter-stock correlation, and the biased attention to coarse- and fine-grained features during multi-scale sampling. To overcome these challenges, we introduce MS-HGFN (Multi-Scale Hierarchical Graph Fusion Network). The model features a hierarchical GNN module that forms dynamic graphs by learning patterns from intra-attributes and features from inter-attributes over different time scales, thus comprehensively capturing spatio-temporal dependencies. Additionally, a top-down gating approach facilitates the integration of multi-scale spatio-temporal features, preserving critical coarse- and fine-grained features without too much interference. Experiments utilizing real-world datasets from U.S. and Chinese stock markets demonstrate that MS-HGFN outperforms both traditional and advanced models, yielding up to a 1.4% improvement in prediction accuracy and enhanced stability in return simulations. The code is available at https://anonymous.4open.science/r/MS-HGFN.
Authors: Wang Hao, Kuang Zhang, Hou Chengyu, Yuan Zhonghao, Tan Chenxing, Fu Weifeng, Zhu Yangying
Abstract: Time series classification holds broad application value in communications, information countermeasures, finance, and medicine. However, state-of-the-art (SOTA) methods-including HIVE-COTE, Proximity Forest, and TS-CHIEF-exhibit high computational complexity, coupled with lengthy parameter tuning and training cycles. In contrast, lightweight solutions like ROCKET (Random Convolutional Kernel Transform) offer greater efficiency but leave substantial room for improvement in kernel selection and computational overhead. To address these challenges, we propose a feature extraction approach based on Hadamard convolutional transform, utilizing column or row vectors of Hadamard matrices as convolution kernels with extended lengths of varying sizes. This enhancement maintains full compatibility with existing methods (e.g., ROCKET) while leveraging kernel orthogonality to boost computational efficiency, robustness, and adaptability. Comprehensive experiments on multi-domain datasets-focusing on the UCR time series dataset-demonstrate SOTA performance: F1-score improved by at least 5% vs. ROCKET, with 50% shorter training time than miniROCKET (fastest ROCKET variant) under identical hyperparameters, enabling deployment on ultra-low-power embedded devices. All code is available on GitHub.
Authors: Zhicheng Wang, Chen Ju, Xu Chen, Shuai Xiao, Jinsong Lan, Xiaoyong Zhu, Ying Chen, Zhiguo Cao
Abstract: Embedding models are a cornerstone of modern AI. Driven by Multimodal Large Language Models (MLLMs), they have made great progress in architecture and data curation, while the holistic paradigm is still limited to SSC, i.e., single input, singular embedding, contrastive supervision, which collapses rich, multifaceted inputs into monolithic embeddings and fails to fully exploit MLLM capabilities. In this paper, we tailor one Parallel Decoupling Framework (PDF) for multimodal embedding learning, by utilizing the proprietary steerability of MLLMs, i.e., their ability to flexibly generate quite differentiated response under explicit instructions. Concretely, PDF conditions a shared MLLM backbone on distinct, learnable prefixes to roll out multiple parallel paths for one input, then relies on these paths to obtain parallel embeddings. To promote full parallel diversity, we employ Mutual Information Minimization (MIM) as an explicit constraint, coupled with per-path contrastive supervision to maintain semantic alignment. Such dual-objectives force PDF to yield robust semantic coverage and a generalizable embedding space. Ultimately, the remarkable embedding space are accessible at inference via one single forward pass, incurring negligible computational overhead. We instantiate PDF on multiple MLLM backbones and prove its effectiveness on MMEB benchmark. Significant gains are consistently achieved across various resolutions and model sizes, e.g., boosting the VLM2Vec-LLaVA-1.6-LR model by a remarkable +8.9% (7B), while the VLM2Vec-Qwen2VL models by +4.2% (2B) and +3.1% (7B). In terms of efficiency, our 2B model surpasses its baseline by +2.6% using only half the computational budget.
Authors: Nat\'alia Ribeiro Marinho, Richard Loendersloot, Frank Grooteman, Jan Willem Wiegman, Uraz Odyurt, Tiedo Tinga
Abstract: Energy estimation is critical to impact identification on aerospace composites, where low-velocity impacts can induce internal damage that is undetectable at the surface. Current methodologies for energy prediction are often constrained by data sparsity, signal noise, complex feature interdependencies, non-linear dynamics, massive design spaces, and the ill-posed nature of the inverse problem. This study introduces a physics-informed framework that embeds domain knowledge into machine learning through a dedicated input space. The approach combines observational biases, which guide the design of physics-motivated features, with targeted feature selection to retain only the most informative indicators. Features are extracted from time, frequency, and time-frequency domains to capture complementary aspects of the structural response. A structured feature selection process integrating statistical significance, correlation filtering, dimensionality reduction, and noise robustness ensures physical relevance and interpretability. Exploratory data analysis further reveals domain-specific trends, yielding a reduced feature set that captures essential dynamic phenomena such as amplitude scaling, spectral redistribution, and transient signal behaviour. Together, these steps produce a compact set of energy-sensitive indicators with both statistical robustness and physical significance, resulting in impact energy predictions that remain interpretable and traceable to measurable structural responses. Using this optimised input space, a fully-connected neural network is trained and validated with experimental data from multiple impact scenarios, including pristine and damaged states. The resulting model demonstrates significantly improved impact energy prediction accuracy, reducing errors by a factor of three compared to conventional time-series techniques and purely data-driven models.
Authors: Daniel Busbib, Ami Wiesel
Abstract: We consider covariance estimation under Toeplitz structure. Numerous sophisticated optimization methods have been developed to maximize the Gaussian log-likelihood under Toeplitz constraints. In contrast, recent advances in deep learning demonstrate the surprising power of simple gradient descent (GD) applied to overparameterized models. Motivated by this trend, we revisit Toeplitz covariance estimation through the lens of overparameterized GD. We model the $P\times P$ covariance as a sum of $K$ complex sinusoids with learnable parameters and optimize them via GD. We show that when $K = P$, GD may converge to suboptimal solutions. However, mild overparameterization ($K = 2P$ or $4P$) consistently enables global convergence from random initializations. We further propose an accelerated GD variant with separate learning rates for amplitudes and frequencies. When frequencies are fixed and only amplitudes are optimized, we prove that the optimization landscape is asymptotically benign and any stationary point recovers the true covariance. Finally, numerical experiments demonstrate that overparameterized GD can match or exceed the accuracy of state-of-the-art methods in challenging settings, while remaining simple and scalable.
Authors: Chengying Huan, Ziheng Meng, Yongchao Liu, Zhengyi Yang, Yun Zhu, Yue Yun, Shipeng Li, Rong Gu, Xiabao Wu, Haitao Zhang, Chuntao Hong, Shaonan Ma, Guihai Chen, Chen Tian
Abstract: Graph Chain-of-Thought (Graph-CoT) enables large language models (LLMs) to perform step-by-step reasoning over graph-structured knowledge, but existing pipelines suffer from low accuracy, excessive token usage, high latency, and low throughput due to single-agent monolithic prompts, repeated context re-encoding, and inefficient serving execution. We present GLM, the first multi-agent Graph-CoT system co-designed with an optimized LLM serving architecture. GLM decomposes reasoning into specialized agents for classification, reasoning, action generation, and graph retrieval, enabling branching and selective context sharing to reduce prompt length and reasoning iterations while preserving reasoning quality, thereby improving accuracy and reducing overall token consumption. To scale inference, we introduce a Graph-CoT-aware LLM inference mechanism with graph-specific KV-cache management, priority-based eviction, and pipelined execution to improve serving efficiency. Experiments demonstrate that GLM improves answer accuracy by up to 38%, reduces token cost by up to 95.7%, lowers inference latency by 90.3%, and achieves up to 15.1x higher throughput compared to state-of-the-art Graph-CoT baselines, enabling efficient adoption for complex real-world reasoning at scale.
Authors: Xiaopeng Ke, Yihan Yu, Ruyue Zhang, Zhishuo Zhou, Fangzhou Shi, Chang Men, Zhengdan Zhu
Abstract: Counterfactual causal inference faces significant challenges when extended to multi-category, multi-valued treatments, where complex cross-effects between heterogeneous interventions are difficult to model. Existing methodologies remain constrained to binary or single-type treatments and suffer from restrictive assumptions, limited scalability, and inadequate evaluation frameworks for complex intervention scenarios. We present XTNet, a novel network architecture for multi-category, multi-valued treatment effect estimation. Our approach introduces a cross-effect estimation module with dynamic masking mechanisms to capture treatment interactions without restrictive structural assumptions. The architecture employs a decomposition strategy separating basic effects from cross-treatment interactions, enabling efficient modeling of combinatorial treatment spaces. We also propose MCMV-AUCC, a suitable evaluation metric that accounts for treatment costs and interaction effects. Extensive experiments on synthetic and real-world datasets demonstrate that XTNet consistently outperforms state-of-the-art baselines in both ranking accuracy and effect estimation quality. The results of the real-world A/B test further confirm its effectiveness.
Authors: Hossein Abdi, Mingfei Sun, Wei Pan
Abstract: Vision-language pre-trained models, such as CLIP, have established new benchmarks in multimodal data mining. In such models, few-shot fine-tuning is a major challenge to achieve optimal performance on both in-distribution (ID) and out-of-distribution (OOD) datasets, especially when labeled data is scarce. Most existing fine-tuning approaches rely on first-order gradient-based optimizers, which typically suffer from slow convergence, sensitivity to step-size hyperparameters, and poor generalization in OOD settings. In contrast, second-order methods utilize local curvature information of the loss landscape to adjust the update step size. This is particularly beneficial for CLIP models, whose non-convex loss functions often contain sharp critical points. In such cases, natural gradient direction can offer more substantial and efficient per-iteration updates when fine-tuning with limited data. Natural Gradient Descent (NGD) is obtained by preconditioning the standard gradient with the inverse Fisher Information Matrix (FIM), which is computationally expensive for large models. To address this, we propose a Bayesian approximation of NGD using a Kalman filter for CLIP models. Our method combines the benefits of second-order optimization with Bayesian inference, which enhances generalization while providing uncertainty quantification. Extensive experiments conducted on diverse image classification datasets demonstrate that our algorithm consistently achieves superior--or comparable--ID performance and improved OOD robustness compared to state-of-the-art baselines. To the best of our knowledge, this work represents the first successful application of Kalman filtering to fine-tuning CLIP-based models, which enables more robust and efficient learning in vision-language tasks.
Authors: Jungyeon Koh, Hyun Jong Yang
Abstract: The growing demand for on-device large language model (LLM) inference highlights the need for efficient mobile edge computing (MEC) solutions, especially in resource-constrained settings. Speculative decoding offers a promising solution by partitioning token generation between a lightweight draft model on mobile devices and a powerful target model on edge servers, but suffers from communication overhead and asynchronous delays. This paper is the first to propose a unified framework that jointly optimizes user association and resource allocation (UARA) to support efficient parallel speculative decoding. We solve the UARA problem using a multi-agent deep reinforcement learning algorithm. To evaluate our approach under realistic conditions, we conduct experiments using the Sionna simulator. Results show that our method achieves up to 28.0% and an average of 23.7% reduction in end-to-end latency without compromising inference accuracy, enabling scalable and low-latency LLM services in MEC systems.
Authors: Obaidullah Zaland, Feras M. Awaysheh, Sawsan Al Zubi, Abdul Rahman Safi, Monowar Bhuyan
Abstract: Federated learning (FL) has emerged as a transformative paradigm for edge intelligence, enabling collaborative model training while preserving data privacy across distributed personal devices. However, the inherent volatility of edge environments, characterized by dynamic resource availability and heterogeneous client capabilities, poses significant challenges for achieving high accuracy and fairness in client participation. This paper investigates the fundamental trade-off between model accuracy and fairness in highly volatile edge environments. This paper provides an extensive empirical evaluation of fairness-based client selection algorithms such as RBFF and RBCSF against random and greedy client selection regarding fairness, model performance, and time, in three benchmarking datasets (CIFAR10, FashionMNIST, and EMNIST). This work aims to shed light on the fairness-performance and fairness-speed trade-offs in a volatile edge environment and explore potential future research opportunities to address existing pitfalls in \textit{fair client selection} strategies in FL. Our results indicate that more equitable client selection algorithms, while providing a marginally better opportunity among clients, can result in slower global training in volatile environments\footnote{The code for our experiments can be found at https://github.com/obaidullahzaland/FairFL_FLTA.
Authors: Dmitrij Schlesinger, Boris Flach
Abstract: One of the main challenges in distributed learning arises from the difficulty of handling heterogeneous local models and data. In light of the recent success of generative models, we propose to meet this challenge by building on the idea of exchanging synthetic data instead of sharing model parameters. Local models can then be treated as ``black boxes'' with the ability to learn their parameters from data and to generate data according to these parameters. Moreover, if the local models admit semi-supervised learning, we can extend the approach by enabling local models on different probability spaces. This allows to handle heterogeneous data with different modalities. We formulate the learning of the local models as a cooperative game starting from the principles of game theory. We prove the existence of a unique Nash equilibrium for exponential family local models and show that the proposed learning approach converges to this equilibrium. We demonstrate the advantages of our approach on standard benchmark vision datasets for image classification and conditional generation.
Authors: Ameya S. Bhave, Navnil Choudhury, Kanad Basu
Abstract: Quantum computing requires effective error correction strategies to mitigate noise and decoherence. Quantum Low-Density Parity-Check (QLDPC) codes have emerged as a promising solution for scalable Quantum Error Correction (QEC) applications by supporting constant-rate encoding and a sparse parity-check structure. However, decoding QLDPC codes via traditional approaches such as Belief Propagation (BP) suffers from poor convergence in the presence of short cycles. Machine learning techniques like Graph Neural Networks (GNNs) utilize learned message passing over their node features; however, they are restricted to pairwise interactions on Tanner graphs, which limits their ability to capture higher-order correlations. In this work, we propose HyperNQ, the first Hypergraph Neural Network (HGNN)- based QLDPC decoder that captures higher-order stabilizer constraints by utilizing hyperedges-thus enabling highly expressive and compact decoding. We use a two-stage message passing scheme and evaluate the decoder over the pseudo-threshold region. Below the pseudo-threshold mark, HyperNQ improves the Logical Error Rate (LER) up to 84% over BP and 50% over GNN-based strategies, demonstrating enhanced performance over the existing state-of-the-art decoders.
Authors: Song Gao, Shusen Jing, Shuai Zhang, Yue Wang, Xiangwei Zhou, Songyang Zhang
Abstract: Recent advancements in large artificial intelligence models (LAMs) are driving significant innovations in mobile edge computing within next-generation wireless networks. However, the substantial demands for computational resources and large-scale training data required to train LAMs conflict with the limited storage and computational capacity of edge devices, posing significant challenges to training and deploying LAMs at the edge. In this work, we introduce the Networked Mixture-of-Experts (NMoE) system, in which clients infer collaboratively by distributing tasks to suitable neighbors based on their expertise and aggregate the returned results. For training the NMoE, we propose a federated learning framework that integrates both supervised and self-supervised learning to balance personalization and generalization, while preserving communication efficiency and data privacy. We conduct extensive experiments to demonstrate the efficacy of the proposed NMoE system, providing insights and benchmarks for the NMoE training algorithms.
Authors: Mei-Chin Pang, Suraj Adhikari, Takuma Kasahara, Nagihiro Haba, Saneyuki Ohno
Abstract: Battery safety is critical in applications ranging from consumer electronics to electric vehicles and aircraft, where undetected anomalies could trigger safety hazards or costly downtime. In this study, we present OSBAD as an open-source benchmark for anomaly detection frameworks in battery applications. By benchmarking 15 diverse algorithms encompassing statistical, distance-based, and unsupervised machine-learning methods, OSBAD enables a systematic comparison of anomaly detection methods across heterogeneous datasets. In addition, we demonstrate how a physics- and statistics-informed feature transformation workflow enhances anomaly separability by decomposing collective anomalies into point anomalies. To address a major bottleneck in unsupervised anomaly detection due to incomplete labels, we propose a Bayesian optimization pipeline that facilitates automated hyperparameter tuning based on transfer-learning and regression proxies. Through validation on datasets covering both liquid and solid-state chemistries, we further demonstrate the cross-chemistry generalization capability of OSBAD to identify irregularities across different electrochemical systems. By making benchmarking database with open-source reproducible anomaly detection workflows available to the community, OSBAD establishes a unified foundation for developing safe, scalable, and transferable anomaly detection tools in battery analytics. This research underscores the significance of physics- and statistics-informed feature engineering as well as model selection with probabilistic hyperparameter tuning, in advancing trustworthy, data-driven diagnostics for safety-critical energy systems.
Authors: Mian Wu, Gavin Zhang, Sewon Min, Sergey Levine, Aviral Kumar
Abstract: Open-ended generation tasks require outputs to satisfy diverse and often implicit task-specific evaluation rubrics. The sheer number of relevant rubrics leads to prohibitively high verification costs and incomplete assessments of a response, making reinforcement learning (RL) post-training with rubric-based rewards difficult to scale. This problem is exacerbated by the fact that often the best way to combine these rubrics into one single reward is also highly prompt-specific. We propose Reinforcement Learning with Adversarial Critic (RLAC), a post-training approach that addresses these challenges via dynamic rubric verification. Our approach employs a large language model (LLM) as a critic that dynamically identifies only the most likely failure modes (e.g., a factual error or unhandled edge case), which are then verified by an external validator to optimize both generator and critic jointly. By training both the generator and the critic, this game enhances the critic's error detection and the generator's output quality while reducing required verifications. Our experiments demonstrate that RLAC improves factual accuracy in text generation and correctness in code generation, while also outperforming exhaustive verification and reward model methods. We show that dynamic critics are more effective than fixed critics, showcasing the potential of RLAC for scaling RL post-training to free-form generation tasks.
Authors: Vi Retault, Yoha\"i-Eliel Berreby
Abstract: When fine-tuning language models on new tasks, catastrophic forgetting -- performance degradation on previously-learned tasks -- is a ubiquitous problem. While Parameter-Efficient Fine-Tuning (PEFT) methods like LoRA address this through low-rank adapters, sparse adaptation offers an alternative that doesn't impose rank constraints. We introduce Random Initialization of Gated Sparse Adapters (RIGSA), which starts from randomly-initialized full-rank adapters, gates them with a ReZero analog, and sparsifies them with iterative magnitude pruning. We evaluate RIGSA on SmolLM2-1.7B-Instruct using a novel vision-in-text task (Textual MNIST) and measure forgetting on PIQA, HellaSwag, and GSM8k. SmolLM2-1.7B-Instruct initially performs around chance level on Textual MNIST, and is capable of learning the task through RIGSA, 4-bit QLoRA and random masking. In spite of having more trainable parameters than QLoRA, the RIGSA configurations that we studied displayed less forgetting than QLoRA, particularly on GSM8k, though it performs comparably to random masking.
Authors: Gabriel Nobis, Maximilian Springenberg, Arina Belova, Rembert Daems, Christoph Knochenhauer, Manfred Opper, Tolga Birdal, Wojciech Samek
Abstract: We present Fractional Diffusion Bridge Models (FDBM), a novel generative diffusion bridge framework driven by an approximation of the rich and non-Markovian fractional Brownian motion (fBM). Real stochastic processes exhibit a degree of memory effects (correlations in time), long-range dependencies, roughness and anomalous diffusion phenomena that are not captured in standard diffusion or bridge modeling due to the use of Brownian motion (BM). As a remedy, leveraging a recent Markovian approximation of fBM (MA-fBM), we construct FDBM that enable tractable inference while preserving the non-Markovian nature of fBM. We prove the existence of a coupling-preserving generative diffusion bridge and leverage it for future state prediction from paired training data. We then extend our formulation to the Schr\"{o}dinger bridge problem and derive a principled loss function to learn the unpaired data translation. We evaluate FDBM on both tasks: predicting future protein conformations from aligned data, and unpaired image translation. In both settings, FDBM achieves superior performance compared to the Brownian baselines, yielding lower root mean squared deviation (RMSD) of C$_\alpha$ atomic positions in protein structure prediction and lower Fr\'echet Inception Distance (FID) in unpaired image translation.
Authors: Prateek Chanda, Shrey Modi, Ganesh Ramakrishnan
Abstract: In a distributed machine learning setting like Federated Learning where there are multiple clients involved which update their individual weights to a single central server, often training on the entire individual client's dataset for each client becomes cumbersome. To address this issue we propose $\methodprop$: a personalized coreset weighted federated learning setup where the training updates for each individual clients are forwarded to the central server based on only individual client coreset based representative data points instead of the entire client data. Through theoretical analysis we present how the average generalization error is minimax optimal up to logarithm bounds (upper bounded by $\mathcal{O}(n_k^{-\frac{2 \beta}{2 \beta+\boldsymbol{\Lambda}}} \log ^{2 \delta^{\prime}}(n_k))$) and lower bounds of $\mathcal{O}(n_k^{-\frac{2 \beta}{2 \beta+\boldsymbol{\Lambda}}})$, and how the overall generalization error on the data likelihood differs from a vanilla Federated Learning setup as a closed form function ${\boldsymbol{\Im}}(\boldsymbol{w}, n_k)$ of the coreset weights $\boldsymbol{w}$ and coreset sample size $n_k$. Our experiments on different benchmark datasets based on a variety of recent personalized federated learning architectures show significant gains as compared to random sampling on the training data followed by federated learning, thereby indicating how intelligently selecting such training samples can help in performance. Additionally, through experiments on medical datasets our proposed method showcases some gains as compared to other submodular optimization based approaches used for subset selection on client's data.
Authors: Viraj Patel, Lisa Kreusser, Katharine Fraser
Abstract: Blood flow is sensitive to disease and provides insight into cardiac function, making flow field analysis valuable for diagnosis. However, while safer than radiation-based imaging and more suitable for patients with medical implants, ultrasound suffers from attenuation with depth, limiting the quality of the image. Despite advances in echocardiographic particle image velocimetry (EchoPIV), accurately measuring blood velocity remains challenging due to the technique's limitations and the complexity of blood flow dynamics. Physics-informed machine learning can enhance accuracy and robustness, particularly in scenarios where noisy or incomplete data challenge purely data-driven approaches. We present a physics-informed neural field model with multi-scale Fourier Feature encoding for estimating blood flow from sparse and noisy ultrasound data without requiring ground truth supervision. We demonstrate that this model achieves consistently low mean squared error in denoising and inpainting both synthetic and real datasets, verified against reference flow fields and ground truth flow rate measurements. While physics-informed neural fields have been widely used to reconstruct medical images, applications to medical flow reconstruction are mostly prominent in Flow MRI. In this work, we adapt methods that have proven effective in other imaging modalities to address the specific challenge of ultrasound-based flow reconstruction.
Authors: Maryam Bagherian
Abstract: Tensor decomposition faces fundamental challenges in analyzing high-dimensional data, where traditional methods based on reconstruction and fixed-rank constraints often fail to capture semantically meaningful structures. This paper introduces a no-rank tensor decomposition framework grounded in metric learning, which replaces reconstruction objectives with a discriminative, similarity-based optimization. The proposed approach learns data-driven embeddings by optimizing a triplet loss with diversity and uniformity regularization, creating a feature space where distance directly reflects semantic similarity. We provide theoretical guarantees for the framework's convergence and establish bounds on its metric properties. Evaluations across diverse domains --including face recognition (LFW, Olivetti), brain connectivity analysis (ABIDE), and simulated data (galaxy morphology, crystal structures)-- demonstrate that our method outperforms baseline techniques, including PCA, t-SNE, UMAP, and tensor decomposition baselines (CP and Tucker). Results show substantial improvements in clustering metrics (Silhouette Score, Davies--Bouldin Index, Calinski--Harabasz Index, Separation Ratio, Adjusted Rand Index, Normalized Mutual Information) and reveal a fundamental trade-off: while metric learning optimizes global class separation, it deliberately transforms local geometry to align with semantic relationships. Crucially, our approach achieves superior performance with smaller training datasets compared to transformer-based methods, offering an efficient alternative for domains with limited labeled data. This work establishes metric learning as a paradigm for tensor-based analysis, prioritizing semantic relevance over pixel-level fidelity while providing computational advantages in data-scarce scenarios.
Authors: Hamed Fard, Mahsa Kholghi, Benedikt Gro{\ss}, Gerhard Wunder
Abstract: Ultra-wideband (UWB) localization delivers centimeter-scale accuracy but is vulnerable to jamming attacks, creating security risks for asset tracking and intrusion detection in smart buildings. Although machine learning (ML) and deep learning (DL) methods have improved tag localization, localizing malicious jammers within a single room and across changing indoor layouts remains largely unexplored. Two novel UWB datasets, collected under original and modified room configurations, are introduced to establish comprehensive ML/DL baselines. Performance is rigorously evaluated using a variety of classification and regression metrics. On the source dataset with the collected UWB features, Random Forest achieves the highest F1-macro score of 0.95 and XGBoost achieves the lowest mean Euclidean error of 20.16 cm. However, deploying these source-trained models in the modified room layout led to severe performance degradation, with XGBoost's mean Euclidean error increasing tenfold to 207.99 cm, demonstrating significant domain shift. To mitigate this degradation, a domain-adversarial ConvNeXt autoencoder (A-CNT) is proposed that leverages a gradient-reversal layer to align CIR-derived features across domains. The A-CNT framework restores localization performance by reducing the mean Euclidean error to 34.67 cm. This represents a 77 percent improvement over non-adversarial transfer learning and an 83 percent improvement over the best baseline, restoring the fraction of samples within 30 cm to 0.56. Overall, the results demonstrate that adversarial feature alignment enables robust and transferable indoor jammer localization despite environmental changes. Code and dataset available at https://github.com/afbf4c8996f/Jammer-Loc
Authors: Paul Setinek, Gianluca Galletti, Johannes Brandstetter
Abstract: Scaling laws describe how model performance grows with data, parameters and compute. While large datasets can usually be collected at relatively low cost in domains such as language or vision, scientific machine learning is often limited by the high expense of generating training data through numerical simulations. However, by adjusting modeling assumptions and approximations, simulation fidelity can be traded for computational cost, an aspect absent in other domains. We investigate this trade-off between data fidelity and cost in neural surrogates using low- and high-fidelity Reynolds-Averaged Navier-Stokes (RANS) simulations. Reformulating classical scaling laws, we decompose the dataset axis into compute budget and dataset composition. Our experiments reveal compute-performance scaling behavior and exhibit budget-dependent optimal fidelity mixes for the given dataset configuration. These findings provide the first study of empirical scaling laws for multi-fidelity neural surrogate datasets and offer practical considerations for compute-efficient dataset generation in scientific machine learning.
Authors: Jay Mohta, Kenan Emir Ak, Dimitrios Dimitriadis, Yan Xu, Mingwei Shen
Abstract: Vision-Language Models (VLMs) suffer from catastrophic forgetting when sequentially fine-tuned on new tasks, degrading performance on previously learned foundational and task-specific capabilities. While multi-task learning can mitigate forgetting, it requires simultaneous access to all datasets and imposes computational overhead that scales linearly with the number of tasks. In this work, we introduce a routing-based approach that enables the integration of new tasks while preserving the foundational knowledge acquired during pretraining. We evaluate our method using InternVL-2 models (2B and 8B parameters) and demonstrate that routing preserves the model's foundational capabilities by maintaining performance on general-purpose benchmarks such as ChartQA, MMBench, and DocVQA, while simultaneously improving accuracy on specialized tasks. Importantly, our approach achieves this without requiring concurrent access to data from all tasks, avoiding the significant computational and data overhead associated with traditional multi-task learning. We further conduct extensive ablation studies to evaluate the scalability and robustness of routing-based learning, showing that the approach is resilient to a growing number of tasks and performs particularly well when new tasks are semantically related. Finally, we show that the routing mechanism enables superior cross-modal transfer between language and vision capabilities, allowing knowledge learned in one modality to enhance performance in another capability not achieved by existing continual learning methods.
Authors: Ekdeep Singh Lubana, Can Rager, Sai Sumedh R. Hindupur, Valerie Costa, Greta Tuckute, Oam Patel, Sonia Krishna Murthy, Thomas Fel, Daniel Wurgaft, Eric J. Bigelow, Johnny Lin, Demba Ba, Martin Wattenberg, Fernanda Viegas, Melanie Weber, Aaron Mueller
Abstract: Recovering meaningful concepts from language model activations is a central aim of interpretability. While existing feature extraction methods aim to identify concepts that are independent directions, it is unclear if this assumption can capture the rich temporal structure of language. Specifically, via a Bayesian lens, we demonstrate that Sparse Autoencoders (SAEs) impose priors that assume independence of concepts across time, implying stationarity. Meanwhile, language model representations exhibit rich temporal dynamics, including systematic growth in conceptual dimensionality, context-dependent correlations, and pronounced non-stationarity, in direct conflict with the priors of SAEs. Taking inspiration from computational neuroscience, we introduce a new interpretability objective -- Temporal Feature Analysis -- which possesses a temporal inductive bias to decompose representations at a given time into two parts: a predictable component, which can be inferred from the context, and a residual component, which captures novel information unexplained by the context. Temporal Feature Analyzers correctly parse garden path sentences, identify event boundaries, and more broadly delineate abstract, slow-moving information from novel, fast-moving information, while existing SAEs show significant pitfalls in all the above tasks. Overall, our results underscore the need for inductive biases that match the data in designing robust interpretability tools.
Authors: Isabela Suaza-Sierra, Hernan A. Moreno, Luis A De la Fuente, Thomas M. Neeson
Abstract: Accurate prediction of Reservoir Water Temperature (RWT) is vital for sustainable water management, ecosystem health, and climate resilience. Yet, prediction alone offers limited insight into the governing physical processes. To bridge this gap, we integrated explainable machine learning (ML) with symbolic modeling to uncover the drivers of RWT dynamics across ten reservoirs in the Red River Basin, USA, using over 10,000 depth-resolved temperature profiles. We first employed ensemble and neural models, including Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Multilayer Perceptron (MLP), achieving high predictive skill (best RMSE = 1.20 degree Celsius, R^2 = 0.97). Using SHAP (SHapley Additive exPlanations), we quantified the contribution of physical drivers such as air temperature, depth, wind, and lake volume, revealing consistent patterns across reservoirs. To translate these data-driven insights into compact analytical expressions, we developed Kolmogorov Arnold Networks (KANs) to symbolically approximate RWT. Ten progressively complex KAN equations were derived, improving from R^2 = 0.84 using a single predictor (7-day antecedent air temperature) to R^2 = 0.92 with ten predictors, though gains diminished beyond five, highlighting a balance between simplicity and accuracy. The resulting equations, dominated by linear and rational forms, incrementally captured nonlinear behavior while preserving interpretability. Depth consistently emerged as a secondary but critical predictor, whereas precipitation had limited effect. By coupling predictive accuracy with explanatory power, this framework demonstrates how KANs and explainable ML can transform black-box models into transparent surrogates that advance both prediction and understanding of reservoir thermal dynamics.
Authors: Zhi Wang, Chicheng Zhang, Ramya Korlakai Vinayak
Abstract: In lifelong learning, a learner faces a sequence of tasks with shared structure and aims to identify and leverage it to accelerate learning. We study the setting where such structure is captured by a common representation of data. Unlike multi-task learning or learning-to-learn, where tasks are available upfront to learn the representation, lifelong learning requires the learner to make use of its existing knowledge while continually gathering partial information in an online fashion. In this paper, we consider a generalized framework of lifelong representation learning. We propose a simple algorithm that uses multi-task empirical risk minimization as a subroutine and establish a sample complexity bound based on a new notion we introduce--the task-eluder dimension. Our result applies to a wide range of learning problems involving general function classes. As concrete examples, we instantiate our result on classification and regression tasks under noise.
Authors: Bettina Hanlon, Angel Garcia Fernandez
Abstract: This paper presents a coordinate ascent algorithm to learn dynamic and measurement models in dynamic state estimation using maximum likelihood estimation in a supervised manner. In particular, the dynamic and measurement models are assumed to be Gaussian and the algorithm learns the neural network parameters that model the dynamic and measurement functions, and also the noise covariance matrices. The trained dynamic and measurement models are then used with a non-linear Kalman filter algorithm to estimate the state during the testing phase.
Authors: Octavio Vega, Javad Komijani, Aida El-Khadra, Marina Marinkovic
Abstract: Near the critical point, Markov Chain Monte Carlo (MCMC) simulations of lattice quantum field theories (LQFT) become increasingly inefficient due to critical slowing down. In this work, we investigate score-based symmetry-preserving diffusion models as an alternative strategy to sample two-dimensional $\phi^4$ and ${\rm U}(1)$ lattice field theories. We develop score networks that are equivariant to a range of group transformations, including global $\mathbb{Z}_2$ reflections, local ${\rm U}(1)$ rotations, and periodic translations $\mathbb{T}$. The score networks are trained using an augmented training scheme, which significantly improves sample quality in the simulated field theories. We also demonstrate empirically that our symmetry-aware models outperform generic score networks in sample quality, expressivity, and effective sample size.
Authors: Jinwoo Baek
Abstract: Matrix Phylogeny introduces compact spectral fingerprints (CSF/ASF) that characterize matrices at the family level. These fingerprints are low-dimensional, eigendecomposition-free descriptors built from Chebyshev trace moments estimated by Hutchinson sketches. A simple affine rescaling to [-1,1] makes them permutation/similarity invariant and robust to global scaling. Across synthetic and real tests, we observe phylogenetic compactness: only a few moments are needed. CSF with K=3-5 already yields perfect clustering (ARI=1.0; silhouettes ~0.89) on four synthetic families and a five-family set including BA vs ER, while ASF adapts the dimension on demand (median K*~9). On a SuiteSparse mini-benchmark (Hutchinson p~100), both CSF-H and ASF-H reach ARI=1.0. Against strong alternatives (eigenvalue histograms + Wasserstein, heat-kernel traces, WL-subtree), CSF-K=5 matches or exceeds accuracy while avoiding eigendecompositions and using far fewer features (K<=10 vs 64/9153). The descriptors are stable to noise (log-log slope ~1.03, R^2~0.993) and support a practical trap->recommend pipeline for automated preconditioner selection. In an adversarial E6+ setting with a probe-and-switch mechanism, our physics-guided recommender attains near-oracle iteration counts (p90 regret=0), whereas a Frobenius 1-NN baseline exhibits large spikes (p90~34-60). CSF/ASF deliver compact (K<=10), fast, invariant fingerprints that enable scalable, structure-aware search and recommendation over large matrix repositories. We recommend CSF with K=5 by default, and ASF when domain-specific adaptivity is desired.
Authors: Daria D. Tyurina, Sergey V. Stasenko, Konstantin V. Lushnikov, Maria V. Vedunova
Abstract: This study introduces a novel method for predicting cognitive age using psychophysiological tests. To determine cognitive age, subjects were asked to complete a series of psychological tests measuring various cognitive functions, including reaction time and cognitive conflict, short-term memory, verbal functions, and color and spatial perception. Based on the tests completed, the average completion time, proportion of correct answers, average absolute delta of the color campimetry test, number of guessed words in the M\"unsterberg matrix, and other parameters were calculated for each subject. The obtained characteristics of the subjects were preprocessed and used to train a machine learning algorithm implementing a regression task for predicting a person's cognitive age. These findings contribute to the field of remote screening using mobile devices for human health for diagnosing and monitoring cognitive aging.
Authors: Haotian Hang, Yueyang Shen, Vicky Zhu, Jose Cruz, Michelle Li
Abstract: In the context of global sustainability mandates, corporate carbon disclosure has emerged as a critical mechanism for aligning business strategy with environmental responsibility. The Carbon Disclosure Project (CDP) hosts the world's largest longitudinal dataset of climate-related survey responses, combining structured indicators with open-ended narratives, but the heterogeneity and free-form nature of these disclosures present significant analytical challenges for benchmarking, compliance monitoring, and investment screening. This paper proposes a novel decision-support framework that leverages large language models (LLMs) to assess corporate climate disclosure quality at scale. It develops a master rubric that harmonizes narrative scoring across 11 years of CDP data (2010-2020), enabling cross-sector and cross-country benchmarking. By integrating rubric-guided scoring with percentile-based normalization, our method identifies temporal trends, strategic alignment patterns, and inconsistencies in disclosure across industries and regions. Results reveal that sectors such as technology and countries like Germany consistently demonstrate higher rubric alignment, while others exhibit volatility or superficial engagement, offering insights that inform key decision-making processes for investors, regulators, and corporate environmental, social, and governance (ESG) strategists. The proposed LLM-based approach transforms unstructured disclosures into quantifiable, interpretable, comparable, and actionable intelligence, advancing the capabilities of AI-enabled decision support systems (DSSs) in the domain of climate governance.
Authors: Tadisetty Sai Yashwanth
Abstract: Floating-point non-associativity makes fundamental deep learning operations, such as matrix multiplication (matmul) on GPUs, inherently non-deterministic. Despite this, the statistical structure of the resulting numerical error remains poorly understood. A common working assumption is that these errors behave as independent and identically distributed (i.i.d.) Gaussian noise. In this paper, we empirically test this assumption and show that it fails to describe real GPU behavior. By comparing outputs of single-input and batched matmuls, we find that while the i.i.d. model predicts non-zero output instability, empirical results show a 0.00% prediction flip rate. Through covariance analysis, we uncover the cause: the floating-point error is structured and highly correlated. For float16, nearly 50% of the total error variance lies in off-diagonal terms, revealing that the noise behaves as a coordinated, directional perturbation rather than random static. This result challenges the prevailing stochastic view of numerical noise and provides a principled foundation for analyzing deep learning reliability under hardware non-determinism.
Authors: Josu Eguiluz Casta\~neira, Axel Brando, Migle Laukyte, Marc Serra-Vidal
Abstract: Artificial intelligence (AI) now permeates critical infrastructures and decision-making systems where failures produce social, economic, and democratic harm. This position paper challenges the entrenched belief that regulation and innovation are opposites. As evidenced by analogies from aviation, pharmaceuticals, and welfare systems and recent cases of synthetic misinformation, bias and unaccountable decision-making, the absence of well-designed regulation has already created immeasurable damage. Regulation, when thoughtful and adaptive, is not a brake on innovation--it is its foundation. The present position paper examines the EU AI Act as a model of risk-based, responsibility-driven regulation that addresses the Collingridge Dilemma: acting early enough to prevent harm, yet flexibly enough to sustain innovation. Its adaptive mechanisms--regulatory sandboxes, small and medium enterprises (SMEs) support, real-world testing, fundamental rights impact assessment (FRIA) -- demonstrate how regulation can accelerate responsibly, rather than delay, technological progress. The position paper summarises how governance tools transform perceived burdens into tangible advantages: legal certainty, consumer trust, and ethical competitiveness. Ultimately, the paper reframes progress: innovation and regulation advance together. By embedding transparency, impact assessments, accountability, and AI literacy into design and deployment, the EU framework defines what responsible innovation truly means--technological ambition disciplined by democratic values and fundamental rights.
Authors: Aditya Akella
Abstract: Recent advances in learnable reward shaping have shown promise in single-agent reinforcement learning by automatically discovering effective feedback signals. However, the effectiveness of decentralized learnable reward shaping in cooperative multi-agent settings remains poorly understood. We propose DMARL-RSA, a fully decentralized system where each agent learns individual reward shaping, and evaluate it on cooperative navigation tasks in the simple_spread_v3 environment. Despite sophisticated reward learning, DMARL-RSA achieves only -24.20 +/- 0.09 average reward, compared to MAPPO with centralized training at 1.92 +/- 0.87--a 26.12-point gap. DMARL-RSA performs similarly to simple independent learning (IPPO: -23.19 +/- 0.96), indicating that advanced reward shaping cannot overcome fundamental decentralized coordination limitations. Interestingly, decentralized methods achieve higher landmark coverage (0.888 +/- 0.029 for DMARL-RSA, 0.960 +/- 0.045 for IPPO out of 3 total) but worse overall performance than centralized MAPPO (0.273 +/- 0.008 landmark coverage)--revealing a coordination paradox between local optimization and global performance. Analysis identifies three critical barriers: (1) non-stationarity from concurrent policy updates, (2) exponential credit assignment complexity, and (3) misalignment between individual reward optimization and global objectives. These results establish empirical limits for decentralized reward learning and underscore the necessity of centralized coordination for effective multi-agent cooperation.
Authors: Krishna Kumar Neelakanta Pillai Santha Kumari Amma
Abstract: Dynamic pricing in retail requires policies that adapt to shifting demand while coordinating decisions across related products. We present a systematic empirical study of multi-agent reinforcement learning for retail price optimization, comparing a strong MAPPO baseline with a graph-attention-augmented variant (MAPPO+GAT) that leverages learned interactions among products. Using a simulated pricing environment derived from real transaction data, we evaluate profit, stability across random seeds, fairness across products, and training efficiency under a standardized evaluation protocol. The results indicate that MAPPO provides a robust and reproducible foundation for portfolio-level price control, and that MAPPO+GAT further enhances performance by sharing information over the product graph without inducing excessive price volatility. These results indicate that graph-integrated MARL provides a more scalable and stable solution than independent learners for dynamic retail pricing, offering practical advantages in multi-product decision-making.
Authors: NVIDIA, :, Arslan Ali, Junjie Bai, Maciej Bala, Yogesh Balaji, Aaron Blakeman, Tiffany Cai, Jiaxin Cao, Tianshi Cao, Elizabeth Cha, Yu-Wei Chao, Prithvijit Chattopadhyay, Mike Chen, Yongxin Chen, Yu Chen, Shuai Cheng, Yin Cui, Jenna Diamond, Yifan Ding, Jiaojiao Fan, Linxi Fan, Liang Feng, Francesco Ferroni, Sanja Fidler, Xiao Fu, Ruiyuan Gao, Yunhao Ge, Jinwei Gu, Aryaman Gupta, Siddharth Gururani, Imad El Hanafi, Ali Hassani, Zekun Hao, Jacob Huffman, Joel Jang, Pooya Jannaty, Jan Kautz, Grace Lam, Xuan Li, Zhaoshuo Li, Maosheng Liao, Chen-Hsuan Lin, Tsung-Yi Lin, Yen-Chen Lin, Huan Ling, Ming-Yu Liu, Xian Liu, Yifan Lu, Alice Luo, Qianli Ma, Hanzi Mao, Kaichun Mo, Seungjun Nah, Yashraj Narang, Abhijeet Panaskar, Lindsey Pavao, Trung Pham, Morteza Ramezanali, Fitsum Reda, Scott Reed, Xuanchi Ren, Haonan Shao, Yue Shen, Stella Shi, Shuran Song, Bartosz Stefaniak, Shangkun Sun, Shitao Tang, Sameena Tasmeen, Lyne Tchapmi, Wei-Cheng Tseng, Jibin Varghese, Andrew Z. Wang, Hao Wang, Haoxiang Wang, Heng Wang, Ting-Chun Wang, Fangyin Wei, Jiashu Xu, Dinghao Yang, Xiaodong Yang, Haotian Ye, Seonghyeon Ye, Xiaohui Zeng, Jing Zhang, Qinsheng Zhang, Kaiwen Zheng, Andrew Zhu, Yuke Zhu
Abstract: We introduce [Cosmos-Predict2.5], the latest generation of the Cosmos World Foundation Models for Physical AI. Built on a flow-based architecture, [Cosmos-Predict2.5] unifies Text2World, Image2World, and Video2World generation in a single model and leverages [Cosmos-Reason1], a Physical AI vision-language model, to provide richer text grounding and finer control of world simulation. Trained on 200M curated video clips and refined with reinforcement learning-based post-training, [Cosmos-Predict2.5] achieves substantial improvements over [Cosmos-Predict1] in video quality and instruction alignment, with models released at 2B and 14B scales. These capabilities enable more reliable synthetic data generation, policy evaluation, and closed-loop simulation for robotics and autonomous systems. We further extend the family with [Cosmos-Transfer2.5], a control-net style framework for Sim2Real and Real2Real world translation. Despite being 3.5$\times$ smaller than [Cosmos-Transfer1], it delivers higher fidelity and robust long-horizon video generation. Together, these advances establish [Cosmos-Predict2.5] and [Cosmos-Transfer2.5] as versatile tools for scaling embodied intelligence. To accelerate research and deployment in Physical AI, we release source code, pretrained checkpoints, and curated benchmarks under the NVIDIA Open Model License at https://github.com/nvidia-cosmos/cosmos-predict2.5 and https://github.com/nvidia-cosmos/cosmos-transfer2.5. We hope these open resources lower the barrier to adoption and foster innovation in building the next generation of embodied intelligence.
URLs: https://github.com/nvidia-cosmos/cosmos-predict2.5, https://github.com/nvidia-cosmos/cosmos-transfer2.5.
Authors: Pradeep M, Ritesh Pallod, Satyen Abrol, Muthu Raman, Ian Anderson
Abstract: Generative AI is reshaping fashion by enabling virtual looks and avatars making it essential to find real products that best match AI-generated styles. We propose an end-to-end product search system that has been deployed in a real-world, internet scale which ensures that AI-generated looks presented to users are matched with the most visually and semantically similar products from the indexed vector space. The search pipeline is composed of four key components: query generation, vectorization, candidate retrieval, and reranking based on AI-generated looks. Recommendation quality is evaluated using human-judged accuracy scores. The system currently serves more than 350,000 AI Looks in production per day, covering diverse product categories across global markets of over 12 million products. In our experiments, we observed that across multiple annotators and categories, CLIP outperformed alternative models by a small relative margin of 3--7\% in mean opinion scores. These improvements, though modest in absolute numbers, resulted in noticeably better user perception matches, establishing CLIP as the most reliable backbone for production deployment.
Authors: Qianhui Li, Weiya Wang, Qianqi Zhao, Tong Qu, Jing He, Xuhong Qiang, Jingwen Hou, Ke Chen, Bao Zhang, Qi Wang
Abstract: Quarter level cell (QLC) 3D NAND flash memory is emerging as the predominant storage solution in the era of artificial intelligence. QLC 3D NAND flash stores 4 bit per cell to expand the storage density, resulting in narrower read margins. Constrained to read margins, QLC always suffers from lateral charge migration (LCM), which caused by non-uniform charge density across adjacent memory cells. To suppress charge density gap between cells, there are some algorithm in form of intra-page data mapping such as WBVM, DVDS. However, we observe inter-page data arrangements also approach the suppression. Thus, we proposed an intelligent model PDA-LSTM to arrange intra-page data for LCM suppression, which is a physics-knowledge-driven neural network model. PDA-LSTM applies a long-short term memory (LSTM) neural network to compute a data arrangement probability matrix from input page data pattern. The arrangement is to minimize the global impacts derived from the LCM among wordlines. Since each page data can be arranged only once, we design a transformation from output matrix of LSTM network to non-repetitive sequence generation probability matrix to assist training process. The arranged data pattern can decrease the bit error rate (BER) during data retention. In addition, PDA-LSTM do not need extra flag bits to record data transport of 3D NAND flash compared with WBVM, DVDS. The experiment results show that the PDA-LSTM reduces the average BER by 80.4% compared with strategy without data arrangement, and by 18.4%, 15.2% compared respectively with WBVM and DVDS with code-length 64.
Authors: Masfiqur Rahaman, Maoyejatun Hasana, Shahad Shahriar Rahman, MD Sajid Mostafiz Noor, Razin Reaz Abedin, Md Toki Tahmid, Duncan Watson Parris, Tanzeem Choudhury, A. B. M. Alim Al Islam, Tauhidur Rahman
Abstract: Cycle rickshaw pullers are highly vulnerable to extreme heat, yet little is known about how their physiological biomarkers respond under such conditions. This study collected real-time weather and physiological data using wearable sensors from 100 rickshaw pullers in Dhaka, Bangladesh. In addition, interviews with 12 pullers explored their knowledge, perceptions, and experiences related to climate change. We developed a Linear Gaussian Bayesian Network (LGBN) regression model to predict key physiological biomarkers based on activity, weather, and demographic features. The model achieved normalized mean absolute error values of 0.82, 0.47, 0.65, and 0.67 for skin temperature, relative cardiac cost, skin conductance response, and skin conductance level, respectively. Using projections from 18 CMIP6 climate models, we layered the LGBN on future climate forecasts to analyze survivability for current (2023-2025) and future years (2026-2100). Based on thresholds of WBGT above 31.1{\deg}C and skin temperature above 35{\deg}C, 32% of rickshaw pullers already face high heat exposure risk. By 2026-2030, this percentage may rise to 37% with average exposure lasting nearly 12 minutes, or about two-thirds of the trip duration. A thematic analysis of interviews complements these findings, showing that rickshaw pullers recognize their increasing climate vulnerability and express concern about its effects on health and occupational survivability.
Authors: NVIDIA, :, Yan Wang, Wenjie Luo, Junjie Bai, Yulong Cao, Tong Che, Ke Chen, Yuxiao Chen, Jenna Diamond, Yifan Ding, Wenhao Ding, Liang Feng, Greg Heinrich, Jack Huang, Peter Karkus, Boyi Li, Pinyi Li, Tsung-Yi Lin, Dongran Liu, Ming-Yu Liu, Langechuan Liu, Zhijian Liu, Jason Lu, Yunxiang Mao, Pavlo Molchanov, Lindsey Pavao, Zhenghao Peng, Mike Ranzinger, Ed Schmerling, Shida Shen, Yunfei Shi, Sarah Tariq, Ran Tian, Tilman Wekel, Xinshuo Weng, Tianjun Xiao, Eric Yang, Xiaodong Yang, Yurong You, Xiaohui Zeng, Wenyuan Zhang, Boris Ivanovic, Marco Pavone
Abstract: End-to-end architectures trained via imitation learning have advanced autonomous driving by scaling model size and data, yet performance remains brittle in safety-critical long-tail scenarios where supervision is sparse and causal understanding is limited. To address this, we introduce Alpamayo-R1 (AR1), a vision-language-action model (VLA) that integrates Chain of Causation reasoning with trajectory planning to enhance decision-making in complex driving scenarios. Our approach features three key innovations: (1) the Chain of Causation (CoC) dataset, built through a hybrid auto-labeling and human-in-the-loop pipeline producing decision-grounded, causally linked reasoning traces aligned with driving behaviors; (2) a modular VLA architecture combining Cosmos-Reason, a Vision-Language Model pre-trained for Physical AI applications, with a diffusion-based trajectory decoder that generates dynamically feasible plans in real time; (3) a multi-stage training strategy using supervised fine-tuning to elicit reasoning and reinforcement learning (RL) to optimize reasoning quality via large reasoning model feedback and enforce reasoning-action consistency. Evaluation shows AR1 achieves up to a 12% improvement in planning accuracy on challenging cases compared to a trajectory-only baseline, with a 35% reduction in off-road rate and 25% reduction in close encounter rate in closed-loop simulation. RL post-training improves reasoning quality by 45% as measured by a large reasoning model critic and reasoning-action consistency by 37%. Model scaling from 0.5B to 7B parameters shows consistent improvements. On-vehicle road tests confirm real-time performance (99 ms latency) and successful urban deployment. By bridging interpretable reasoning with precise control, AR1 demonstrates a practical path towards Level 4 autonomous driving. We plan to release AR1 models and a subset of the CoC in a future update.
Authors: Shunya Minami, Tatsuya Ishigaki, Ikko Hamamura, Taku Mikuriya, Youmi Ma, Naoaki Okazaki, Hiroya Takamura, Yohichi Suzuki, Tadashi Kadowaki
Abstract: Large language models are now integrated into many scientific workflows, accelerating data analysis, hypothesis generation, and design space exploration. In parallel with this growth, there is a growing need to carefully evaluate whether models accurately capture domain-specific knowledge and notation, since general-purpose benchmarks rarely reflect these requirements. This gap is especially clear in quantum science, which features non-intuitive phenomena and requires advanced mathematics. In this study, we introduce QuantumBench, a benchmark for the quantum domain that systematically examine how well LLMs understand and can be applied to this non-intuitive field. Using publicly available materials, we compiled approximately 800 questions with their answers spanning nine areas related to quantum science and organized them into an eight-option multiple-choice dataset. With this benchmark, we evaluate several existing LLMs and analyze their performance in the quantum domain, including sensitivity to changes in question format. QuantumBench is the first LLM evaluation dataset built for the quantum domain, and it is intended to guide the effective use of LLMs in quantum research.
Authors: Nils Porsche, Flurin M\"uller-Diesing, Sweta Banerjee, Miguel Goncalves, Marc Aubreville
Abstract: Confocal laser endomicroscopy (CLE) is a non-invasive, real-time imaging modality that can be used for in-situ, in-vivo imaging and the microstructural analysis of mucous structures. The diagnosis using CLE is, however, complicated by images being hard to interpret for non-experienced physicians. Utilizing machine learning as an augmentative tool would hence be beneficial, but is complicated by the shortage of histopathology-correlated CLE imaging sequences with respect to the plurality of patterns in this domain, leading to overfitting of machine learning models. To overcome this, self-supervised learning (SSL) can be employed on larger unlabeled datasets. CLE is a video-based modality with high inter-frame correlation, leading to a non-stratified data distribution for SSL training. In this work, we propose a filter functionality on CLE video sequences to reduce the dataset redundancy in SSL training and improve SSL training convergence and training efficiency. We use four state-of-the-art baseline networks and a SSL teacher-student network with a vision transformer small backbone for the evaluation. These networks were evaluated on downstream tasks for a sinonasal tumor dataset and a squamous cell carcinoma of the skin dataset. On both datasets, we found the highest test accuracy on the filtered SSL-pretrained model, with 67.48% and 73.52%, both considerably outperforming their non-SSL baselines. Our results show that SSL is an effective method for CLE pretraining. Further, we show that our proposed CLE video filter can be utilized to improve training efficiency in self-supervised scenarios, resulting in a reduction of 67% in training time.
Authors: Hanae Elmekki, Amanda Spilkin, Ehsan Zakeri, Antonela Mariel Zanuttini, Ahmed Alagha, Hani Sami, Jamal Bentahar, Lyes Kadem, Wen-Fang Xie, Philippe Pibarot, Rabeb Mizouni, Hadi Otrok, Azzam Mourad, Sami Muhaidat
Abstract: Cardiac ultrasound (US) is among the most widely used diagnostic tools in cardiology for assessing heart health, but its effectiveness is limited by operator dependence, time constraints, and human error. The shortage of trained professionals, especially in remote areas, further restricts access. These issues underscore the need for automated solutions that can ensure consistent, and accessible cardiac imaging regardless of operator skill or location. Recent progress in artificial intelligence (AI), especially in deep reinforcement learning (DRL), has gained attention for enabling autonomous decision-making. However, existing DRL-based approaches to cardiac US scanning lack reproducibility, rely on proprietary data, and use simplified models. Motivated by these gaps, we present the first end-to-end framework that integrates generative AI and DRL to enable autonomous and reproducible cardiac US scanning. The framework comprises two components: (i) a conditional generative simulator combining Generative Adversarial Networks (GANs) with Variational Autoencoders (VAEs), that models the cardiac US environment producing realistic action-conditioned images; and (ii) a DRL module that leverages this simulator to learn autonomous, accurate scanning policies. The proposed framework delivers AI-driven guidance through expert-validated models that classify image type and assess quality, supports conditional generation of realistic US images, and establishes a reproducible foundation extendable to other organs. To ensure reproducibility, a publicly available dataset of real cardiac US scans is released. The solution is validated through several experiments. The VAE-GAN is benchmarked against existing GAN variants, with performance assessed using qualitative and quantitative approaches, while the DRL-based scanning system is evaluated under varying configurations to demonstrate effectiveness.
Authors: Gaby Maroun, Salah Eddine Bekhouche, Fadi Dornaika
Abstract: Age estimation from facial images is a complex and multifaceted challenge in computer vision. In this study, we present a novel hybrid architecture that combines ConvNeXt, a state-of-the-art advancement of convolutional neural networks (CNNs), with Vision Transformers (ViT). While each model independently delivers excellent performance on a variety of tasks, their integration leverages the complementary strengths of the CNNs localized feature extraction capabilities and the Transformers global attention mechanisms. Our proposed ConvNeXt-ViT hybrid solution was thoroughly evaluated on benchmark age estimation datasets, including MORPH II, CACD, and AFAD, and achieved superior performance in terms of mean absolute error (MAE). To address computational constraints, we leverage pre-trained models and systematically explore different configurations, using linear layers and advanced regularization techniques to optimize the architecture. Comprehensive ablation studies highlight the critical role of individual components and training strategies, and in particular emphasize the importance of adapted attention mechanisms within the CNN framework to improve the model focus on age-relevant facial features. The results show that the ConvNeXt-ViT hybrid not only outperforms traditional methods, but also provides a robust foundation for future advances in age estimation and related visual tasks. This work underscores the transformative potential of hybrid architectures and represents a promising direction for the seamless integration of CNNs and transformers to address complex computer vision challenges.
Authors: Michael D. Moffitt
Abstract: The Abstraction and Reasoning Corpus remains one of the most compelling and challenging benchmarks for tracking progress toward achieving Artificial General Intelligence. In contrast to other evaluation datasets designed to assess an agent's task-specific skills or accumulated knowledge, the ARC-AGI suite is specifically targeted at measuring skill acquisition efficiency, a trait that has (so far) been lacking in even the most sophisticated machine learning systems. For algorithms that require extensive intra-task exemplars, a significant constraint imposed by ARC-AGI is the modest cardinality of its demonstration set, comprising a small number of $\langle$ input, output $\rangle$ grids per task specifying the corresponding transformation. To embellish the space of viable sample pairs, this paper introduces ARC-GEN, an open-source procedural generator aimed at extending the original ARC-AGI training dataset as faithfully as possible. Unlike prior efforts, our generator is both exhaustive (covering all four-hundred tasks) and mimetic (more closely honoring the distributional properties and characteristics embodied in the initial ARC-AGI-1 release). We also discuss the use of this generator in establishing a static benchmark suite to verify the correctness of programs submitted to the 2025 Google Code Golf Championship.
Authors: Xiang Li, Till Jahnke, Rebecca Boll, Jiaqi Han, Minkai Xu, Michael Meyer, Maria Novella Piancastelli, Daniel Rolles, Artem Rudenko, Florian Trinter, Thomas J. A. Wolf, Jana B. Thayer, James P. Cryan, Stefano Ermon, Phay J. Ho
Abstract: Capturing the structural changes that molecules undergo during chemical reactions in real space and time is a long-standing dream and an essential prerequisite for understanding and ultimately controlling femtochemistry. A key approach to tackle this challenging task is Coulomb explosion imaging, which benefited decisively from recently emerging high-repetition-rate X-ray free-electron laser sources. With this technique, information on the molecular structure is inferred from the momentum distributions of the ions produced by the rapid Coulomb explosion of molecules. Retrieving molecular structures from these distributions poses a highly non-linear inverse problem that remains unsolved for molecules consisting of more than a few atoms. Here, we address this challenge using a diffusion-based Transformer neural network. We show that the network reconstructs unknown molecular geometries from ion-momentum distributions with a mean absolute error below one Bohr radius, which is half the length of a typical chemical bond.
Authors: Nicky Pochinkov, Yulia Volkova, Anna Vasileva, Sai V R Chereddy
Abstract: Interpretability studies in language models often investigate forward-looking representations of activations. However, as language models become capable of doing ever longer time horizon tasks, methods for understanding activations often remain limited to testing specific concepts or tokens. We develop a framework of Residual Stream Decoders as a method of probing model activations for paragraph-scale and document-scale plans. We test several methods and find information can be decoded equivalent to 5+ tokens of future context in small models. These results lay the groundwork for better monitoring of language models and better understanding how they might encode longer-term planning information.
Authors: Ziliang Chen, Xin Huang, Quanlong Guan, Liang Lin, Weiqi Luo
Abstract: The vision community is undergoing the unprecedented progress with the emergence of Vision-Language Pretraining Models (VLMs). Prompt learning plays as the holy grail of accessing VLMs since it enables their fast adaptation to downstream tasks with limited resources. Whereas existing researches milling around single-prompt paradigms, rarely investigate the technical potential behind their multi-prompt learning counterparts. This paper aims to provide a principled retrospect for vision-language multi-prompt learning. We extend the recent constant modality gap phenomenon to learnable prompts and then, justify the superiority of vision-language transfer with multi-prompt augmentation, empirically and theoretically. In terms of this observation, we propose an Energy-based Multi-prompt Learning (EMPL) to generate multiple prompt embeddings by drawing instances from an energy-based distribution, which is implicitly defined by VLMs. So our EMPL is not only parameter-efficient but also rigorously lead to the balance between in-domain and out-of-domain open-vocabulary generalization. Comprehensive experiments have been conducted to justify our claims and the excellence of EMPL.
Authors: Faranak Akbarifar, Nooshin Maghsoodi, Sean P Dukelow, Stephen Scott, Parvin Mousavi
Abstract: Purpose: Visually Guided Reaching (VGR) on the Kinarm robot yields sensitive kinematic biomarkers but requires 40-64 reaches, imposing time and fatigue burdens. We evaluate whether time-series foundation models can replace unrecorded trials from an early subset of reaches while preserving the reliability of standard Kinarm parameters. Methods: We analyzed VGR speed signals from 461 stroke and 599 control participants across 4- and 8-target reaching protocols. We withheld all but the first 8 or 16 reaching trials and used ARIMA, MOMENT, and Chronos models, fine-tuned on 70 percent of subjects, to forecast synthetic trials. We recomputed four kinematic features of reaching (reaction time, movement time, posture speed, maximum speed) on combined recorded plus forecasted trials and compared them to full-length references using ICC(2,1). Results: Chronos forecasts restored ICC >= 0.90 for all parameters with only 8 recorded trials plus forecasts, matching the reliability of 24-28 recorded reaches (Delta ICC <= 0.07). MOMENT yielded intermediate gains, while ARIMA improvements were minimal. Across cohorts and protocols, synthetic trials replaced reaches without materially compromising feature reliability. Conclusion: Foundation-model forecasting can greatly shorten Kinarm VGR assessment time. For the most impaired stroke survivors, sessions drop from 4-5 minutes to about 1 minute while preserving kinematic precision. This forecast-augmented paradigm promises efficient robotic evaluations for assessing motor impairments following stroke.
Authors: Jacqueline Mitchell, Yasser Shaaban
Abstract: ``Vibe coding'' -- the practice of developing software through iteratively conversing with a large language model (LLM) -- has exploded in popularity within the last year. However, developers report key limitations including the accumulation of technical debt, security issues, and code churn to achieve satisfactory results. We argue that these pitfalls result from LLMs' inability to reconcile accumulating human-imposed constraints during vibe coding, with developers inadvertently failing to resolve contradictions because LLMs prioritize user commands over code consistency. Given LLMs' receptiveness to verification-based feedback, we argue that formal methods can mitigate these pitfalls, making vibe coding more reliable. However, we posit that integrating formal methods must transcend existing approaches that combine formal methods and LLMs. We advocate for a side-car system throughout the vibe coding process which: (1) \emph{Autoformalizes} specifications (2) Validates against targets, (3) Delivers \emph{actionable} feedback to the LLM, and (4) Allows intuitive developer influence on specifications.
Authors: Haoming Yan, Xinyu Chen, Yanran Wang, Zhengchao Luo, Weizheng Huang, Hongshuai Wang, Peng Chen, Yuzhi Zhang, Weijie Sun, Jinzhuo Wang, Qihuang Gong, Rui Zhu, Lichen Zhao
Abstract: The discovery of effective molecular modulators is essential for advancing perovskite solar cells (PSCs), but the research process is hindered by the vastness of chemical space and the time-consuming and expensive trial-and-error experimental screening. Concurrently, machine learning (ML) offers significant potential for accelerating materials discovery. However, applying ML to PSCs remains a major challenge due to data scarcity and limitations of traditional quantitative structure-property relationship (QSPR) models. Here, we apply a chemical informed transfer learning framework based on pre-trained deep neural networks, which achieves high accuracy in predicting the molecular modulator's effect on the power conversion efficiency (PCE) of PSCs. This framework is established through systematical benchmarking of diverse molecular representations, enabling lowcost and high-throughput virtual screening over 79,043 commercially available molecules. Furthermore, we leverage interpretability techniques to visualize the learned chemical representation and experimentally characterize the resulting modulator-perovskite interactions. The top molecular modulators identified by the framework are subsequently validated experimentally, delivering a remarkably improved champion PCE of 26.91% in PSCs.
Authors: Mitchell L. Prevett, Francis K. C. Hui, Zhi Yang Tho, A. H. Welsh, Anton H. Westveld
Abstract: Linear mixed models are widely used for clustered data, but their reliance on parametric forms limits flexibility in complex and high-dimensional settings. In contrast, gradient boosting methods achieve high predictive accuracy through nonparametric estimation, but do not accommodate clustered data structures or provide uncertainty quantification. We introduce Gradient Boosted Mixed Models (GBMixed), a framework and algorithm that extends boosting to jointly estimate mean and variance components via likelihood-based gradients. In addition to nonparametric mean estimation, the method models both random effects and residual variances as potentially covariate-dependent functions using flexible base learners such as regression trees or splines, enabling nonparametric estimation while maintaining interpretability. Simulations and real-world applications demonstrate accurate recovery of variance components, calibrated prediction intervals, and improved predictive accuracy relative to standard linear mixed models and nonparametric methods. GBMixed provides heteroscedastic uncertainty quantification and introduces boosting for heterogeneous random effects. This enables covariate-dependent shrinkage for cluster-specific predictions to adapt between population and cluster-level data. Under standard causal assumptions, the framework enables estimation of heterogeneous treatment effects with reliable uncertainty quantification.
Authors: Zongyang Du, Shreeram Suresh Chandra, Ismail Rasim Ulgen, Aurosweta Mahapatra, Ali N. Salman, Carlos Busso, Berrak Sisman
Abstract: Everyday speech conveys far more than words, it reflects who we are, how we feel, and the circumstances surrounding our interactions. Yet, most existing speech datasets are acted, limited in scale, and fail to capture the expressive richness of real-life communication. With the rise of large neural networks, several large-scale speech corpora have emerged and been widely adopted across various speech processing tasks. However, the field of voice conversion (VC) still lacks large-scale, expressive, and real-life speech resources suitable for modeling natural prosody and emotion. To fill this gap, we release NaturalVoices (NV), the first large-scale spontaneous podcast dataset specifically designed for emotion-aware voice conversion. It comprises 5,049 hours of spontaneous podcast recordings with automatic annotations for emotion (categorical and attribute-based), speech quality, transcripts, speaker identity, and sound events. The dataset captures expressive emotional variation across thousands of speakers, diverse topics, and natural speaking styles. We also provide an open-source pipeline with modular annotation tools and flexible filtering, enabling researchers to construct customized subsets for a wide range of VC tasks. Experiments demonstrate that NaturalVoices supports the development of robust and generalizable VC models capable of producing natural, expressive speech, while revealing limitations of current architectures when applied to large-scale spontaneous data. These results suggest that NaturalVoices is both a valuable resource and a challenging benchmark for advancing the field of voice conversion. Dataset is available at: https://huggingface.co/JHU-SmileLab
Authors: Christian Prothmann, Vijay Gadepally, Jeremy Kepner, Koley Borchard, Luca Carlone, Zachary Folcik, J. Daniel Grith, Michael Houle, Jonathan P. How, Nathan Hughes, Ifueko Igbinedion, Hayden Jananthan, Tejas Jayashankar, Michael Jones, Sertac Karaman, Binoy G. Kurien, Alejandro Lancho, Giovanni Lavezzi, Gary C. F. Lee, Charles E. Leiserson, Richard Linares, Lindsey McEvoy, Peter Michaleas, Chasen Milner, Alex Pentland, Yury Polyanskiy, Jovan Popovich, Jeffrey Price, Tim W. Reid, Stephanie Riley, Siddharth Samsi, Peter Saunders, Olga Simek, Mark S. Veillette, Amir Weiss, Gregory W. Wornell, Daniela Rus, Scott T. Ruppel
Abstract: The DAF-MIT AI Accelerator is a collaboration between the United States Department of the Air Force (DAF) and the Massachusetts Institute of Technology (MIT). This program pioneers fundamental advances in artificial intelligence (AI) to expand the competitive advantage of the United States in the defense and civilian sectors. In recent years, AI Accelerator projects have developed and launched public challenge problems aimed at advancing AI research in priority areas. Hallmarks of AI Accelerator challenges include large, publicly available, and AI-ready datasets to stimulate open-source solutions and engage the wider academic and private sector AI ecosystem. This article supplements our previous publication, which introduced AI Accelerator challenges. We provide an update on how ongoing and new challenges have successfully contributed to AI research and applications of AI technologies.
Authors: Shounak Paul, Dhananjay Ghumare, Pawan Goyal, Saptarshi Ghosh, Ashutosh Modi
Abstract: Identifying/retrieving relevant statutes and prior cases/precedents for a given legal situation are common tasks exercised by law practitioners. Researchers to date have addressed the two tasks independently, thus developing completely different datasets and models for each task; however, both retrieval tasks are inherently related, e.g., similar cases tend to cite similar statutes (due to similar factual situation). In this paper, we address this gap. We propose IL-PCR (Indian Legal corpus for Prior Case and Statute Retrieval), which is a unique corpus that provides a common testbed for developing models for both the tasks (Statute Retrieval and Precedent Retrieval) that can exploit the dependence between the two. We experiment extensively with several baseline models on the tasks, including lexical models, semantic models and ensemble based on GNNs. Further, to exploit the dependence between the two tasks, we develop an LLM-based re-ranking approach that gives the best performance.
Authors: Abhinav Joshi, Vaibhav Sharma, Sanjeet Singh, Ashutosh Modi
Abstract: Sign language translation remains a challenging task due to the scarcity of large-scale, sentence-aligned datasets. Prior arts have focused on various feature extraction and architectural changes to support neural machine translation for sign languages. We propose POSESTITCH-SLT, a novel pre-training scheme that is inspired by linguistic-templates-based sentence generation technique. With translation comparison on two sign language datasets, How2Sign and iSign, we show that a simple transformer-based encoder-decoder architecture outperforms the prior art when considering template-generated sentence pairs in training. We achieve BLEU-4 score improvements from 1.97 to 4.56 on How2Sign and from 0.55 to 3.43 on iSign, surpassing prior state-of-the-art methods for pose-based gloss-free translation. The results demonstrate the effectiveness of template-driven synthetic supervision in low-resource sign language settings.
Authors: Meituan LongCat Team, Bairui Wang, Bayan, Bin Xiao, Bo Zhang, Bolin Rong, Borun Chen, Chang Wan, Chao Zhang, Chen Huang, Chen Chen, Chen Chen, Chengxu Yang, Chengzuo Yang, Cong Han, Dandan Peng, Delian Ruan, Detai Xin, Disong Wang, Dongchao Yang, Fanfan Liu, Fengjiao Chen, Fengyu Yang, Gan Dong, Gang Huang, Gang Xu, Guanglu Wan, Guoqiang Tan, Guoqiao Yu, Haibo Qiu, Hao Lu, Hongbo Liu, Hongyu Xiang, Jiaheng Wu, Jian Yang, Jiaxing Liu, Jing Huang, Jingang Wang, Jinrui Ding, Juchao Jiang, Jun Kuang, Jun Wang, Junhui Mei, Ke Ding, Kefeng Zhang, Lei Chen, Liang Shi, Limeng Qiao, Liming Zheng, Lin Ma, Liuyang Guo, Liya Ma, Luying Sun, Man Gao, Mengshen Zhu, Miao Cao, Minliang Lin, Nuo Xu, Peng Shi, Qi Zhang, Qian Fang, Qian Wang, Qian Yang, Quanxiu Wang, Rongxiang Weng, Rongxin Guo, Ruoxuan Liang, Senbin Yang, Shanbo Xu, Shanglin Lei, Shengze Ye, Shimin Chen, Shuaiqi Chen, Shujie Hu, Shuo Li, Siqi Yang, Siyu Xu, Siyu Ren, Song Li, Songxiang Liu, Tianhao Bai, Tianye Dai, Wei Hong, Wei Wang, Weixiao Zhao, Wengang Cao, Wenlong Zhu, Wenlong He, Xi Su, Xi Nan, Xiaohan Zhao, Xiaohao Wang, Xiaoyu Zhao, Xiaoyu Wang, Xiaoyu Li, Xin Pan, Xin Chen, Xiusong Sun, Xu Xiang, Xudong Xing, Xuezhi Cao, Xunliang Cai, Yang Yang, Yanli Tan, Yao Yao, Yerui Sun, Yi Chen, Yifan Lu, Yin Gong, Yining Zhang, Yitian Chen, Yiyang Gan, Yuchen Tang, Yuchen Xie, Yueqian Wang, Yuewen Zheng, Yufei Zhang, Yufeng Zhong, Yulei Qian, Yuqi Peng, Yuwei Jiang, Zeyang Hu, Zheng Zhang, Zhengkun Tian, Zhiqing Hong, Zhixiong Zeng, Zhuqi Mi, Ziran Li, Ziwen Wang, Ziyi Zhao, Ziyuan Zhuang, Zizhe Zhao
Abstract: We introduce LongCat-Flash-Omni, a state-of-the-art open-source omni-modal model with 560 billion parameters, excelling at real-time audio-visual interaction. By adopting a curriculum-inspired progressive training strategy that transitions from simpler to increasingly complex modality sequence modeling tasks, LongCat-Flash-Omni attains comprehensive multimodal capabilities while maintaining strong unimodal capability. Building upon LongCat-Flash, which adopts a high-performance Shortcut-connected Mixture-of-Experts (MoE) architecture with zero-computation experts, LongCat-Flash-Omni integrates efficient multimodal perception and speech reconstruction modules. Despite its immense size of 560B parameters (with 27B activated), LongCat-Flash-Omni achieves low-latency real-time audio-visual interaction. For training infrastructure, we developed a modality-decoupled parallelism scheme specifically designed to manage the data and model heterogeneity inherent in large-scale multimodal training. This innovative approach demonstrates exceptional efficiency by sustaining over 90% of the throughput achieved by text-only training. Extensive evaluations show that LongCat-Flash-Omni achieves state-of-the-art performance on omni-modal benchmarks among open-source models. Furthermore, it delivers highly competitive results across a wide range of modality-specific tasks, including text, image, and video understanding, as well as audio understanding and generation. We provide a comprehensive overview of the model architecture design, training procedures, and data strategies, and open-source the model to foster future research and development in the community.
Authors: Weidong Zhang, Pak Lun Kevin Ding, Huan Liu
Abstract: Lightweight vision classification models such as MobileNet, ShuffleNet, and EfficientNet are increasingly deployed in mobile and embedded systems, yet their performance has been predominantly benchmarked on ImageNet. This raises critical questions: Do models that excel on ImageNet also generalize across other domains? How can cross-dataset robustness be systematically quantified? And which architectural elements consistently drive generalization under tight resource constraints? Here, we present the first systematic evaluation of 11 lightweight vision models (2.5M parameters), trained under a fixed 100-epoch schedule across 7 diverse datasets. We introduce the Cross-Dataset Score (xScore), a unified metric that quantifies the consistency and robustness of model performance across diverse visual domains. Our results show that (1) ImageNet accuracy does not reliably predict performance on fine-grained or medical datasets, (2) xScore provides a scalable predictor of mobile model performance that can be estimated from just four datasets, and (3) certain architectural components--such as isotropic convolutions with higher spatial resolution and channel-wise attention--promote broader generalization, while Transformer-based blocks yield little additional benefit, despite incurring higher parameter overhead. This study provides a reproducible framework for evaluating lightweight vision models beyond ImageNet, highlights key design principles for mobile-friendly architectures, and guides the development of future models that generalize robustly across diverse application domains.
Authors: Siva Sai, Manish Prasad, Animesh Bhargava, Vinay Chamola, Rajkumar Buyya
Abstract: The rapid growth of Internet of Medical Things (IoMT) devices has resulted in significant security risks, particularly the risk of malware attacks on resource-constrained devices. Conventional deep learning methods are impractical due to resource limitations, while Federated Learning (FL) suffers from high communication overhead and vulnerability to non-IID (heterogeneous) data. In this paper, we propose a split learning (SL) based framework for IoT malware detection through image-based classification. By dividing the neural network training between the clients and an edge server, the framework reduces computational burden on resource-constrained clients while ensuring data privacy. We formulate a joint optimization problem that balances computation cost and communication efficiency by using a game-theoretic approach for attaining better training performance. Experimental evaluations show that the proposed framework outperforms popular FL methods in terms of accuracy (+6.35%), F1-score (+5.03%), high convergence speed (+14.96%), and less resource consumption (33.83%). These results establish the potential of SL as a scalable and secure paradigm for next-generation IoT security.
Authors: Hendrio Braganca, Diego Kreutz, Vanderson Rocha, Joner Assolin, and Eduardo Feitosa
Abstract: We present MH-1M, one of the most comprehensive and up-to-date datasets for advanced Android malware research. The dataset comprises 1,340,515 applications, encompassing a wide range of features and extensive metadata. To ensure accurate malware classification, we employ the VirusTotal API, integrating multiple detection engines for comprehensive and reliable assessment. Our GitHub, Figshare, and Harvard Dataverse repositories provide open access to the processed dataset and its extensive supplementary metadata, totaling more than 400 GB of data and including the outputs of the feature extraction pipeline as well as the corresponding VirusTotal reports. Our findings underscore the MH-1M dataset's invaluable role in understanding the evolving landscape of malware.
Authors: Amir Ziashahabi, Narges Ghasemi, Sajjad Shahabi, John Krumm, Salman Avestimehr, Cyrus Shahabi
Abstract: Accurate and up-to-date geospatial data are essential for urban planning, infrastructure monitoring, and environmental management. Yet, automating urban monitoring remains difficult because curated datasets of specific urban features and their changes are scarce. We introduce OSMGen, a generative framework that creates realistic satellite imagery directly from raw OpenStreetMap (OSM) data. Unlike prior work that relies on raster tiles, OSMGen uses the full richness of OSM JSON, including vector geometries, semantic tags, location, and time, giving fine-grained control over how scenes are generated. A central feature of the framework is the ability to produce consistent before-after image pairs: user edits to OSM inputs translate into targeted visual changes, while the rest of the scene is preserved. This makes it possible to generate training data that addresses scarcity and class imbalance, and to give planners a simple way to preview proposed interventions by editing map data. More broadly, OSMGen produces paired (JSON, image) data for both static and changed states, paving the way toward a closed-loop system where satellite imagery can automatically drive structured OSM updates. Source code is available at https://github.com/amir-zsh/OSMGen.
Authors: Kayua Oleques Paim, Rodrigo Brandao Mansilha, Diego Kreutz, Muriel Figueredo Franco, Weverton Cordeiro
Abstract: The rapid proliferation of Large Language Models (LLMs) has raised significant concerns about their security against adversarial attacks. In this work, we propose a novel approach to crafting universal jailbreaks and data extraction attacks by exploiting latent space discontinuities, an architectural vulnerability related to the sparsity of training data. Unlike previous methods, our technique generalizes across various models and interfaces, proving highly effective in seven state-of-the-art LLMs and one image generation model. Initial results indicate that when these discontinuities are exploited, they can consistently and profoundly compromise model behavior, even in the presence of layered defenses. The findings suggest that this strategy has substantial potential as a systemic attack vector.
Authors: Adrita Rahman Tory, Khondokar Fida Hasan, Md Saifur Rahman, Nickolaos Koroniotis, Mohammad Ali Moni
Abstract: Network Intrusion Detection Systems (NIDS) developed us- ing publicly available datasets predominantly focus on enterprise environ- ments, raising concerns about their effectiveness for converged Informa- tion Technology (IT) and Operational Technology (OT) in energy infras- tructures. This study evaluates the representativeness of five widely used datasets: CIC-IDS2017, SWaT, WADI, Sherlock, and CIC-Modbus2023 against network-detectable MITRE ATT&CK techniques extracted from documented energy sector incidents. Using a structured five-step analyt- ical approach, this article successfully developed and performed a gap analysis that identified 94 network observable techniques from an initial pool of 274 ATT&CK techniques. Sherlock dataset exhibited the high- est mean coverage (0.56), followed closely by CIC-IDS2017 (0.55), while SWaT and WADI recorded the lowest scores (0.38). Combining CIC- IDS2017, Sherlock, and CIC-Modbus2023 achieved an aggregate coverage of 92%, highlighting their complementary strengths. The analysis identi- fies critical gaps, particularly in lateral movement and industrial protocol manipulation, providing a clear pathway for dataset enhancement and more robust NIDS evaluation in hybrid IT/OT energy environments.
Authors: Kayua Oleques Paim, Angelo Gaspar Diniz Nogueira, Diego Kreutz, Weverton Cordeiro, Rodrigo Brandao Mansilha
Abstract: High-quality data scarcity hinders malware detection, limiting ML performance. We introduce MalDataGen, an open-source modular framework for generating high-fidelity synthetic tabular data using modular deep learning models (e.g., WGAN-GP, VQ-VAE). Evaluated via dual validation (TR-TS/TS-TR), seven classifiers, and utility metrics, MalDataGen outperforms benchmarks like SDV while preserving data utility. Its flexible design enables seamless integration into detection pipelines, offering a practical solution for cybersecurity applications.
Authors: Krishna Prasath Logakannan, Shridhar Vashishtha, Jacob Hochhalter, Shandian Zhe, Robert M. Kirby
Abstract: Digital twins are developed to model the behavior of a specific physical asset (or twin), and they can consist of high-fidelity physics-based models or surrogates. A highly accurate surrogate is often preferred over multi-physics models as they enable forecasting the physical twin future state in real-time. To adapt to a specific physical twin, the digital twin model must be updated using in-service data from that physical twin. Here, we extend Gaussian process (GP) models to include derivative data, for improved accuracy, with dynamic updating to ingest physical twin data during service. Including derivative data, however, comes at a prohibitive cost of increased covariance matrix dimension. We circumvent this issue by using a sparse GP approximation, for which we develop extensions to incorporate derivatives. Numerical experiments demonstrate that the prediction accuracy of the derivative-enhanced sparse GP method produces improved models upon dynamic data additions. Lastly, we apply the developed algorithm within a DT framework to model fatigue crack growth in an aerospace vehicle.
Authors: Mina Taraghi, Yann Pequignot, Amin Nikanjam, Mohamed Amine Merzouk, Foutse Khomh
Abstract: Organizations are increasingly adopting and adapting Large Language Models (LLMs) hosted on public repositories such as HuggingFace. Although these adaptations often improve performance on specialized downstream tasks, recent evidence indicates that they can also degrade a model's safety or fairness. Since different fine-tuning techniques may exert distinct effects on these critical dimensions, this study undertakes a systematic assessment of their trade-offs. Four widely used Parameter-Efficient Fine-Tuning methods, LoRA, IA3, Prompt-Tuning, and P-Tuning, are applied to four instruction-tuned model families (Meta-Llama-3-8B, Qwen2.5-7B, Mistral-7B, and Gemma-7B). In total, 235 fine-tuned variants are evaluated across eleven safety hazard categories and nine demographic fairness dimensions. The results show that adapter-based approaches (LoRA, IA3) tend to improve safety scores and are the least disruptive to fairness, retaining higher accuracy and lower bias scores. In contrast, prompt-based methods (Prompt-Tuning and P-Tuning) generally reduce safety and cause larger fairness regressions, with decreased accuracy and increased bias. Alignment shifts are strongly moderated by base model type: LLaMA remains stable, Qwen records modest gains, Gemma experiences the steepest safety decline, and Mistral, which is released without an internal moderation layer, displays the greatest variance. Improvements in safety do not necessarily translate into improvements in fairness, and no single configuration optimizes all fairness metrics simultaneously, indicating an inherent trade-off between these objectives. These findings suggest a practical guideline for safety-critical deployments: begin with a well-aligned base model, favour adapter-based PEFT, and conduct category-specific audits of both safety and fairness.
Authors: Naoto Iwase, Hiroki Okuyama, Junichiro Iwasawa
Abstract: Large language models (LLMs) show increasing promise in medical applications, but their ability to detect and correct errors in clinical texts -- a prerequisite for safe deployment -- remains under-evaluated, particularly beyond English. We introduce MedRECT, a cross-lingual benchmark (Japanese/English) that formulates medical error handling as three subtasks: error detection, error localization (sentence extraction), and error correction. MedRECT is built with a scalable, automated pipeline from the Japanese Medical Licensing Examinations (JMLE) and a curated English counterpart, yielding MedRECT-ja (663 texts) and MedRECT-en (458 texts) with comparable error/no-error balance. We evaluate 9 contemporary LLMs spanning proprietary, open-weight, and reasoning families. Key findings: (i) reasoning models substantially outperform standard architectures, with up to 13.5% relative improvement in error detection and 51.0% in sentence extraction; (ii) cross-lingual evaluation reveals 5-10% performance gaps from English to Japanese, with smaller disparities for reasoning models; (iii) targeted LoRA fine-tuning yields asymmetric improvements in error correction performance (Japanese: +0.078, English: +0.168) while preserving reasoning capabilities; and (iv) our fine-tuned model exceeds human expert performance on structured medical error correction tasks. To our knowledge, MedRECT is the first comprehensive cross-lingual benchmark for medical error correction, providing a reproducible framework and resources for developing safer medical LLMs across languages.
Authors: Andrea Angino, Matteo Aurina, Alena Kopani\v{c}\'akov\'a, Matthias Voigt, Marco Donatelli, Rolf Krause
Abstract: We introduce two multifidelity trust-region methods based on the Magical Trust Region (MTR) framework. MTR augments the classical trust-region step with a secondary, informative direction. In our approaches, the secondary ``magical'' directions are determined by solving coarse trust-region subproblems based on low-fidelity objective models. The first proposed method, Sketched Trust-Region (STR), constructs this secondary direction using a sketched matrix to reduce the dimensionality of the trust-region subproblem. The second method, SVD Trust-Region (SVDTR), defines the magical direction via a truncated singular value decomposition of the dataset, capturing the leading directions of variability. Several numerical examples illustrate the potential gain in efficiency.
Authors: Xin Yao, Haiyang Zhao, Yimin Chen, Jiawei Guo, Kecheng Huang, Ming Zhao
Abstract: The Contrastive Language-Image Pretraining (CLIP) model has significantly advanced vision-language modeling by aligning image-text pairs from large-scale web data through self-supervised contrastive learning. Yet, its reliance on uncurated Internet-sourced data exposes it to data poisoning and backdoor risks. While existing studies primarily investigate image-based attacks, the text modality, which is equally central to CLIP's training, remains underexplored. In this work, we introduce ToxicTextCLIP, a framework for generating high-quality adversarial texts that target CLIP during the pre-training phase. The framework addresses two key challenges: semantic misalignment caused by background inconsistency with the target class, and the scarcity of background-consistent texts. To this end, ToxicTextCLIP iteratively applies: 1) a background-aware selector that prioritizes texts with background content aligned to the target class, and 2) a background-driven augmenter that generates semantically coherent and diverse poisoned samples. Extensive experiments on classification and retrieval tasks show that ToxicTextCLIP achieves up to 95.83% poisoning success and 98.68% backdoor Hit@1, while bypassing RoCLIP, CleanCLIP and SafeCLIP defenses. The source code can be accessed via https://github.com/xinyaocse/ToxicTextCLIP/.
Authors: Xiaolong Li, Zhi-Qin John Xu, Yan Ren, Tianming Qiu, Xiaowen Wang
Abstract: Accurate segmentation of pediatric brain tumors in multi-parametric magnetic resonance imaging (mpMRI) is critical for diagnosis, treatment planning, and monitoring, yet faces unique challenges due to limited data, high anatomical variability, and heterogeneous imaging across institutions. In this work, we present an advanced nnU-Net framework tailored for BraTS 2025 Task-6 (PED), the largest public dataset of pre-treatment pediatric high-grade gliomas. Our contributions include: (1) a widened residual encoder with squeeze-and-excitation (SE) attention; (2) 3D depthwise separable convolutions; (3) a specificity-driven regularization term; and (4) small-scale Gaussian weight initialization. We further refine predictions with two postprocessing steps. Our models achieved first place on the Task-6 validation leaderboard, attaining lesion-wise Dice scores of 0.759 (CC), 0.967 (ED), 0.826 (ET), 0.910 (NET), 0.928 (TC) and 0.928 (WT).
Authors: Weihao Bo, Yanpeng Sun, Yu Wang, Xinyu Zhang, Zechao Li
Abstract: In this paper, we introduce FedMGP, a new paradigm for personalized federated prompt learning in vision-language models. FedMGP equips each client with multiple groups of paired textual and visual prompts, enabling the model to capture diverse, fine-grained semantic and instance-level cues. A diversity loss is introduced to drive each prompt group to specialize in distinct and complementary semantic aspects, ensuring that the groups collectively cover a broader range of local characteristics. During communication, FedMGP employs a dynamic prompt aggregation strategy based on similarity-guided probabilistic sampling: each client computes the cosine similarity between its prompt groups and the global prompts from the previous round, then samples s groups via a softmax-weighted distribution. This soft selection mechanism preferentially aggregates semantically aligned knowledge while still enabling exploration of underrepresented patterns effectively balancing the preservation of common knowledge with client-specific features. Notably, FedMGP maintains parameter efficiency by redistributing a fixed prompt capacity across multiple groups, achieving state-of-the-art performance with the lowest communication parameters among all federated prompt learning methods. Theoretical analysis shows that our dynamic aggregation strategy promotes robust global representation learning by reinforcing shared semantics while suppressing client-specific noise. Extensive experiments demonstrate that FedMGP consistently outperforms prior approaches in both personalization and domain generalization across diverse federated vision-language benchmarks. The code will be released on https://github.com/weihao-bo/FedMGP.git.
Authors: Gero Junike, Marco Oesting
Abstract: Neural networks are able to approximate any continuous function on a compact set. However, it is not obvious how to quantify the error of the neural network, i.e., the remaining bias between the function and the neural network. Here, we propose the application of extreme value theory to quantify large values of the error, which are typically relevant in applications. The distribution of the error beyond some threshold is approximately generalized Pareto distributed. We provide a new estimator of the shape parameter of the Pareto distribution suitable to describe the error of neural networks. Numerical experiments are provided.
Authors: Peter Atandoh, Jie Zou, Weikang Guo, Jiwei Wei, Zheng Wang
Abstract: Sentiment analysis using deep learning and pre-trained language models (PLMs) has gained significant traction due to their ability to capture rich contextual representations. However, existing approaches often underperform in scenarios involving nuanced emotional cues, domain shifts, and imbalanced sentiment distributions. We argue that these limitations stem from inadequate semantic grounding, poor generalization to diverse linguistic patterns, and biases toward dominant sentiment classes. To overcome these challenges, we propose CISEA-MRFE, a novel PLM-based framework integrating Contextual Instruction (CI), Semantic Enhancement Augmentation (SEA), and Multi-Refined Feature Extraction (MRFE). CI injects domain-aware directives to guide sentiment disambiguation; SEA improves robustness through sentiment-consistent paraphrastic augmentation; and MRFE combines a Scale-Adaptive Depthwise Encoder (SADE) for multi-scale feature specialization with an Emotion Evaluator Context Encoder (EECE) for affect-aware sequence modeling. Experimental results on four benchmark datasets demonstrate that CISEA-MRFE consistently outperforms strong baselines, achieving relative improvements in accuracy of up to 4.6% on IMDb, 6.5% on Yelp, 30.3% on Twitter, and 4.1% on Amazon. These results validate the effectiveness and generalization ability of our approach for sentiment classification across varied domains.
Authors: Qiang Li, Ningjing Zeng, Lina Yu
Abstract: Several studies have employed reinforcement learning (RL) to address the challenges of regional adaptive traffic signal control (ATSC) and achieved promising results. In this field, existing research predominantly adopts multi-agent frameworks. However, the adoption of multi-agent frameworks presents challenges for scalability. Instead, the Traffic signal control (TSC) problem necessitates a single-agent framework. TSC inherently relies on centralized management by a single control center, which can monitor traffic conditions across all roads in the study area and coordinate the control of all intersections. This work proposes a single-agent RL-based regional ATSC model compatible with probe vehicle technology. Key components of the RL design include state, action, and reward function definitions. To facilitate learning and manage congestion, both state and reward functions are defined based on queue length, with action designed to regulate queue dynamics. The queue length definition used in this study differs slightly from conventional definitions but is closely correlated with congestion states. More importantly, it allows for reliable estimation using link travel time data from probe vehicles. With probe vehicle data already covering most urban roads, this feature enhances the proposed method's potential for widespread deployment. The method was comprehensively evaluated using the SUMO simulation platform. Experimental results demonstrate that the proposed model effectively mitigates large-scale regional congestion levels via coordinated multi-intersection control.
Authors: Massinissa Merouani, Islem Kara Bernou, Riyadh Baghdadi
Abstract: Automatic code optimization remains a difficult challenge, particularly for complex loop nests on modern hardware. This paper investigates a novel approach to code optimization where Large Language Models (LLMs) guide the process through a closed-loop interaction with a compiler. We present ComPilot, an experimental framework that leverages off-the-shelf LLMs, without any task-specific fine-tuning, as interactive optimization agents. ComPilot establishes a feedback loop where an LLM proposes transformations for a given loop nest to a compiler. The compiler attempts the transformations, reporting back legality status and measured speedup or slowdown. The LLM utilizes this concrete feedback to iteratively refine its optimization strategy. Our extensive evaluation across the PolyBench benchmark suite demonstrates the effectiveness of this zero-shot approach. ComPilot achieves geometric mean speedups of 2.66x (single run) and 3.54x (best-of-5 runs) over the original code. Furthermore, ComPilot demonstrates competitive performance against the state-of-the-art Pluto polyhedral optimizer, outperforming it in many cases. This experimental study demonstrates that general-purpose LLMs can effectively guide the code optimization process when grounded by compiler feedback, opening promising research directions for agentic AI in code optimization.
Authors: Mark Kocherovsky, Illya Bakurov, Wolfgang Banzhaf
Abstract: While crossover is a critical and often indispensable component in other forms of Genetic Programming, such as Linear- and Tree-based, it has consistently been claimed that it deteriorates search performance in CGP. As a result, a mutation-alone $(1+\lambda)$ evolutionary strategy has become the canonical approach for CGP. Although several operators have been developed that demonstrate an increased performance over the canonical method, a general solution to the problem is still lacking. In this paper, we compare basic crossover methods, namely one-point and uniform, to variants in which nodes are ``preserved,'' including the subgraph crossover developed by Roman Kalkreuth, the difference being that when ``node preservation'' is active, crossover is not allowed to break apart instructions. We also compare a node mutation operator to the traditional point mutation; the former simply replaces an entire node with a new one. We find that node preservation in both mutation and crossover improves search using symbolic regression benchmark problems, moving the field towards a general solution to CGP crossover.
Authors: Zicheng Xu, Guanchu Wang, Yu-Neng Chuang, Guangyao Zheng, Alexander S. Szalay, Zirui Liu, Vladimir Braverman
Abstract: Large Reasoning Models (LRMs) demonstrate strong performance on complex reasoning tasks, yet they often suffer from overthinking, producing excessively long chain-of-thought (CoT) traces that increase inference cost and may degrade accuracy. Our analysis reveals a clear anti-correlation between reasoning length and accuracy, where across multiple stochastic decodes, the short reasoning paths consistently achieve the highest correctness, while longer ones accumulate errors and repetitions. These short optimal reasoning paths can be found ideally through full enumeration of the reasoning space. However, the tree-structured reasoning space grows exponentially with sequence length, rendering exhaustive exploration infeasible. To address this, we propose DTS, a model-agnostic decoding framework that sketches the reasoning space by selectively branching at high-entropy tokens and applies early stopping to select the shortest completed reasoning path. This approach approximates the optimal solution that enhances both efficiency and accuracy, without requiring additional training or supervision. Experiments on AIME2024 and AIME2025 datasets with DeepSeek-R1-Distill-Qwen-7B and 1.5B show that DTS improves accuracy by up to 8%, reduces average reasoning length by 23%, and decreases repetition frequency by 12%, demonstrating DTS's ability for scalable and efficient LRM reasoning.
Authors: Zhiyang Ning, Benjamin Peherstorfer
Abstract: Uncertainty quantification in digital twins is critical to enable reliable and credible predictions beyond available data. A key challenge is that ensemble-based approaches can become prohibitively expensive when embedded in control and data assimilation loops in digital twins, even when reduced models are used. We introduce a reduced modeling approach that advances in time the mean and covariance of the reduced solution distribution induced by the initial condition uncertainties, which eliminates the need to maintain and propagate a costly ensemble of reduced solutions. The mean and covariance dynamics are obtained as a moment closure from Neural Galerkin schemes on pre-trained neural networks, which can be interpreted as filtered Neural Galerkin dynamics analogous to Gaussian filtering and the extended Kalman filter. Numerical experiments demonstrate that filtered Neural Galerkin schemes achieve more than one order of magnitude speedup compared to ensemble-based uncertainty propagation.
Authors: Weijie Su
Abstract: In this paper, we introduce a model for analyzing deep learning optimization over a single iteration by leveraging the matrix structure of the weights. We derive the model by assuming isotropy of curvature, including the second-order Hessian and higher-order terms, of the loss function across all perturbation directions; hence, we call it the isotropic curvature model. This model is a convex optimization program amenable to analysis, which allows us to understand how an update on the weights in the form of a matrix relates to the change in the total loss function. As an application, we use the isotropic curvature model to analyze the recently introduced Muon optimizer and other matrix-gradient methods for training language models. First, we show that under a general growth condition on the curvature, the optimal update matrix is obtained by making the spectrum of the original gradient matrix more homogeneous -- that is, making its singular values closer in ratio -- which in particular improves the conditioning of the update matrix. Next, we show that the orthogonalized gradient becomes optimal for the isotropic curvature model when the curvature exhibits a phase transition in growth. Taken together, these results suggest that the gradient orthogonalization employed in Muon and other related methods is directionally correct but may not be strictly optimal. Finally, we discuss future research on how to leverage the isotropic curvature model for designing new optimization methods for training deep learning and language models.
Authors: Mehmet Yigit Avci, Pedro Borges, Virginia Fernandez, Paul Wright, Mehmet Yigitsoy, Sebastien Ourselin, Jorge Cardoso
Abstract: Magnetic Resonance Imaging suffers from substantial data heterogeneity and the absence of standardized contrast labels across scanners, protocols, and institutions, which severely limits large-scale automated analysis. A unified representation of MRI contrast would enable a wide range of downstream utilities, from automatic sequence recognition to harmonization and quality control, without relying on manual annotations. To this end, we introduce MR-CLIP, a metadata-guided framework that learns MRI contrast representations by aligning volumetric images with their DICOM acquisition parameters. The resulting embeddings shows distinct clusters of MRI sequences and outperform supervised 3D baselines under data scarcity in few-shot sequence classification. Moreover, MR-CLIP enables unsupervised data quality control by identifying corrupted or inconsistent metadata through image-metadata embedding distances. By transforming routinely available acquisition metadata into a supervisory signal, MR-CLIP provides a scalable foundation for label-efficient MRI analysis across diverse clinical datasets.
Authors: Haoting Zhang, Haoxian Chen, Donglin Zhan, Hanyang Zhao, Henry Lam, Wenpin Tang, David Yao, Zeyu Zheng
Abstract: The field of simulation optimization (SO) encompasses various methods developed to optimize complex, expensive-to-sample stochastic systems. Established methods include, but are not limited to, ranking-and-selection for finite alternatives and surrogate-based methods for continuous domains, with broad applications in engineering and operations management. The recent advent of large language models (LLMs) offers a new paradigm for exploiting system structure and automating the strategic selection and composition of these established SO methods into a tailored optimization procedure. This work introduces SOCRATES (Simulation Optimization with Correlated Replicas and Adaptive Trajectory Evaluations), a novel two-stage procedure that leverages LLMs to automate the design of tailored SO algorithms. The first stage constructs an ensemble of digital replicas of the real system. An LLM is employed to implement causal discovery from a textual description of the system, generating a structural `skeleton' that guides the sample-efficient learning of the replicas. In the second stage, this replica ensemble is used as an inexpensive testbed to evaluate a set of baseline SO algorithms. An LLM then acts as a meta-optimizer, analyzing the performance trajectories of these algorithms to iteratively revise and compose a final, hybrid optimization schedule. This schedule is designed to be adaptive, with the ability to be updated during the final execution on the real system when the optimization performance deviates from expectations. By integrating LLM-driven reasoning with LLM-assisted trajectory-aware meta-optimization, SOCRATES creates an effective and sample-efficient solution for complex SO optimization problems.
Authors: Ritik Raj, Hong Wang, Tushar Krishna
Abstract: Agentic AI frameworks add a decision-making orchestrator embedded with external tools, including web search, Python interpreter, contextual database, and others, on top of monolithic LLMs, turning them from passive text oracles into autonomous problem-solvers that can plan, call tools, remember past steps, and adapt on the fly. This paper aims to characterize and understand the system bottlenecks introduced by agentic AI workloads from a largely overlooked CPU-centric perspective. We first systematically characterize Agentic AI on the basis of orchestrator/decision making component, inference path dynamics and repetitiveness of the agentic flow which directly influences the system-level performance. Thereafter, based on the characterization, we choose five representative agentic AI workloads- Haystack RAG, Toolformer, ChemCrow, Langchain and SWE-Agent to profile latency, throughput and energy metrics and demystify the significant impact of CPUs on these metrics relative to GPUs. We observe that - 1. Tool processing on CPUs can take up to 90.6% of the total latency; 2. Agentic throughput gets bottlenecked either by CPU factors - coherence, synchronization and over-subscription of cores or GPU factors - main memory capacity and bandwidth; \circled{3} CPU dynamic energy consumes up to 44% of the total dynamic energy at large batch sizes. Based on the profiling insights, we present two key optimizations- 1. CPU and GPU-Aware Micro-batching (CGAM) and 2. Mixed Agentic Workload Scheduling (MAWS) for homogeneous and heterogeneous agentic workloads respectively to demonstrate the potential to improve the performance, efficiency, and scalability of agentic AI. We achieve up to 2.1x and 1.41x P50 latency speedup compared to the multi-processing benchmark for homogeneous and heterogeneous agentic workloads respectively.
Authors: Andrew G. Moore
Abstract: Computation with the Ising model is central to future computing technologies like quantum annealing, adiabatic quantum computing, and thermodynamic classical computing. Traditionally, computed values have been equated with ground states. This paper generalizes computation with ground states to computation with spin averages, allowing computations to take place at high temperatures. It then introduces a systematic correspondence between Ising devices and neural networks and a simple method to run trained feed-forward neural networks on Ising-type hardware. Finally, a mathematical proof is offered that these implementations are always successful.
Authors: Kokila Kasuni Perera, Frank Neumann, Aneta Neumann
Abstract: Bayesian optimisation (BO) is a surrogate-based optimisation technique that efficiently solves expensive black-box functions with small evaluation budgets. Recent studies consider trust regions to improve the scalability of BO approaches when the problem space scales to more dimensions. Motivated by this research, we explore the effectiveness of trust region-based BO algorithms for diversity optimisation in different dimensional black box problems. We propose diversity optimisation approaches extending TuRBO1, which is the first BO method that uses a trust region-based approach for scalability. We extend TuRBO1 as divTuRBO1, which finds an optimal solution while maintaining a given distance threshold relative to a reference solution set. We propose two approaches to find diverse solutions for black-box functions by combining divTuRBO1 runs in a sequential and an interleaving fashion. We conduct experimental investigations on the proposed algorithms and compare their performance with that of the baseline method, ROBOT (rank-ordered Bayesian optimisation with trust regions). We evaluate proposed algorithms on benchmark functions with dimensions 2 to 20. Experimental investigations demonstrate that the proposed methods perform well, particularly in larger dimensions, even with a limited evaluation budget.
Authors: Peiru Wu, Maojun Zhai, Lingzhu Zhang
Abstract: Measuring similarity in urban spatial networks is key to understanding cities as complex systems. Yet most existing methods are not tailored for spatial networks and struggle to differentiate them effectively. We propose GCA-Sim, a similarity-evaluation framework based on graph cellular automata. Each submodel measures similarity by the divergence between value distributions recorded at multiple stages of an information evolution process. We find that some propagation rules magnify differences among network signals; we call this "network resonance." With an improved differentiable logic-gate network, we learn several submodels that induce network resonance. We evaluate similarity through clustering performance on fifty city-level and fifty district-level road networks. The submodels in this framework outperform existing methods, with Silhouette scores above 0.9. Using the best submodel, we further observe that planning-led street networks are less internally homogeneous than organically grown ones; morphological categories from different domains contribute with comparable importance; and degree, as a basic topological signal, becomes increasingly aligned with land value and related variables over iterations.
Authors: Raymond M. Xiong, Panyu Chen, Tianze Dong, Jian Lu, Benjamin Goldstein, Danyang Zhuo, Anru R. Zhang
Abstract: Electronic health records (EHRs) are central to modern healthcare delivery and research; yet, many researchers lack the database expertise necessary to write complex SQL queries or generate effective visualizations, limiting efficient data use and scientific discovery. To address this barrier, we introduce CELEC, a large language model (LLM)-powered framework for automated EHR data extraction and analytics. CELEC translates natural language queries into SQL using a prompting strategy that integrates schema information, few-shot demonstrations, and chain-of-thought reasoning, which together improve accuracy and robustness. On a subset of the EHRSQL benchmark, CELEC achieves execution accuracy comparable to prior systems while maintaining low latency, cost efficiency, and strict privacy by exposing only database metadata to the LLM. CELEC also adheres to strict privacy protocols: the LLM accesses only database metadata (e.g., table and column names), while all query execution occurs securely within the institutional environment, ensuring that no patient-level data is ever transmitted to or shared with the LLM. Ablation studies confirm that each component of the SQL generation pipeline, particularly the few-shot demonstrations, plays a critical role in performance. By lowering technical barriers and enabling medical researchers to query EHR databases directly, CELEC streamlines research workflows and accelerates biomedical discovery.
Authors: Ran Yan, Youhe Jiang, Tianyuan Wu, Jiaxuan Gao, Zhiyu Mei, Wei Fu, Haohui Mai, Wei Wang, Yi Wu, Binhang Yuan
Abstract: Maximizing training throughput and cost-efficiency of RL for LLMs is essential to democratize this advanced technique. One promising but challenging approach is to deploy such a computational workflow over heterogeneous GPUs. Unlike conventional large-scale LLM pretraining, RL training generally decomposes into three coupled stages, i.e., rollout generation, reward computation, and policy/value updates, which exhibit markedly different compute intensities, memory footprints, and communication patterns. Recent research shows that fully asynchronous RL training can disaggregate these stages across disjoint hardware pools without sacrificing training stability, creating a great opportunity for real-world heterogeneous deployment. To this end, we present AReaL-Hex, a heterogeneity-aware asynchronous RL training system that effectively schedules how to execute rollout generation and policy model training over heterogeneous GPUs while enforcing data staleness bounds. Concretely, we use a two-phase scheduler: (i) a constrained search with MILP to select per-stage parallelization strategies and workload assignments given a resource budget, and (ii) a graph-partitioning step that allocates heterogeneous GPUs and interconnects to maximize end-to-end throughput. Built atop a fully asynchronous RL architecture, AReaL-Hex maps HBM-I/O-bound generation and compute-bound optimization to more cost-efficient resources and balances their producer-consumer interactions to avoid both idleness and stale rollout trajectories. On the mathematical reasoning task with various model scales (1.5B, 7B, and 14B), compared to homogeneous deployments of state-of-the-art asynchronous RL systems: (i) When maintaining the same total budgets, AReaL-Hex delivers up to 1.50x higher training throughput; (ii) When achieving the same training throughput, AReaL-Hex results in up to 1.46x reduction in training cost.
Authors: Jie JW Wu, Ayanda Patrick Herlihy, Ahmad Saleem Mirza, Ali Afoud, Fatemeh Fard
Abstract: With the software industry shifting toward a data-driven culture, online A/B testing is a key tool for evaluating new technologies. However, deploying such experiments requires substantial resources, may negatively impact users, and involves long data collection periods. To address this, \textit{off-policy evaluation (OPE)}, or offline A/B testing, uses logged data to assess technologies and is fundamental in Reinforcement Learning, making it crucial in domains where online testing is costly or risky, such as healthcare, recommender systems, education, dialog systems, and robotics. Despite advances in coding LLMs and agentic AI, little is known about leveraging them to optimize OPE results. We investigate whether LLMs and LLM-based agents can improve OPE performance via code optimization. We propose \textit{GrowthHacker}, a benchmark with agent and baseline methods on large-scale real-world datasets, which iteratively optimizes code, evaluates results, and begins new optimization cycles. We collected datasets, established protocols, implemented baselines for OPE on the Open Bandit Pipeline (OBP)~\cite{saito2021openbanditdatasetpipeline} and Scope-RL~\cite{kiyohara2023scope}, and developed the \textit{two_agent} framework, which reduces system complexity while preserving optimization effectiveness. Results show the two_agent framework achieves 100% reliability and the highest average improvement of 106.7% among positive outcomes. Both two_agent and CrewAI reach 45% success rates, outperforming AutoGen's 34%. These findings demonstrate the feasibility of LLM-based agents as automated "growth hackers" to enhance OPE systems, with implications for scaling data-driven decision-making in production.
Authors: Shijie Zhou, Viet Dac Lai, Hao Tan, Jihyung Kil, Wanrong Zhu, Changyou Chen, Ruiyi Zhang
Abstract: Graphical user interface (GUI) grounding is a key function of computer-use agents, which maps natural-language instructions to actionable screen regions. Existing approaches based on Multimodal Large Language Models (MLLMs) typically formulate it as a text-based coordinate generation task, yet directly generating precise coordinates from visual inputs remains challenging and computationally intensive. An intuitive way to implement GUI grounding is to first select visual patches relevant to the instructions and then determine the precise click location within those patches. Based on the observations that general MLLMs have some native grounding capability, nested within their attentions, we propose GUI-AIMA, an attention-based and coordinate-free supervised fine-tuning framework for efficient GUI grounding. GUI-AIMA aligns the intrinsic multimodal attention of MLLMs with patch-wise grounding signals. These signals are calculated adaptively for diverse user instructions by multi-head aggregation on simplified query-visual attention matrices. Besides, its coordinate-free manner can easily integrate a plug-and-play zoom-in stage. GUI-AIMA-3B was trained with only 85k screenshots, demonstrating exceptional data efficiency and verifying that light training can trigger the native grounding capability of MLLMs. It achieves state-of-the-art performance among 3B models, attaining an average accuracy of 58.6% on ScreenSpot-Pro and 62.2% on OSWorld-G. Project page: https://github.com/sjz5202/GUI-AIMA
Authors: Stella Kombo, Masih Haseli, Skylar Wei, Joel W. Burdick
Abstract: Autonomous systems often must predict the motions of nearby agents from partial and noisy data. This paper asks and answers the question: "can we learn, in real-time, a nonlinear predictive model of another agent's motions?" Our online framework denoises and forecasts such dynamics using a modified sliding-window Hankel Dynamic Mode Decomposition (Hankel-DMD). Partial noisy measurements are embedded into a Hankel matrix, while an associated Page matrix enables singular-value hard thresholding (SVHT) to estimate the effective rank. A Cadzow projection enforces structured low-rank consistency, yielding a denoised trajectory and local noise variance estimates. From this representation, a time-varying Hankel-DMD lifted linear predictor is constructed for multi-step forecasts. The residual analysis provides variance-tracking signals that can support downstream estimators and risk-aware planning. We validate the approach in simulation under Gaussian and heavy-tailed noise, and experimentally on a dynamic crane testbed. Results show that the method achieves stable variance-aware denoising and short-horizon prediction suitable for integration into real-time control frameworks.
Authors: Zhexiao Huang, Weihao He, Shutao Deng, Junzhe Chen, Chao Yuan, Hongxin Wang, Changsheng Zhou
Abstract: Out-of-distribution (OOD) detection is essential for deploying deep learning models in open-world environments. Existing approaches, such as energy-based scoring and gradient-projection methods, typically rely on high-dimensional representations to separate in-distribution (ID) and OOD samples. We introduce P-OCS (Perturbations in the Orthogonal Complement Subspace), a lightweight and theoretically grounded method that operates in the orthogonal complement of the principal subspace defined by ID features. P-OCS applies a single projected perturbation restricted to this complementary subspace, enhancing subtle ID-OOD distinctions while preserving the geometry of ID representations. We show that a one-step update is sufficient in the small-perturbation regime and provide convergence guarantees for the resulting detection score. Experiments across multiple architectures and datasets demonstrate that P-OCS achieves state-of-the-art OOD detection with negligible computational cost and without requiring model retraining, access to OOD data, or changes to model architecture.
Authors: Yu Liu, Zhijie Liu, Zedong Yang, You-Fu Li, He Kong
Abstract: Predicting pedestrian crossing intentions is crucial for the navigation of mobile robots and intelligent vehicles. Although recent deep learning-based models have shown significant success in forecasting intentions, few consider incomplete observation under occlusion scenarios. To tackle this challenge, we propose an Occlusion-Aware Diffusion Model (ODM) that reconstructs occluded motion patterns and leverages them to guide future intention prediction. During the denoising stage, we introduce an occlusion-aware diffusion transformer architecture to estimate noise features associated with occluded patterns, thereby enhancing the model's ability to capture contextual relationships in occluded semantic scenarios. Furthermore, an occlusion mask-guided reverse process is introduced to effectively utilize observation information, reducing the accumulation of prediction errors and enhancing the accuracy of reconstructed motion features. The performance of the proposed method under various occlusion scenarios is comprehensively evaluated and compared with existing methods on popular benchmarks, namely PIE and JAAD. Extensive experimental results demonstrate that the proposed method achieves more robust performance than existing methods in the literature.
Authors: Amirabbas Afzali, Amirreza Velae, Iman Ahmadi, Mohammad Aliannejadi
Abstract: The presence of social biases in large language models (LLMs) has become a significant concern in AI research. These biases, often embedded in training data, can perpetuate harmful stereotypes and distort decision-making processes. When LLMs are integrated into ranking systems, they can propagate these biases, leading to unfair outcomes in critical applications such as search engines and recommendation systems. Backpack Language Models, unlike traditional transformer-based models that treat text sequences as monolithic structures, generate outputs as weighted combinations of non-contextual, learned word aspects, also known as senses. Leveraging this architecture, we propose a framework for debiasing ranking tasks. Our experimental results show that this framework effectively mitigates gender bias in text retrieval and ranking with minimal degradation in performance.
Authors: Hyeon Hwang, Yewon Cho, Chanwoong Yoon, Yein Park, Minju Song, Kyungjae Lee, Gangwoo Kim, Jaewoo Kang
Abstract: Step-by-step reasoning has become a standard approach for large language models (LLMs) to tackle complex tasks. While this paradigm has proven effective, it raises a fundamental question: How can we verify that an LLM's reasoning is accurately grounded in knowledge? To address this question, we introduce a novel evaluation suite that systematically assesses the knowledge grounding of intermediate reasoning. Our framework comprises three key components. (1) Principal Knowledge Collection, a large-scale repository of atomic knowledge essential for reasoning. Based on the collection, we propose (2) knowledge-grounded evaluation metrics designed to measure how well models recall and apply prerequisite knowledge in reasoning. These metrics are computed by our (3) evaluator LLM, a lightweight model optimized for cost-effective and reliable metric computation. Our evaluation suite demonstrates remarkable effectiveness in identifying missing or misapplied knowledge elements, providing crucial insights for uncovering fundamental reasoning deficiencies in LLMs. Beyond evaluation, we demonstrate how these metrics can be integrated into preference optimization, showcasing further applications of knowledge-grounded evaluation.
Authors: Simone Sarrocco, Philippe C. Cattin, Peter M. Maloca, Paul Friedrich, Philippe Valmaggia
Abstract: Purpose: To evaluate deep learning (DL) models for enhancing vitreous optical coherence tomography (OCT) image quality and reducing acquisition time. Methods: Conditional Denoising Diffusion Probabilistic Models (cDDPMs), Brownian Bridge Diffusion Models (BBDMs), U-Net, Pix2Pix, and Vector-Quantised Generative Adversarial Network (VQ-GAN) were used to generate high-quality spectral-domain (SD) vitreous OCT images. Inputs were SD ART10 images, and outputs were compared to pseudoART100 images obtained by averaging ten ART10 images per eye location. Model performance was assessed using image quality metrics and Visual Turing Tests, where ophthalmologists ranked generated images and evaluated anatomical fidelity. The best model's performance was further tested within the manually segmented vitreous on newly acquired data. Results: U-Net achieved the highest Peak Signal-to-Noise Ratio (PSNR: 30.230) and Structural Similarity Index Measure (SSIM: 0.820), followed by cDDPM. For Learned Perceptual Image Patch Similarity (LPIPS), Pix2Pix (0.697) and cDDPM (0.753) performed best. In the first Visual Turing Test, cDDPM ranked highest (3.07); in the second (best model only), cDDPM achieved a 32.9% fool rate and 85.7% anatomical preservation. On newly acquired data, cDDPM generated vitreous regions more similar in PSNR to the ART100 reference than true ART1 or ART10 B-scans and achieved higher PSNR on whole images when conditioned on ART1 than ART10. Conclusions: Results reveal discrepancies between quantitative metrics and clinical evaluation, highlighting the need for combined assessment. cDDPM showed strong potential for generating clinically meaningful vitreous OCT images while reducing acquisition time fourfold. Translational Relevance: cDDPMs show promise for clinical integration, supporting faster, higher-quality vitreous imaging. Dataset and code will be made publicly available.
Authors: Kyriakos Georgiou, Gianluca Fabiani, Constantinos Siettos, Athanasios N. Yannacopoulos
Abstract: We deal with the solution of the forward problem for high-dimensional parabolic PDEs with random feature (projection) neural networks (RFNNs). We first prove that there exists a single-hidden layer neural network with randomized heat-kernels arising from the fundamental solution (Green's functions) of the heat operator, that we call HEATNET, that provides an unbiased universal approximator to the solution of parabolic PDEs in arbitrary (high) dimensions, with the rate of convergence being analogous to the ${O}(N^{-1/2})$, where $N$ is the size of HEATNET. Thus, HEATNETs are explainable schemes, based on the analytical framework of parabolic PDEs, exploiting insights from physics-informed neural networks aided by numerical and functional analysis, and the structure of the corresponding solution operators. Importantly, we show how HEATNETs can be scaled up for the efficient numerical solution of arbitrary high-dimensional parabolic PDEs using suitable transformations and importance Monte Carlo sampling of the integral representation of the solution, in order to deal with the singularities of the heat kernel around the collocation points. We evaluate the performance of HEATNETs through benchmark linear parabolic problems up to 2,000 dimensions. We show that HEATNETs result in remarkable accuracy with the order of the approximation error ranging from $1.0E-05$ to $1.0E-07$ for problems up to 500 dimensions, and of the order of $1.0E-04$ to $1.0E-03$ for 1,000 to 2,000 dimensions, with a relatively low number (up to 15,000) of features.
Authors: Hasan Abdulla
Abstract: This study examines machine learning techniques like Decision Trees, Support Vector Machines, Logistic Regression, Neural Networks, and ensemble methods to detect Android malware. The study evaluates these models on a dataset of Android applications and analyzes their accuracy, efficiency, and real-world applicability. Key findings show that ensemble methods demonstrate superior performance, but there are trade-offs between model interpretability, efficiency, and accuracy. Given its increasing threat, the insights guide future research and practical use of ML to combat Android malware.
Authors: Mahdi Maleki, Reza Agahzadeh Ayoubi, Marouan Mizmizi, Umberto Spagnolini
Abstract: We investigate how fully-passive electromagnetic skins (EMSs) can be engineered to enhance channel charting (CC) in dense urban environments. We employ two complementary state-of-the-art CC techniques, semi-supervised t-distributed stochastic neighbor embedding (t-SNE) and a semi-supervised Autoencoder (AE), to verify the consistency of results across nonparametric and parametric mappings. We show that the accuracy of CC hinges on a balance between signal-to-noise ratio (SNR) and spatial dissimilarity: EMS codebooks that only maximize gain, as in conventional Reconfigurable Intelligent Surface (RIS) optimization, suppress location fingerprints and degrade CC, while randomized phases increase diversity but reduce SNR. To address this trade-off, we design static EMS phase profiles via a quantile-driven criterion that targets worst-case users and improves both trustworthiness and continuity. In a 3D ray-traced city at 30 GHz, the proposed EMS reduces the 90th-percentile localization error from > 50 m to < 25 m for both t-SNE and AE-based CC, and decreases severe trajectory dropouts by over 4x under 15% supervision. The improvements hold consistently across the evaluated configurations, establishing static, pre-configured EMS as a practical enabler of CC without reconfiguration overheads.
Authors: Tianming Liu, Jirong Yang, Yafeng Yin, Manzi Li, Linghao Wang, Zheng Zhu
Abstract: Effective modeling of how human travelers learn and adjust their travel behavior from interacting with transportation systems is critical for system assessment and planning. However, this task is also difficult due to the complex cognition and decision-making involved in such behavior. Recent research has begun to leverage Large Language Model (LLM) agents for this task. Building on this, we introduce a novel dual-agent framework that enables continuous learning and alignment between LLM agents and human travelers on learning and adaptation behavior from online data streams. Our approach involves a set of LLM traveler agents, equipped with a memory system and a learnable persona, which serve as simulators for human travelers. To ensure behavioral alignment, we introduce an LLM calibration agent that leverages the reasoning and analytical capabilities of LLMs to train the personas of these traveler agents. Working together, this dual-agent system is designed to track and align the underlying decision-making mechanisms of travelers and produce realistic, adaptive simulations. Using a real-world dataset from a day-to-day route choice experiment, we show our approach significantly outperforms existing LLM-based methods in both individual behavioral alignment and aggregate simulation accuracy. Furthermore, we demonstrate that our method moves beyond simple behavioral mimicry to capture the evolution of underlying learning processes, a deeper alignment that fosters robust generalization. Overall, our framework provides a new approach for creating adaptive and behaviorally realistic agents to simulate travelers' learning and adaptation that can benefit transportation simulation and policy analysis.
Authors: Julian Streit, Franziska Weindel, Reinhard Heckel
Abstract: We consider the reconstruction of a codeword from multiple noisy copies that are independently corrupted by insertions, deletions, and substitutions. This problem arises, for example, in DNA data storage. A common code construction uses a concatenated coding scheme that combines an outer linear block code with an inner code, which can be either a nonlinear marker code or a convolutional code. Outer decoding is done with Belief Propagation, and inner decoding is done with the Bahl-Cocke-Jelinek-Raviv (BCJR) algorithm. However, the BCJR algorithm scales exponentially with the number of noisy copies, which makes it infeasible to reconstruct a codeword from more than about four copies. In this work, we introduce BCJRFormer, a transformer-based neural inner decoder. BCJRFormer achieves error rates comparable to the BCJR algorithm for binary and quaternary single-message transmissions of marker codes. Importantly, BCJRFormer scales quadratically with the number of noisy copies. This property makes BCJRFormer well-suited for DNA data storage, where multiple reads of the same DNA strand occur. To lower error rates, we replace the Belief Propagation outer decoder with a transformer-based decoder. Together, these modifications yield an efficient and performant end-to-end transformer-based pipeline for decoding multiple noisy copies affected by insertion, deletion, and substitution errors. Additionally, we propose a novel cross-attending transformer architecture called ConvBCJRFormer. This architecture extends BCJRFormer to decode transmissions of convolutional codewords, serving as an initial step toward joint inner and outer decoding for more general linear code classes.
Authors: Hassan Ugail, Ismail Lujain Jaleel
Abstract: Art authentication of Francisco Goya's works presents complex computational challenges due to his heterogeneous stylistic evolution and extensive historical patterns of forgery. We introduce a novel multimodal machine learning framework that applies identical feature extraction techniques to both visual and X-ray radiographic images of Goya paintings. The unified feature extraction pipeline incorporates Grey-Level Co-occurrence Matrix descriptors, Local Binary Patterns, entropy measures, energy calculations, and colour distribution analysis applied consistently across both imaging modalities. The extracted features from both visual and X-ray images are processed through an optimised One-Class Support Vector Machine with hyperparameter tuning. Using a dataset of 24 authenticated Goya paintings with corresponding X-ray images, split into an 80/20 train-test configuration with 10-fold cross-validation, the framework achieves 97.8% classification accuracy with a 0.022 false positive rate. Case study analysis of ``Un Gigante'' demonstrates the practical efficacy of our pipeline, achieving 92.3% authentication confidence through unified multimodal feature analysis. Our results indicate substantial performance improvement over single-modal approaches, establishing the effectiveness of applying identical computational methods to both visual and radiographic imagery in art authentication applications.
Authors: Bowen Chen (DK), Jayesh Gajbhar (DK), Gregory Dusek (DK), Rob Redmon (DK), Patrick Hogan (DK), Paul Liu (DK), DelWayne Bohnenstiehl (DK), Dongkuan (DK), Xu, Ruoying He
Abstract: Artificial intelligence is transforming the sciences, yet general conversational AI systems often generate unverified "hallucinations" undermining scientific rigor. We present OceanAI, a conversational platform that integrates the natural-language fluency of open-source large language models (LLMs) with real-time, parameterized access to authoritative oceanographic data streams hosted by the National Oceanic and Atmospheric Administration (NOAA). Each query such as "What was Boston Harbor's highest water level in 2024?" triggers real-time API calls that identify, parse, and synthesize relevant datasets into reproducible natural-language responses and data visualizations. In a blind comparison with three widely used AI chat-interface products, only OceanAI produced NOAA-sourced values with original data references; others either declined to answer or provided unsupported results. Designed for extensibility, OceanAI connects to multiple NOAA data products and variables, supporting applications in marine hazard forecasting, ecosystem assessment, and water-quality monitoring. By grounding outputs and verifiable observations, OceanAI advances transparency, reproducibility, and trust, offering a scalable framework for AI-enabled decision support within the oceans. A public demonstration is available at https://oceanai.ai4ocean.xyz.
Authors: Ay\c{s}e Selin Okatan, Mustafa \.Ilhan Akba\c{s}, Laxima Niure Kandel, Berker Pek\"oz
Abstract: We analyze subliminal transfer in Transformer models, where a teacher embeds hidden traits that can be linearly decoded by a student without degrading main-task performance. Prior work often attributes transferability to global representational similarity, typically quantified with Centered Kernel Alignment (CKA). Using synthetic corpora with disentangled public and private labels, we distill students under matched and independent random initializations. We find that transfer strength hinges on alignment within a trait-discriminative subspace: same-seed students inherit this alignment and show higher leakage {\tau \approx} 0.24, whereas different-seed students--despite global CKA > 0.9--exhibit substantially reduced excess accuracy {\tau \approx} 0.12 - 0.13. We formalize this with subspace-level CKA diagnostic and residualized probes, showing that leakage tracks alignment within the trait-discriminative subspace rather than global representational similarity. Security controls (projection penalty, adversarial reversal, right-for-the-wrong-reasons regularization) reduce leakage in same-base models without impairing public-task fidelity. These results establish seed-induced uniqueness as a resilience property and argue for subspace-aware diagnostics for secure multi-model deployments.
Authors: Mihailo Stojnic
Abstract: Recent studies suggest that asymmetric binary perceptron (ABP) likely exhibits the so-called statistical-computational gap characterized with the appearance of two phase transitioning constraint density thresholds: \textbf{\emph{(i)}} the \emph{satisfiability threshold} $\alpha_c$, below/above which ABP succeeds/fails to operate as a storage memory; and \textbf{\emph{(ii)}} \emph{algorithmic threshold} $\alpha_a$, below/above which one can/cannot efficiently determine ABP's weight so that it operates as a storage memory. We consider a particular parametric utilization of \emph{fully lifted random duality theory} (fl RDT) [85] and study its potential ABP's algorithmic implications. A remarkable structural parametric change is uncovered as one progresses through fl RDT lifting levels. On the first two levels, the so-called $\c$ sequence -- a key parametric fl RDT component -- is of the (natural) decreasing type. A change of such phenomenology on higher levels is then connected to the $\alpha_c$ -- $\alpha_a$ threshold change. Namely, on the second level concrete numerical values give for the critical constraint density $\alpha=\alpha_c\approx 0.8331$. While progressing through higher levels decreases this estimate, already on the fifth level we observe a satisfactory level of convergence and obtain $\alpha\approx 0.7764$. This allows to draw two striking parallels: \textbf{\emph{(i)}} the obtained constraint density estimate is in a remarkable agrement with range $\alpha\in (0.77,0.78)$ of clustering defragmentation (believed to be responsible for failure of locally improving algorithms) [17,88]; and \textbf{\emph{(ii)}} the observed change of $\c$ sequence phenomenology closely matches the one of the negative Hopfield model for which the existence of efficient algorithms that closely approach similar type of threshold has been demonstrated recently [87].
Authors: Charles C. Margossian, Lawrence K. Saul
Abstract: We extend several recent results providing symmetry-based guarantees for variational inference (VI) with location-scale families. VI approximates a target density~$p$ by the best match $q^*$ in a family $Q$ of tractable distributions that in general does not contain $p$. It is known that VI can recover key properties of $p$, such as its mean and correlation matrix, when $p$ and $Q$ exhibit certain symmetries and $q^*$ is found by minimizing the reverse Kullback-Leibler divergence. We extend these guarantees in two important directions. First, we provide symmetry-based guarantees for a broader family of divergences, highlighting the properties of variational objectives under which VI provably recovers the mean and correlation matrix. Second, we obtain further guarantees for VI when the target density $p$ exhibits even and elliptical symmetries in some but not all of its coordinates. These partial symmetries arise naturally in Bayesian hierarchical models, where the prior induces a challenging geometry but still possesses axes of symmetry. We illustrate these theoretical results in a number of experimental settings.
Authors: Narges Ghasemi, Amir Ziashahabi, Salman Avestimehr, Cyrus Shahabi
Abstract: Image geolocalization, the task of determining an image's geographic origin, poses significant challenges, largely due to visual similarities across disparate locations and the large search space. To address these issues, we propose a hierarchical sequence prediction approach inspired by how humans narrow down locations from broad regions to specific addresses. Analogously, our model predicts geographic tokens hierarchically, first identifying a general region and then sequentially refining predictions to increasingly precise locations. Rather than relying on explicit semantic partitions, our method uses S2 cells, a nested, multiresolution global grid, and sequentially predicts finer-level cells conditioned on visual inputs and previous predictions. This procedure mirrors autoregressive text generation in large language models. Much like in language modeling, final performance depends not only on training but also on inference-time strategy. We investigate multiple top-down traversal methods for autoregressive sampling, incorporating techniques from test-time compute scaling used in language models. Specifically, we integrate beam search and multi-sample inference while exploring various selection strategies to determine the final output. This enables the model to manage uncertainty by exploring multiple plausible paths through the hierarchy. We evaluate our method on the Im2GPS3k and YFCC4k datasets against two distinct sets of baselines: those that operate without a Multimodal Large Language Model (MLLM) and those that leverage one. In the MLLM-free setting, our model surpasses other comparable baselines on nearly all metrics, achieving state-of-the-art performance with accuracy gains of up to 13.9%. When augmented with an MLLM, our model outperforms all baselines, setting a new state-of-the-art across all metrics. The source code is available at https://github.com/NNargesNN/GeoToken.
Authors: Md. Abid Hasan Rafi, Mst. Fatematuj Johora, Pankaj Bhowmik
Abstract: The emergence of 5G and 6G networks has established network slicing as a significant part of future service-oriented architectures, demanding refined identification methods supported by robust datasets. The article presents SliceVision-F2I, a dataset of synthetic samples for studying feature visualization in network slicing for next-generation networking systems. The dataset transforms multivariate Key Performance Indicator (KPI) vectors into visual representations through four distinct encoding methods: physically inspired mappings, Perlin noise, neural wallpapering, and fractal branching. For each encoding method, 30,000 samples are generated, each comprising a raw KPI vector and a corresponding RGB image at low-resolution pixels. The dataset simulates realistic and noisy network conditions to reflect operational uncertainties and measurement imperfections. SliceVision-F2I is suitable for tasks involving visual learning, network state classification, anomaly detection, and benchmarking of image-based machine learning techniques applied to network data. The dataset is publicly available and can be reused in various research contexts, including multivariate time series analysis, synthetic data generation, and feature-to-image transformations.
Authors: Alex Boyd, Andrew Warrington, Taha Kass-Hout, Parminder Bhatia, Danica Xiao
Abstract: Foundational marked temporal point process (MTPP) models, such as the Hawkes process, often use inexpressive model families in order to offer interpretable parameterizations of event data. On the other hand, neural MTPPs models forego this interpretability in favor of absolute predictive performance. In this work, we present a new family MTPP models: the hyper Hawkes process (HHP), which aims to be as flexible and performant as neural MTPPs, while retaining interpretable aspects. To achieve this, the HHP extends the classical Hawkes process to increase its expressivity by first expanding the dimension of the process into a latent space, and then introducing a hypernetwork to allow time- and data-dependent dynamics. These extensions define a highly performant MTPP family, achieving state-of-the-art performance across a range of benchmark tasks and metrics. Furthermore, by retaining the linearity of the recurrence, albeit now piecewise and conditionally linear, the HHP also retains much of the structure of the original Hawkes process, which we exploit to create direct probes into how the model creates predictions. HHP models therefore offer both state-of-the-art predictions, while also providing an opportunity to ``open the box'' and inspect how predictions were generated.
Authors: Y. Isabel Liu, Bowen Li, Benjamin Eysenbach, Tom Silver
Abstract: Long-horizon decision-making with sparse rewards and continuous states and actions remains a fundamental challenge in AI and robotics. Task and motion planning (TAMP) is a model-based framework that addresses this challenge by planning hierarchically with abstract actions (options). These options are manually defined, limiting the agent to behaviors that we as human engineers know how to program (pick, place, move). In this work, we propose Shortcut Learning for Abstract Planning (SLAP), a method that leverages existing TAMP options to automatically discover new ones. Our key idea is to use model-free reinforcement learning (RL) to learn shortcuts in the abstract planning graph induced by the existing options in TAMP. Without any additional assumptions or inputs, shortcut learning leads to shorter solutions than pure planning, and higher task success rates than flat and hierarchical RL. Qualitatively, SLAP discovers dynamic physical improvisations (e.g., slap, wiggle, wipe) that differ significantly from the manually-defined ones. In experiments in four simulated robotic environments, we show that SLAP solves and generalizes to a wide range of tasks, reducing overall plan lengths by over 50% and consistently outperforming planning and RL baselines.
Authors: Li Raymond, Salim Flora, Wang Sijin, Wright Brendan
Abstract: Charge carrier dynamics critically affect the efficiency and stability of organic photovoltaic devices, but they are challenging to model with traditional analytical methods. We introduce \b{eta}-Linearly Decoded Latent Ordinary Differential Equations (\b{eta}-LLODE), a machine learning framework that disentangles and reconstructs extraction dynamics from time-resolved charge extraction measurements of P3HT:PCBM cells. This model enables the isolated analysis of the underlying charge carrier behaviour, which was found to be well described by a compressed exponential decay. Furthermore, the learnt interpretable latent space enables simulation, including both interpolation and extrapolation of experimental measurement conditions, offering a predictive tool for solar cell research to support device study and optimisation.
Authors: Yoshihiro Maruyama
Abstract: We propose CatEquiv, a category-equivariant neural network for Human Activity Recognition (HAR) from inertial sensors that systematically encodes temporal, amplitude, and structural symmetries. In particular, we introduce the categorical symmetry product where cyclic time shifts, positive gains and the sensor-hierarchy poset together capture the categorical symmetry structure of the data. CatEquiv achieves equivariance with respect to the categorical symmetry product. On UCI-HAR under out-of-distribution perturbations, CatEquiv attains markedly higher robustness compared with circularly padded CNNs and plain CNNs. These results demonstrate that enforcing categorical symmetries yields strong invariance and generalization without additional model capacity.
Authors: Md Talha Mohsin, Ismail Abdulrashid
Abstract: Medical imaging relies heavily on large, labeled datasets. But, unfortunately, they are not always easily accessible in clinical settings. Additionally, many practitioners often face various structural obstacles like limited data availability, fragmented data systems, and unbalanced datasets. These barriers often lead to the increased diagnostic uncertainty, underrepresentation of certain conditions, reduced model robustness, and biased diagnostic decisions. In response to these challenges, approaches such as transfer learning, meta-learning, and multimodal fusion have made great strides. However, they still need a solid theoretical justification for why they succeed or fail in situations where data is scarce. To address this gap, we propose a unified theoretical framework that characterizes learning and inference under low-resource medical imaging conditions. We first formalize the learning objective under few-shot conditions and compute sample complexity constraints to estimate the smallest quantity of data needed to achieve clinically reliable accuracy. Then based on ideas from PAC-learning and PAC-Bayesian theory, we explain how multimodal integration encourages generalization and quantifies uncertainty under sparse supervision. We further propose a formal metric for explanation stability, offering interpretability guarantees under low-data conditions. Taken together, the proposed framework establishes a principled foundation for constructing dependable, data-efficient diagnostic systems by jointly characterizing sample efficiency, uncertainty quantification, and interpretability in a unified theoretical setting.
Authors: Sinho Chewi, Aram-Alexandre Pooladian, Matthew S. Zhang
Abstract: In this short note, we characterize stability of the Kim--Milman flow map -- also known as the probability flow ODE -- with respect to variations in the target measure. Rather than the Wasserstein distance, we show that stability holds with respect to the relative Fisher information
Authors: Emaad Manzoor, Eva Ascarza, Oded Netzer
Abstract: Salespeople frequently face the dynamic screening decision of whether to persist in a conversation or abandon it to pursue the next lead. Yet, little is known about how these decisions are made, whether they are efficient, or how to improve them. We study these decisions in the context of high-volume outbound sales where leads are ample, but time is scarce and failure is common. We formalize the dynamic screening decision as an optimal stopping problem and develop a generative language model-based sequential decision agent - a stopping agent - that learns whether and when to quit conversations by imitating a retrospectively-inferred optimal stopping policy. Our approach handles high-dimensional textual states, scales to large language models, and works with both open-source and proprietary language models. When applied to calls from a large European telecommunications firm, our stopping agent reduces the time spent on failed calls by 54% while preserving nearly all sales; reallocating the time saved increases expected sales by up to 37%. Upon examining the linguistic cues that drive salespeople's quitting decisions, we find that they tend to overweight a few salient expressions of consumer disinterest and mispredict call failure risk, suggesting cognitive bounds on their ability to make real-time conversational decisions. Our findings highlight the potential of artificial intelligence algorithms to correct cognitively-bounded human decisions and improve salesforce efficiency.
Authors: Ru Wang, Wei Huang, Qi Cao, Yusuke Iwasawa, Yutaka Matsuo, Jiaxian Guo
Abstract: Test-time reinforcement learning (TTRL) offers a label-free paradigm for adapting models using only synthetic signals at inference, but its success hinges on constructing reliable learning signals. Standard approaches such as majority voting often collapse to spurious yet popular answers. We introduce Self-Harmony, a framework built on a simple intuition: the correct answer should remain stable across both an original question and its paraphrase. Self-Harmony operationalizes this by employing a single model in two complementary roles: a Solver to produce answers and a Reframer to rephrase the input. Based on this, we further propose a pseudo-label method: instead of majority voting, it aggregates answer frequencies across these original and reframed views using the harmonic mean. This is a process that naturally selects for solutions stable under reframing, thereby avoiding the common trap of favoring view-dependent, spurious answers. Crucially, this requires no human supervision or auxiliary models. Across diverse reasoning benchmarks, Self-Harmony achieves state-of-the-art results at the label-free test-time setting, ranking first in 28 of 30 settings across multiple methods. Beyond accuracy, it demonstrates unprecedented robustness, with zero training failures in all experiments, underscoring its stability and reliability.
Authors: Jicong Fan
Abstract: Missing data is a fundamental challenge in data science, significantly hindering analysis and decision-making across a wide range of disciplines, including healthcare, bioinformatics, social science, e-commerce, and industrial monitoring. Despite decades of research and numerous imputation methods, the literature remains fragmented across fields, creating a critical need for a comprehensive synthesis that connects statistical foundations with modern machine learning advances. This work systematically reviews core concepts-including missingness mechanisms, single versus multiple imputation, and different imputation goals-and examines problem characteristics across various domains. It provides a thorough categorization of imputation methods, spanning classical techniques (e.g., regression, the EM algorithm) to modern approaches like low-rank and high-rank matrix completion, deep learning models (autoencoders, GANs, diffusion models, graph neural networks), and large language models. Special attention is given to methods for complex data types, such as tensors, time series, streaming data, graph-structured data, categorical data, and multimodal data. Beyond methodology, we investigate the crucial integration of imputation with downstream tasks like classification, clustering, and anomaly detection, examining both sequential pipelines and joint optimization frameworks. The review also assesses theoretical guarantees, benchmarking resources, and evaluation metrics. Finally, we identify critical challenges and future directions, emphasizing model selection and hyperparameter optimization, the growing importance of privacy-preserving imputation via federated learning, and the pursuit of generalizable models that can adapt across domains and data types, thereby outlining a roadmap for future research.
Authors: Hans Gundlach, Hrvoje Kukina, Jayson Lynch, Neil Thompson
Abstract: Quantum computing technology is advancing rapidly. Yet, even accounting for these trends, a quantum leap would be needed for quantum computers to mean- ingfully impact deep learning over the coming decade or two. We arrive at this conclusion based on a first-of-its-kind survey of quantum algorithms and how they match potential deep learning applications. This survey reveals three important areas where quantum computing could potentially accelerate deep learning, each of which faces a challenging roadblock to realizing its potential. First, quantum algorithms for matrix multiplication and other algorithms central to deep learning offer small theoretical improvements in the number of operations needed, but this advantage is overwhelmed on practical problem sizes by how slowly quantum computers do each operation. Second, some promising quantum algorithms depend on practical Quantum Random Access Memory (QRAM), which is underdeveloped. Finally, there are quantum algorithms that offer large theoretical advantages, but which are only applicable to special cases, limiting their practical benefits. In each of these areas, we support our arguments using quantitative forecasts of quantum advantage that build on the work by Choi et al. [2023] as well as new research on limitations and quantum hardware trends. Our analysis outlines the current scope of quantum deep learning and points to research directions that could lead to greater practical advances in the field.
Authors: Joonghyuk Shin, Zhengqi Li, Richard Zhang, Jun-Yan Zhu, Jaesik Park, Eli Schechtman, Xun Huang
Abstract: Current motion-conditioned video generation methods suffer from prohibitive latency (minutes per video) and non-causal processing that prevents real-time interaction. We present MotionStream, enabling sub-second latency with up to 29 FPS streaming generation on a single GPU. Our approach begins by augmenting a text-to-video model with motion control, which generates high-quality videos that adhere to the global text prompt and local motion guidance, but does not perform inference on the fly. As such, we distill this bidirectional teacher into a causal student through Self Forcing with Distribution Matching Distillation, enabling real-time streaming inference. Several key challenges arise when generating videos of long, potentially infinite time-horizons: (1) bridging the domain gap from training on finite length and extrapolating to infinite horizons, (2) sustaining high quality by preventing error accumulation, and (3) maintaining fast inference, without incurring growth in computational cost due to increasing context windows. A key to our approach is introducing carefully designed sliding-window causal attention, combined with attention sinks. By incorporating self-rollout with attention sinks and KV cache rolling during training, we properly simulate inference-time extrapolations with a fixed context window, enabling constant-speed generation of arbitrarily long videos. Our models achieve state-of-the-art results in motion following and video quality while being two orders of magnitude faster, uniquely enabling infinite-length streaming. With MotionStream, users can paint trajectories, control cameras, or transfer motion, and see results unfold in real-time, delivering a truly interactive experience.
Authors: Samet Demir, Zafer Dogan
Abstract: Pretrained Transformers excel at in-context learning (ICL), inferring new tasks from only a handful of examples. Yet, their ICL performance can degrade sharply under distribution shift between pretraining and test data, a regime increasingly common in real-world deployments. While recent empirical work hints that adjusting the attention temperature in the softmax can enhance Transformer performance, the attention temperature's role in ICL under distribution shift remains unexplored. This paper provides the first theoretical and empirical study of attention temperature for ICL under distribution shift. Using a simplified but expressive "linearized softmax" framework, we derive closed-form generalization error expressions and prove that shifts in input covariance or label noise substantially impair ICL, but that an optimal attention temperature exists which minimizes this error. We then validate our predictions through extensive simulations on linear regression tasks and large-scale experiments with GPT-2 and LLaMA2-7B on question-answering benchmarks. Our results establish attention temperature as a principled and powerful mechanism for improving the robustness of ICL in pretrained Transformers, advancing theoretical understanding and providing actionable guidance for selecting attention temperature in practice.
Authors: Tomer Shoham, Moshe Shenfeld, Noa Velner-Harris, Katrina Ligett
Abstract: We introduce a simple, general framework that takes any differentially private estimator of any arbitrary quantity as a black box, and from it constructs a differentially private nonparametric confidence interval of that quantity. Our approach repeatedly subsamples the data, applies the private estimator to each subsample, and then post-processes the resulting empirical CDF to a confidence interval. Our analysis uses the randomness from the subsampling to achieve privacy amplification. Under mild assumptions, the empirical CDF we obtain approaches the CDF of the private statistic as the sample size grows. We use this to show that the confidence intervals we estimate are asymptotically valid, tight, and equivalent to their non-private counterparts. We provide empirical evidence that our method performs well compared with the (less-general) state-of-the-art algorithms.
Authors: Hongyin Zhang, Shuo Zhang, Junxi Jin, Qixin Zeng, Runze Li, Donglin Wang
Abstract: Vision-Language-Action (VLA) models have recently emerged as powerful general-purpose policies for robotic manipulation, benefiting from large-scale multi-modal pre-training. However, they often fail to generalize reliably in out-of-distribution deployments, where unavoidable disturbances such as observation noise, sensor errors, or actuation perturbations become prevalent. While recent Reinforcement Learning (RL)-based post-training provides a practical means to adapt pre-trained VLA models, existing methods mainly emphasize reward maximization and overlook robustness to environmental uncertainty. In this work, we introduce RobustVLA, a lightweight online RL post-training method designed to explicitly enhance the resilience of VLA models. Through a systematic robustness analysis, we identify two key regularizations: Jacobian regularization, which mitigates sensitivity to observation noise, and smoothness regularization, which stabilizes policies under action perturbations. Extensive experiments across diverse robotic environments demonstrate that RobustVLA significantly outperforms prior state-of-the-art methods in robustness and reliability. Our results highlight the importance of principled robustness-aware RL post-training as a key step toward improving the reliability and robustness of VLA models.
Authors: Pavan Kumar Perepu
Abstract: Mathematical expressions (MEs) have complex two-dimensional structures in which symbols can be present at any nested depth like superscripts, subscripts, above, below etc. As MEs are represented using LaTeX format, several text retrieval methods based on string matching, vector space models etc., have also been applied for ME retrieval problem in the literature. As these methods are based on syntactic similarity, recently deep learning approaches based on embedding have been used for semantic similarity. In our present work, we have focused on the retrieval of mathematical expressions using deep learning approaches. In our approach, semantic features are extracted from the MEs using a deep recurrent neural network (DRNN) and these features have been used for matching and retrieval. We have trained the network for a classification task which determines the complexity of an ME. ME complexity has been quantified in terms of its nested depth. Based on the nested depth, we have considered three complexity classes of MEs: Simple, Medium and Complex. After training the network, outputs just before the the final fully connected layer are extracted for all the MEs. These outputs form the semantic features of MEs and are stored in a database. For a given ME query, its semantic features are computed using the trained DRNN and matched against the semantic feature database. Matching is performed based on the standard euclidean distance and top 'k' nearest matches are retrieved, where 'k' is a user-defined parameter. Our approach has been illustrated on a database of 829 MEs.
Authors: Reza Karimzadeh, Albert Alonso, Frans Zdyb, Julius B. Kirkegaard, Bulat Ibragimov
Abstract: Faithful yet compact explanations for vision models remain a challenge, as commonly used dense perturbation masks are often fragmented and overfitted, needing careful post-processing. Here, we present a training-free explanation method that replaces dense masks with smooth tunable contours. A star-convex region is parameterized by a truncated Fourier series and optimized under an extremal preserve/delete objective using the classifier gradients. The approach guarantees a single, simply connected mask, cuts the number of free parameters by orders of magnitude, and yields stable boundary updates without cleanup. Restricting solutions to low-dimensional, smooth contours makes the method robust to adversarial masking artifacts. On ImageNet classifiers, it matches the extremal fidelity of dense masks while producing compact, interpretable regions with improved run-to-run consistency. Explicit area control also enables importance contour maps, yielding a transparent fidelity-area profiles. Finally, we extend the approach to multi-contour and show how it can localize multiple objects within the same framework. Across benchmarks, the method achieves higher relevance mass and lower complexity than gradient and perturbation based baselines, with especially strong gains on self-supervised DINO models where it improves relevance mass by over 15% and maintains positive faithfulness correlations.
Authors: Sander Hummerich, Tristan Bereau, Ullrich K\"othe
Abstract: By reducing resolution, coarse-grained models greatly accelerate molecular simulations, unlocking access to long-timescale phenomena, though at the expense of microscopic information. Recovering this fine-grained detail is essential for tasks that depend on atomistic accuracy, making backmapping a central challenge in molecular modeling. We introduce split-flows, a novel flow-based approach that reinterprets backmapping as a continuous-time measure transport across resolutions. Unlike existing generative strategies, split-flows establish a direct probabilistic link between resolutions, enabling expressive conditional sampling of atomistic structures and -- for the first time -- a tractable route to computing mapping entropies, an information-theoretic measure of the irreducible detail lost in coarse-graining. We demonstrate these capabilities on diverse molecular systems, including chignolin, a lipid bilayer, and alanine dipeptide, highlighting split-flows as a principled framework for accurate backmapping and systematic evaluation of coarse-grained models.
Authors: Ayanava Dasgupta, Naqueeb Ahmad Warsi, Masahito Hayashi
Abstract: We develop a framework for quantum differential privacy (QDP) based on quantum hypothesis testing and Blackwell's ordering. This approach characterizes $(\eps,\delta)$-QDP via hypothesis testing divergences and identifies the most informative quantum state pairs under privacy constraints. We apply this to analyze the stability of quantum learning algorithms, generalizing classical results to the case $\delta>0$. Additionally, we study privatized quantum parameter estimation, deriving tight bounds on the quantum Fisher information under QDP. Finally, we establish near-optimal contraction bounds for differentially private quantum channels with respect to the hockey-stick divergence.
Authors: Irched Chafaa, E. Veronica Belmega, Giacomo Bacci
Abstract: Large multiple antenna arrays coupled with accu- rate beamforming are essential in terahertz (THz) communi- cations to ensure link reliability. However, as the number of antennas increases, beam alignment (focusing) and beam tracking in mobile networks incur prohibitive overhead. Additionally, the near-field region expands both with the size of antenna arrays and the carrier frequency, calling for adjustments in the beamforming to account for spherical wavefront instead of the conventional planar wave assumption. In this letter, we introduce a novel beam coherence time for mobile THz networks, to drastically reduce the rate of beam updates. Then, we propose a deep learning model, relying on a simple feedforward neural network with a time-dependent input, to predict the beam coherence time and adjust the beamforming on the fly with minimal overhead. Our numerical results demonstrate the effectiveness of the proposed approach by enabling higher data rates while reducing the overhead, especially at high (i.e., vehicular) mobility.
Authors: Aditya Kapoor, Yash Bhisikar, Benjamin Freed, Jan Peters, Mingfei Sun
Abstract: Effective communication in multi-agent reinforcement learning (MARL) is critical for success but constrained by bandwidth, yet past approaches have been limited to complex gating mechanisms that only decide \textit{whether} to communicate, not \textit{how precisely}. Learning to optimize message precision at the bit-level is fundamentally harder, as the required discretization step breaks gradient flow. We address this by generalizing Differentiable Discrete Communication Learning (DDCL), a framework for end-to-end optimization of discrete messages. Our primary contribution is an extension of DDCL to support unbounded signals, transforming it into a universal, plug-and-play layer for any MARL architecture. We verify our approach with three key results. First, through a qualitative analysis in a controlled environment, we demonstrate \textit{how} agents learn to dynamically modulate message precision according to the informational needs of the task. Second, we integrate our variant of DDCL into four state-of-the-art MARL algorithms, showing it reduces bandwidth by over an order of magnitude while matching or exceeding task performance. Finally, we provide direct evidence for the \enquote{Bitter Lesson} in MARL communication: a simple Transformer-based policy leveraging DDCL matches the performance of complex, specialized architectures, questioning the necessity of bespoke communication designs.
Authors: Nathan J. LeRoy, Donald R. Campbell Jr, Seth Stadick, Oleksandr Khoroshevskyi, Sang-Hoon Park, Ziyang Hu, Nathan C. Sheffield
Abstract: Introduction: Epigenomic datasets from high-throughput sequencing experiments are commonly summarized as genomic intervals. As the volume of this data grows, so does interest in analyzing it through deep learning. However, the heterogeneity of genomic interval data, where each dataset defines its own regions, creates barriers for machine learning methods that require consistent, discrete vocabularies. Methods: We introduce gtars-tokenizers, a high-performance library that maps genomic intervals to a predefined universe or vocabulary of regions, analogous to text tokenization in natural language processing. Built in Rust with bindings for Python, R, CLI, and WebAssembly, gtars-tokenizers implements two overlap methods (BITS and AIList) and integrates seamlessly with modern ML frameworks through Hugging Face-compatible APIs. Results: The gtars-tokenizers package achieves top efficiency for large-scale datasets, while enabling genomic intervals to be processed using standard ML workflows in PyTorch and TensorFlow without ad hoc preprocessing. This token-based approach bridges genomics and machine learning, supporting scalable and standardized analysis of interval data across diverse computational environments. Availability: PyPI and GitHub: https://github.com/databio/gtars.
Authors: Daniel M. Jimenez-Gutierrez, Enrique Zuazua, Joaquin Del Rio, Oleksii Sliusarenko, Xabi Uribe-Etxebarria
Abstract: Detecting malware, especially ransomware, is essential to securing today's interconnected ecosystems, including cloud storage, enterprise file-sharing, and database services. Training high-performing artificial intelligence (AI) detectors requires diverse datasets, which are often distributed across multiple organizations, making centralization necessary. However, centralized learning is often impractical due to security, privacy regulations, data ownership issues, and legal barriers to cross-organizational sharing. Compounding this challenge, ransomware evolves rapidly, demanding models that are both robust and adaptable. In this paper, we evaluate Federated Learning (FL) using the Sherpa.ai FL platform, which enables multiple organizations to collaboratively train a ransomware detection model while keeping raw data local and secure. This paradigm is particularly relevant for cybersecurity companies (including both software and hardware vendors) that deploy ransomware detection or firewall systems across millions of endpoints. In such environments, data cannot be transferred outside the customer's device due to strict security, privacy, or regulatory constraints. Although FL applies broadly to malware threats, we validate the approach using the Ransomware Storage Access Patterns (RanSAP) dataset. Our experiments demonstrate that FL improves ransomware detection accuracy by a relative 9% over server-local models and achieves performance comparable to centralized training. These results indicate that FL offers a scalable, high-performing, and privacy-preserving framework for proactive ransomware detection across organizational and regulatory boundaries.
Authors: Xinyue Yang, Chen Zheng, Yaoyang Hou, Renhao Zhang, Yiyan Zhang, Yanjun Wu, Heng Zhang
Abstract: Configuration tuning is critical for database performance. Although recent advancements in database tuning have shown promising results in throughput and latency improvement, challenges remain. First, the vast knob space makes direct optimization unstable and slow to converge. Second, reinforcement learning pipelines often lack effective warm-start guidance and require long offline training. Third, transferability is limited: when hardware or workloads change, existing models typically require substantial retraining to recover performance. To address these limitations, we propose L2T-Tune, a new LLM-guided hybrid database tuning framework that features a three-stage pipeline: Stage one performs a warm start that simultaneously generates uniform samples across the knob space and logs them into a shared pool; Stage two leverages a large language model to mine and prioritize tuning hints from manuals and community documents for rapid convergence. Stage three uses the warm-start sample pool to reduce the dimensionality of knobs and state features, then fine-tunes the configuration with the Twin Delayed Deep Deterministic Policy Gradient algorithm. We conduct experiments on L2T-Tune and the state-of-the-art models. Compared with the best-performing alternative, our approach improves performance by an average of 37.1% across all workloads, and by up to 73% on TPC-C. Compared with models trained with reinforcement learning, it achieves rapid convergence in the offline tuning stage on a single server. Moreover, during the online tuning stage, it only takes 30 steps to achieve best results.
Authors: Arran Carter, Torben Sell
Abstract: Bayesian neural networks (BNNs) allow rigorous uncertainty quantification in deep learning, but often come at a prohibitive computational cost. We propose three different innovative architectures of partial trace-class Bayesian neural networks (PaTraC BNNs) that enable uncertainty quantification comparable to standard BNNs but use significantly fewer Bayesian parameters. These PaTraC BNNs have computational and statistical advantages over standard Bayesian neural networks in terms of speed and memory requirements. Our proposed methodology therefore facilitates reliable, robust, and scalable uncertainty quantification in neural networks. The three architectures build on trace-class neural network priors which induce an ordering of the neural network parameters, and are thus a natural choice in our framework. In a numerical simulation study, we verify the claimed benefits, and further illustrate the performance of our proposed methodology on a real-world dataset.
Authors: Ayesha Gull, Muhammad Usman Safder, Rania Elbadry, Preslav Nakov, Zhuohan Xie
Abstract: Large Language Models (LLMs) are increasingly being applied to specialized, high-stakes domains like engineering, which demands rigorous evaluation of their complex reasoning capabilities. While current benchmarks assess language understanding, factual recall, mathematics or code generation, none capture the integrative reasoning central to engineering where scientific principles, quantitative modeling and practical constraints must converge. To address this gap, we introduce EngChain, a benchmark for verifiable multi-step engineering problem-solving. EngChain contains 90 problems spanning three engineering branches, organized into 9 domains and 20 distinct areas. The problems are generated from symbolic templates with a high degree of randomization to ensure diversity and eliminate the risk of contamination. With this benchmark, we move beyond final answer accuracy with a two-stage evaluation: we first quantitatively verify the numerical and semantic validity of each reasoning step and then introduce LLM-As-A-Judge, an automated system to qualitatively categorize the identified reasoning errors.
Authors: Congcong Chen, Xinyu Liu, Kaifeng Huang, Lifei Wei, Yang Shi
Abstract: Graph Neural Networks (GNNs) have marked significant impact in traffic state prediction, social recommendation, knowledge-aware question answering and so on. As more and more users move towards cloud computing, it has become a critical issue to unleash the power of GNNs while protecting the privacy in cloud environments. Specifically, the training data and inference data for GNNs need to be protected from being stolen by external adversaries. Meanwhile, the financial cost of cloud computing is another primary concern for users. Therefore, although existing studies have proposed privacy-preserving techniques for GNNs in cloud environments, their additional computational and communication overhead remain relatively high, causing high financial costs that limit their widespread adoption among users. To protect GNN privacy while lowering the additional financial costs, we introduce Panther, a cost-effective privacy-preserving framework for GNN training and inference services in cloud environments. Technically, Panther leverages four-party computation to asynchronously executing the secure array access protocol, and randomly pads the neighbor information of GNN nodes. We prove that Panther can protect privacy for both training and inference of GNN models. Our evaluation shows that Panther reduces the training and inference time by an average of 75.28% and 82.80%, respectively, and communication overhead by an average of 52.61% and 50.26% compared with the state-of-the-art, which is estimated to save an average of 55.05% and 59.00% in financial costs (based on on-demand pricing model) for the GNN training and inference process on Google Cloud Platform.
Authors: Jacob Carlson
Abstract: Social scientists are increasingly turning to unstructured datasets to unlock new empirical insights, e.g., estimating causal effects on text outcomes, measuring beliefs from open-ended survey responses. In such settings, unsupervised analysis is often of interest, in that the researcher does not want to pre-specify the objects of measurement or otherwise artificially delimit the space of measurable concepts; they are interested in discovery. This paper proposes a general and flexible framework for pursuing discovery from unstructured data in a statistically principled way. The framework leverages recent methods from the literature on machine learning interpretability to map unstructured data points to high-dimensional, sparse, and interpretable dictionaries of concepts; computes (test) statistics of these dictionary entries; and then performs selective inference on them using newly developed statistical procedures for high-dimensional exceedance control of the $k$-FWER under arbitrary dependence. The proposed framework has few researcher degrees of freedom, is fully replicable, and is cheap to implement -- both in terms of financial cost and researcher time. Applications to recent descriptive and causal analyses of unstructured data in empirical economics are explored. An open source Jupyter notebook is provided for researchers to implement the framework in their own projects.
Authors: Sharan Maiya, Henning Bartsch, Nathan Lambert, Evan Hubinger
Abstract: The character of the "AI assistant" persona generated by modern chatbot large language models influences both surface-level behavior and apparent values, beliefs, and ethics. These all affect interaction quality, perceived intelligence, and alignment with both developer and user intentions. The shaping of this persona, known as character training, is a critical component of industry post-training, yet remains effectively unstudied in the academic literature. We introduce the first open implementation of character training, leveraging Constitutional AI and a new data pipeline using synthetic introspective data to shape the assistant persona in a more effective and controlled manner than alternatives such as constraining system prompts or activation steering. Specifically, we fine-tune three popular open-weights models using 11 example personas, such as humorous, deeply caring, or even malevolent. To track the effects of our approach, we introduce a method which analyzes revealed preferences, uncovering clear and holistic changes in character. We find these changes are more robust to adversarial prompting than the above two alternatives, while also leading to more coherent and realistic generations. Finally, we demonstrate this fine-tuning has little to no effect on general capabilities as measured by common benchmarks. We describe and open-source our full post-training method, the implementation of which can be found at https://github.com/maiush/OpenCharacterTraining.
Authors: Mirco A. Mannucci
Abstract: A fundamental question in search-guided AI: what topology should guide Monte Carlo Tree Search (MCTS) in puzzle solving? Prior work applied topological features to guide MCTS in ARC-style tasks using grid topology -- the Laplacian spectral properties of cell connectivity -- and found no benefit. We identify the root cause: grid topology is constant across all instances. We propose measuring \emph{solution space topology} instead: the structure of valid color assignments constrained by detected pattern rules. We build this via compatibility graphs where nodes are $(cell, color)$ pairs and edges represent compatible assignments under pattern constraints. Our method: (1) detect pattern rules automatically with 100\% accuracy on 5 types, (2) construct compatibility graphs encoding solution space structure, (3) extract topological features (algebraic connectivity, rigidity, color structure) that vary with task difficulty, (4) integrate these features into MCTS node selection via sibling-normalized scores. We provide formal definitions, a rigorous selection formula, and comprehensive ablations showing that algebraic connectivity is the dominant signal. The work demonstrates that topology matters for search -- but only the \emph{right} topology. For puzzle solving, this is solution space structure, not problem space structure.
Authors: Sekh Mainul Islam, Pepa Atanasova, Isabelle Augenstein
Abstract: Natural Language Explanations (NLEs) describe how Large Language Models (LLMs) make decisions, drawing on both external Context Knowledge (CK) and Parametric Knowledge (PK) stored in model weights. Understanding their interaction is key to assessing the grounding of NLEs, yet it remains underexplored. Prior work has largely examined only single-step generation, typically the final answer, and has modelled PK and CK interaction only as a binary choice in a rank-1 subspace. This overlooks richer forms of interaction, such as complementary or supportive knowledge. We propose a novel rank-2 projection subspace that disentangles PK and CK contributions more accurately and use it for the first multi-step analysis of knowledge interactions across longer NLE sequences. Experiments on four QA datasets and three open-weight instruction-tuned LLMs show that diverse knowledge interactions are poorly represented in a rank-1 subspace but are effectively captured in our rank-2 formulation. Our multi-step analysis reveals that hallucinated NLEs align strongly with the PK direction, context-faithful ones balance PK and CK, and Chain-of-Thought prompting for NLEs shifts generated NLEs toward CK by reducing PK reliance. This work provides the first framework for systematic studies of multi-step knowledge interactions in LLMs through a richer rank-2 subspace disentanglement. Code and data: https://github.com/copenlu/pk-ck-knowledge-disentanglement.
URLs: https://github.com/copenlu/pk-ck-knowledge-disentanglement.
Authors: Jiale Lao, Andreas Zimmerer, Olga Ovcharenko, Tianji Cong, Matthew Russo, Gerardo Vitagliano, Michael Cochez, Fatma \"Ozcan, Gautam Gupta, Thibaud Hottelier, H. V. Jagadish, Kris Kissel, Sebastian Schelter, Andreas Kipf, Immanuel Trummer
Abstract: We present a benchmark targeting a novel class of systems: semantic query processing engines. Those systems rely inherently on generative and reasoning capabilities of state-of-the-art large language models (LLMs). They extend SQL with semantic operators, configured by natural language instructions, that are evaluated via LLMs and enable users to perform various operations on multimodal data. Our benchmark introduces diversity across three key dimensions: scenarios, modalities, and operators. Included are scenarios ranging from movie review analysis to medical question-answering. Within these scenarios, we cover different data modalities, including images, audio, and text. Finally, the queries involve a diverse set of operators, including semantic filters, joins, mappings, ranking, and classification operators. We evaluated our benchmark on three academic systems (LOTUS, Palimpzest, and ThalamusDB) and one industrial system, Google BigQuery. Although these results reflect a snapshot of systems under continuous development, our study offers crucial insights into their current strengths and weaknesses, illuminating promising directions for future research.
Authors: Yi Zhang, Zheng Wang, Chen Zhen, Wenjie Ruan, Qing Guo, Siddartha Khastgir, Carsten Maple, Xingyu Zhao
Abstract: Deep learning models are notoriously vulnerable to imperceptible perturbations. Most existing research centers on adversarial robustness (AR), which evaluates models under worst-case scenarios by examining the existence of deterministic adversarial examples (AEs). In contrast, probabilistic robustness (PR) adopts a statistical perspective, measuring the probability that predictions remain correct under stochastic perturbations. While PR is widely regarded as a practical complement to AR, dedicated training methods for improving PR are still relatively underexplored, albeit with emerging progress. Among the few PR-targeted training methods, we identify three limitations: i non-comparable evaluation protocols; ii limited comparisons to strong AT baselines despite anecdotal PR gains from AT; and iii no unified framework to compare the generalization of these methods. Thus, we introduce PRBench, the first benchmark dedicated to evaluating improvements in PR achieved by different robustness training methods. PRBench empirically compares most common AT and PR-targeted training methods using a comprehensive set of metrics, including clean accuracy, PR and AR performance, training efficiency, and generalization error (GE). We also provide theoretical analysis on the GE of PR performance across different training methods. Main findings revealed by PRBench include: AT methods are more versatile than PR-targeted training methods in terms of improving both AR and PR performance across diverse hyperparameter settings, while PR-targeted training methods consistently yield lower GE and higher clean accuracy. A leaderboard comprising 222 trained models across 7 datasets and 10 model architectures is publicly available at https://tmpspace.github.io/PRBenchLeaderboard/.
Authors: Soufiane Hayou
Abstract: We provide the first proof of learning rate transfer with width in a linear multi-layer perceptron (MLP) parametrized with $\mu$P, a neural network parameterization designed to ``maximize'' feature learning in the infinite-width limit. We show that under $\mu P$, the optimal learning rate converges to a \emph{non-zero constant} as width goes to infinity, providing a theoretical explanation to learning rate transfer. In contrast, we show that this property fails to hold under alternative parametrizations such as Standard Parametrization (SP) and Neural Tangent Parametrization (NTP). We provide intuitive proofs and support the theoretical findings with extensive empirical results.
Authors: Daniel Jimon, Mircea Vaida, Adriana Stan
Abstract: Audio denoising is critical in signal processing, enhancing intelligibility and fidelity for applications like restoring musical recordings. This paper presents a proof-of-concept for adapting a state-of-the-art neural audio codec, the Descript Audio Codec (DAC), for music denoising. This work overcomes the limitations of traditional architectures like U-Nets by training the model on a large-scale, custom-synthesized dataset built from diverse sources. Training is guided by a multi objective loss function that combines time-domain, spectral, and signal-level fidelity metrics. Ultimately, this paper aims to present a PoC for high-fidelity, generative audio restoration.
Authors: Javier Ballesteros-Jerez, Jesus Mart\'inez-G\'omez, Ismael Garc\'ia-Varea, Luis Orozco-Barbosa, Manuel Castillo-Cara
Abstract: We present a hybrid neural network model for inferring the position of mobile robots using Channel State Information (CSI) data from a Massive MIMO system. By leveraging an existing CSI dataset, our approach integrates a Convolutional Neural Network (CNN) with a Multilayer Perceptron (MLP) to form a Hybrid Neural Network (HyNN) that estimates 2D robot positions. CSI readings are converted into synthetic images using the TINTO tool. The localisation solution is integrated with a robotics simulator, and the Robot Operating System (ROS), which facilitates its evaluation through heterogeneous test cases, and the adoption of state estimators like Kalman filters. Our contributions illustrate the potential of our HyNN model in achieving precise indoor localisation and navigation for mobile robots in complex environments. The study follows, and proposes, a generalisable procedure applicable beyond the specific use case studied, making it adaptable to different scenarios and datasets.
Authors: Hao Zhu, Joschka Boedecker
Abstract: We introduce disciplined biconvex programming (DBCP), a modeling framework for specifying and solving biconvex optimization problems. Biconvex optimization problems arise in various applications, including machine learning, signal processing, computational science, and control. Solving a biconvex optimization problem in practice usually resolves to heuristic methods based on alternate convex search (ACS), which iteratively optimizes over one block of variables while keeping the other fixed, so that the resulting subproblems are convex and can be efficiently solved. However, designing and implementing an ACS solver for a specific biconvex optimization problem usually requires significant effort from the user, which can be tedious and error-prone. DBCP extends the principles of disciplined convex programming to biconvex problems, allowing users to specify biconvex optimization problems in a natural way based on a small number of syntax rules. The resulting problem can then be automatically split and transformed into convex subproblems, for which a customized ACS solver is then generated and applied. DBCP allows users to quickly experiment with different biconvex problem formulations, without expertise in convex optimization. We implement DBCP into the open source Python package dbcp, as an extension to the famous domain specific language CVXPY for convex optimization.
Authors: Konrad Staniszewski, Adrian {\L}a\'ncucki
Abstract: Serving large language models (LLMs) at scale necessitates efficient key-value (KV) cache management. KV caches can be reused across conversation turns via shared-prefix prompts that are common in iterative code editing and chat. However, stale caches consume scarce GPU memory, require offloading, or force recomputation. We present KVTC, a lightweight transform coder that compresses KV caches for compact on-GPU and off-GPU storage. Drawing on classical media compression, KVTC combines PCA-based feature decorrelation, adaptive quantization, and entropy coding. It requires only a brief initial calibration and leaves model parameters unchanged. By exploiting redundancies in KV caches, KVTC achieves up to 20$\times$ compression while maintaining reasoning and long-context accuracy, and 40$\times$ or higher for specific use cases. We test KVTC with Llama 3, Mistral NeMo, and R1-Qwen 2.5 models across benchmarks including AIME25, LiveCodeBench, GSM8K, MMLU, Qasper, RULER, and MATH-500. It consistently outperforms inference-time baselines such as token eviction, quantization, and SVD-based methods, while achieving higher compression ratios. These results support KVTC as a practical building block for memory-efficient LLM serving with reusable KV caches.
Authors: Yuetai Li, Huseyin A Inan, Xiang Yue, Wei-Ning Chen, Lukas Wutschitz, Janardhan Kulkarni, Radha Poovendran, Robert Sim, Saravan Rajmohan
Abstract: LLM agents excel in compact environments requiring deep reasoning but remain brittle when operating in broader, more complex contexts that demand robustness across diverse tools and schemas. Building bespoke environments for training is heavy, brittle, and limits progress. In this paper, we demonstrate that LLMs can simulate realistic environment feedback without access to actual testbed data or APIs. Inspired by this capability, we propose two frameworks: Simia-SFT, a pipeline that synthesizes SFT data by amplifying small seed sets into diverse trajectories in an environment-agnostic manner, and Simia-RL, a framework that enables RL training without real environment implementations through LLM-simulated feedback. Fine-tuning open models yields consistent improvements across multiple benchmarks, surpassing GPT-4o and approaching o4-mini on $\tau^2$-Bench. Together, Simia-SFT and Simia-RL enable scalable agent training without environment engineering, replacing heavy and brittle implementations with flexible LLM-based simulation.
Authors: Yang Cai, Constantinos Daskalakis, Haipeng Luo, Chen-Yu Wei, Weiqiang Zheng
Abstract: Learning and computation of equilibria are central problems in algorithmic game theory. In this work, we introduce proximal regret, a new notion of regret based on proximal operators that lies strictly between external and swap regret. When every player employs a no-proximal-regret algorithm in a general convex game, the empirical distribution of play converges to proximal correlated equilibria (PCE), a refinement of coarse correlated equilibria. Our framework unifies several emerging notions in online learning and game theory -- such as gradient equilibrium and semicoarse correlated equilibrium -- and introduces new ones. Our main result shows that the classic Online Gradient Descent (GD) algorithm achieves an optimal $O(\sqrt{T})$ bound on proximal regret, revealing that GD, without modification, minimizes a stronger regret notion than external regret. This provides a new explanation for the empirically superior performance of gradient descent in online learning and games. We further extend our analysis to Mirror Descent in the Bregman setting and to Optimistic Gradient Descent, which yields faster convergence in smooth convex games.
Authors: Yuting Ye, Christine Ho, Ci-Ren Jiang, Wayne Tai Lee, Haiyan Huang
Abstract: Hierarchical multi-label classification (HMC) has gained considerable attention in recent decades. A seminal line of HMC research addresses the problem in two stages: first, training individual classifiers for each class, then integrating these classifiers to provide a unified set of classification results across classes while respecting the given hierarchy. In this article, we focus on the less attended second-stage question while adhering to the given class hierarchy. This involves addressing a key challenge: how to manage the hierarchical constraint and account for statistical differences in the first-stage classifier scores across different classes to make classification decisions that are optimal under a justifiable criterion. To address this challenge, we introduce a new objective function, called CATCH, to ensure reasonable classification performance. To optimize this function, we propose a decision strategy built on a novel metric, the multidimensional Local Precision Rate (mLPR), which reflects the membership chance of an object in a class given all classifier scores and the class hierarchy. Particularly, we demonstrate that, under certain conditions, transforming the classifier scores into mLPRs and comparing mLPR values for all objects against all classes can, in theory, ensure the class hierarchy and maximize CATCH. In practice, we propose an algorithm HierRank to rank estimated mLPRs under the hierarchical constraint, leading to a ranking that maximizes an empirical version of CATCH. Our approach was evaluated on a synthetic dataset and two real datasets, exhibiting superior performance compared to several state-of-the-art methods in terms of improved decision accuracy.
Authors: Weijie Su
Abstract: Machine learning (ML) and artificial intelligence (AI) conferences including NeurIPS and ICML have experienced a significant decline in peer review quality in recent years. To address this growing challenge, we introduce the Isotonic Mechanism, a computationally efficient approach to enhancing the accuracy of noisy review scores by incorporating authors' private assessments of their submissions. Under this mechanism, authors with multiple submissions are required to rank their papers in descending order of perceived quality. Subsequently, the raw review scores are calibrated based on this ranking to produce adjusted scores. We prove that authors are incentivized to truthfully report their rankings because doing so maximizes their expected utility, modeled as an additive convex function over the adjusted scores. Moreover, the adjusted scores are shown to be more accurate than the raw scores, with improvements being particularly significant when the noise level is high and the author has many submissions -- a scenario increasingly prevalent at large-scale ML/AI conferences. We further investigate whether submission quality information beyond a simple ranking can be truthfully elicited from authors. We establish that a necessary condition for truthful elicitation is that the mechanism be based on pairwise comparisons of the author's submissions. This result underscores the optimality of the Isotonic Mechanism, as it elicits the most fine-grained truthful information among all mechanisms we consider. We then present several extensions, including a demonstration that the mechanism maintains truthfulness even when authors have only partial rather than complete information about their submission quality. Finally, we discuss future research directions, focusing on the practical implementation of the mechanism and the further development of a theoretical framework inspired by our mechanism.
Authors: Youngjae Min, Namhoon Cho, Navid Azizan
Abstract: While large machine learning models have shown remarkable performance in various domains, their training typically requires iterating for many passes over the training data. However, due to computational and memory constraints and potential privacy concerns, storing and accessing all the data is impractical in many real-world scenarios where the data arrives in a stream. In this paper, we investigate the problem of one-pass learning, in which a model is trained on sequentially arriving data without retraining on previous datapoints. Motivated by the demonstrated effectiveness of overparameterized models and the phenomenon of benign overfitting, we propose Orthogonal Recursive Fitting (ORFit), an algorithm for one-pass learning which seeks to perfectly fit each new datapoint while minimally altering the predictions on previous datapoints. ORFit updates the parameters in a direction orthogonal to past gradients, similar to orthogonal gradient descent (OGD) in continual learning. We show that, interestingly, ORFit's update leads to an operation similar to the recursive least-squares (RLS) algorithm in adaptive filtering but with significantly improved memory and computational efficiency, i.e., linear, instead of quadratic, in the number of parameters. To further reduce memory usage, we leverage the structure of the streaming data via an incremental principal component analysis (IPCA). We show that using the principal components is minimax optimal, i.e., it minimizes the worst-case forgetting of previous predictions for unknown future updates. Further, we prove that, for overparameterized linear models, the parameter vector obtained by ORFit matches what the standard multi-pass stochastic gradient descent (SGD) would converge to. Finally, we extend our results to the nonlinear setting for highly overparameterized models, relevant for deep learning.
Authors: Bryce McLaughlin, Jann Spiess
Abstract: When an algorithm provides risk assessments, we typically think of them as helpful inputs to human decisions, such as when risk scores are presented to judges or doctors. However, a decision-maker may react not only to the information provided by the algorithm. The decision-maker may also view the algorithmic recommendation as a default action, making it costly for them to deviate, such as when a judge is reluctant to overrule a high-risk assessment for a defendant or a doctor fears the consequences of deviating from recommended procedures. To address such unintended consequences of algorithmic assistance, we propose a model of joint human-machine decision-making. Within this model, we consider the effect and design of algorithmic recommendations when they affect choices not just by shifting beliefs, but also by altering preferences. We motivate this assumption from institutional factors, such as a desire to avoid audits, as well as from well-established models in behavioral science that predict loss aversion relative to a reference point. We show that recommendation-dependent preferences create inefficiencies where the decision-maker is overly responsive to the recommendation. As a remedy, we discuss algorithms that strategically withhold recommendations and show how they can improve the quality of final decisions. Concretely, we prove that an intuitive algorithm achieves minimax optimality by sending recommendations only when it is confident that their implementation would improve over an unassisted baseline decision.
Authors: Yu-Teng Li, Justin Lin, Jeffery Cheng, Pedro Pachuca
Abstract: Reinforcement Learning is a powerful tool to model decision-making processes. However, it relies on an exploration-exploitation trade-off that remains an open challenge for many tasks. In this work, we study neighboring state-based, model-free exploration led by the intuition that, for an early-stage agent, considering actions derived from a bounded region of nearby states may lead to better actions when exploring. We propose two algorithms that choose exploratory actions based on a survey of nearby states, and find that one of our methods, ${\rho}$-explore, consistently outperforms the Double DQN baseline in an discrete environment by 49% in terms of Eval Reward Return.
Authors: Shengming Li, Luping Liu, Runnan Li, Xu Tan
Abstract: Though denoising diffusion probabilistic models (DDPMs) have achieved remarkable generation results, the low sampling efficiency of DDPMs still limits further applications. Since DDPMs can be formulated as diffusion ordinary differential equations (ODEs), various fast sampling methods can be derived from solving diffusion ODEs. However, we notice that previous fast sampling methods with fixed analytical form are not able to robust with the various error patterns in the noise estimated from pretrained diffusion models. In this work, we construct an error-robust Adams solver (ERA-Solver), which utilizes the implicit Adams numerical method that consists of a predictor and a corrector. Different from the traditional predictor based on explicit Adams methods, we leverage a Lagrange interpolation function as the predictor, which is further enhanced with an error-robust strategy to adaptively select the Lagrange bases with lower errors in the estimated noise. The proposed solver can be directly applied to any pretrained diffusion models, without extra training. Experiments on Cifar10, CelebA, LSUN-Church, and ImageNet 64 x 64 (conditional) datasets demonstrate that our proposed ERA-Solver achieves 3.54, 5.06, 5.02, and 5.11 Frechet Inception Distance (FID) for image generation, with only 10 network evaluations.
Authors: Ruiwen Zhou, Minghuan Liu, Kan Ren, Xufang Luo, Weinan Zhang, Dongsheng Li
Abstract: Due to the nature of risk management in learning applicable policies, risk-sensitive reinforcement learning (RSRL) has been realized as an important direction. RSRL is usually achieved by learning risk-sensitive objectives characterized by various risk measures, under the framework of distributional reinforcement learning. However, it remains unclear if the distributional Bellman operator properly optimizes the RSRL objective in the sense of risk measures. In this paper, we prove that the existing RSRL methods do not achieve unbiased optimization and cannot guarantee optimality or even improvements regarding risk measures over accumulated return distributions. To remedy this issue, we further propose a novel algorithm, namely Trajectory Q-Learning (TQL), for RSRL problems with provable policy improvement towards the optimal policy. Based on our new learning architecture, we are free to introduce a general and practical implementation for different risk measures to learn disparate risk-sensitive policies. In the experiments, we verify the learnability of our algorithm and show how our method effectively achieves better performances toward risk-sensitive objectives.
Authors: Barathi Subramanian, Rathinaraja Jeyaraj, Rakhmonov Akhrorjon Akhmadjon Ugli
Abstract: Activation function is a pivotal component of deep learning, facilitating the extraction of intricate data patterns. While classical activation functions like ReLU and its variants are extensively utilized, their static nature and simplicity, despite being advantageous, often limit their effectiveness in specialized tasks. The trainable activation functions also struggle sometimes to adapt to the unique characteristics of the data. Addressing these limitations, we introduce a novel trainable activation function, adaptive piecewise approximated activation linear unit (APALU), to enhance the learning performance of deep learning across a broad range of tasks. It presents a unique set of features that enable it to maintain stability and efficiency in the learning process while adapting to complex data representations. Experiments reveal significant improvements over widely used activation functions for different tasks. In image classification, APALU increases MobileNet and GoogleNet accuracy by 0.37% and 0.04%, respectively, on the CIFAR10 dataset. In anomaly detection, it improves the average area under the curve of One-CLASS Deep SVDD by 0.8% on the MNIST dataset, 1.81% and 1.11% improvements with DifferNet, and knowledge distillation, respectively, on the MVTech dataset. Notably, APALU achieves 100% accuracy on a sign language recognition task with a limited dataset. For regression tasks, APALU enhances the performance of deep neural networks and recurrent neural networks on different datasets. These improvements highlight the robustness and adaptability of APALU across diverse deep-learning applications.
Authors: Barathi Subramanian, Rathinaraja Jeyaraj, Anand Paul
Abstract: Activation functions enable neural networks to learn complex representations by introducing non-linearities. While feedforward models commonly use rectified linear units, sequential models like recurrent neural networks, long short-term memory (LSTMs) and gated recurrent units (GRUs) still rely on Sigmoid and TanH activation functions. However, these classical activation functions often struggle to model sparse patterns when trained on small sequential datasets to effectively capture temporal dependencies. To address this limitation, we propose squared Sigmoid TanH (SST) activation specifically tailored to enhance the learning capability of sequential models under data constraints. SST applies mathematical squaring to amplify differences between strong and weak activations as signals propagate over time, facilitating improved gradient flow and information filtering. We evaluate SST-powered LSTMs and GRUs for diverse applications, such as sign language recognition, regression, and time-series classification tasks, where the dataset is limited. Our experiments demonstrate that SST models consistently outperform RNN-based models with baseline activations, exhibiting improved test accuracy.
Authors: Jiayi Huang, Sangwoo Park, Osvaldo Simeone
Abstract: The application of artificial intelligence (AI) models in fields such as engineering is limited by the known difficulty of quantifying the reliability of an AI's decision. A well-calibrated AI model must correctly report its accuracy on in-distribution (ID) inputs, while also enabling the detection of out-of-distribution (OOD) inputs. A conventional approach to improve calibration is the application of Bayesian ensembling. However, owing to computational limitations and model misspecification, practical ensembling strategies do not necessarily enhance calibration. This paper proposes an extension of variational inference (VI)-based Bayesian learning that integrates calibration regularization for improved ID performance, confidence minimization for OOD detection, and selective calibration to ensure a synergistic use of calibration regularization and confidence minimization. The scheme is constructed successively by first introducing calibration-regularized Bayesian learning (CBNN), then incorporating out-of-distribution confidence minimization (OCM) to yield CBNN-OCM, and finally integrating also selective calibration to produce selective CBNN-OCM (SCBNN-OCM). Selective calibration rejects inputs for which the calibration performance is expected to be insufficient. Numerical results illustrate the trade-offs between ID accuracy, ID calibration, and OOD calibration attained by both frequentist and Bayesian learning methods. Among the main conclusions, SCBNN-OCM is seen to achieve best ID and OOD performance as compared to existing state-of-the-art approaches at the cost of rejecting a sufficiently large number of inputs.
Authors: Xiongxiao Xu, Canyu Chen, Yueqing Liang, Baixiang Huang, Guangji Bai, Liang Zhao, Kai Shu
Abstract: Time series forecasting has made significant advances, including with Transformer-based models. The attention mechanism in Transformer effectively captures temporal dependencies by attending to all past inputs simultaneously. However, its quadratic complexity with respect to sequence length limits the scalability for long-range modeling. Recent state space models (SSMs) such as Mamba offer a promising alternative by achieving linear complexity without attention. Yet, Mamba compresses historical information into a fixed-size latent state, potentially causing information loss and limiting representational effectiveness. This raises a key research question: Can we design a hybrid Mamba-Transformer architecture that is both effective and efficient for time series forecasting? To address it, we adapt a hybrid Mamba-Transformer architecture Mambaformer, originally proposed for language modeling, to the time series domain. Preliminary experiments reveal that naively stacking Mamba and Transformer layers in Mambaformer is suboptimal for time series forecasting, due to an information interference problem. To mitigate this issue, we introduce a new time series decomposition strategy that separates time series into long-range patterns and short-range variations. Then we show that Mamba excels at capturing long-term structures, while Transformer is more effective at modeling short-term dynamics. Building on this insight, we propose State Space Transformer (SST), a multi-scale hybrid model with expert modules: a Mamba expert for long-range patterns and a Transformer expert for short-term variations. SST also employs a multi-scale patching mechanism to adaptively adjust time series resolution: low resolution for long-term patterns and high resolution for short-term variations. Experiments show that SST obtains SOTA performance with linear scalability. The code is at https://github.com/XiongxiaoXu/SST.
Authors: Fran\c{c}ois Rozet, G\'er\^ome Andry, Fran\c{c}ois Lanusse, Gilles Louppe
Abstract: Diffusion models recently proved to be remarkable priors for Bayesian inverse problems. However, training these models typically requires access to large amounts of clean data, which could prove difficult in some settings. In this work, we present DiEM, a novel method based on the expectation-maximization algorithm for training diffusion models from incomplete and noisy observations only. Unlike previous works, DiEM leads to proper diffusion models, which is crucial for downstream tasks. As part of our methods, we propose and motivate an improved posterior sampling scheme for unconditional diffusion models. We present empirical evidence supporting the effectiveness of our approach.
Authors: Sungho Jeon, Xinyue Ma, Kwang In Kim, Myeongjae Jeon
Abstract: Recent rehearsal-free continual learning (CL) methods guided by prompts achieve strong performance on vision tasks with non-stationary data but remain resource-intensive, hindering real-world edge deployment. We introduce resource-efficient prompting (REP), which improves the computational and memory efficiency of prompt-based rehearsal-free continual learning methods while minimizing accuracy trade-offs. Our approach employs swift prompt selection to refine input data using a carefully provisioned model and introduces adaptive token merging (AToM) and adaptive layer dropping (ALD) for efficient prompt updates. AToM and ALD selectively skip data and model layers while preserving task-specific features during the learning of new tasks. Extensive experiments on multiple image classification datasets demonstrate REP's superior resource efficiency over state-of-the-art rehearsal-free CL methods.
Authors: Tianheng Ling, Chao Qian, Gregor Schiele
Abstract: This paper presents the design of a hardware accelerator for Transformers, optimized for on-device time-series forecasting in AIoT systems. It integrates integer-only quantization and Quantization-Aware Training with optimized hardware designs to realize 6-bit and 4-bit quantized Transformer models, which achieved precision comparable to 8-bit quantized models from related research. Utilizing a complete implementation on an embedded FPGA (Xilinx Spartan-7 XC7S15), we examine the feasibility of deploying Transformer models on embedded IoT devices. This includes a thorough analysis of achievable precision, resource utilization, timing, power, and energy consumption for on-device inference. Our results indicate that while sufficient performance can be attained, the optimization process is not trivial. For instance, reducing the quantization bitwidth does not consistently result in decreased latency or energy consumption, underscoring the necessity of systematically exploring various optimization combinations. Compared to an 8-bit quantized Transformer model in related studies, our 4-bit quantized Transformer model increases test loss by only 0.63%, operates up to 132.33x faster, and consumes 48.19x less energy. Relevant source code is provided in the accompanying GitHub repository\footnote{https://github.com/tianheng-ling/TinyTransformer4TS}.
Authors: Liam Schramm, Abdeslam Boularias
Abstract: Diffusion models have seen tremendous success as generative architectures. Recently, they have been shown to be effective at modelling policies for offline reinforcement learning and imitation learning. We explore using diffusion as a model class for the successor state measure (SSM) of a policy. We find that enforcing the Bellman flow constraints leads to a simple Bellman update on the diffusion step distribution.
Authors: Angie Boggust, Hyemin Bang, Hendrik Strobelt, Arvind Satyanarayan
Abstract: While interpretability methods identify a model's learned concepts, they overlook the relationships between concepts that make up its abstractions and inform its ability to generalize to new data. To assess whether models' have learned human-aligned abstractions, we introduce abstraction alignment, a methodology to compare model behavior against formal human knowledge. Abstraction alignment externalizes domain-specific human knowledge as an abstraction graph, a set of pertinent concepts spanning levels of abstraction. Using the abstraction graph as a ground truth, abstraction alignment measures the alignment of a model's behavior by determining how much of its uncertainty is accounted for by the human abstractions. By aggregating abstraction alignment across entire datasets, users can test alignment hypotheses, such as which human concepts the model has learned and where misalignments recur. In evaluations with experts, abstraction alignment differentiates seemingly similar errors, improves the verbosity of existing model-quality metrics, and uncovers improvements to current human abstractions.
Authors: Mark Towers, Ariel Kwiatkowski, Jordan Terry, John U. Balis, Gianluca De Cola, Tristan Deleu, Manuel Goul\~ao, Andreas Kallinteris, Markus Krimmel, Arjun KG, Rodrigo Perez-Vicente, Andrea Pierr\'e, Sander Schulhoff, Jun Jet Tai, Hannah Tan, Omar G. Younis
Abstract: Reinforcement Learning (RL) is a continuously growing field that has the potential to revolutionize many areas of artificial intelligence. However, despite its promise, RL research is often hindered by the lack of standardization in environment and algorithm implementations. This makes it difficult for researchers to compare and build upon each other's work, slowing down progress in the field. Gymnasium is an open-source library that provides a standard API for RL environments, aiming to tackle this issue. Gymnasium's main feature is a set of abstractions that allow for wide interoperability between environments and training algorithms, making it easier for researchers to develop and test RL algorithms. In addition, Gymnasium provides a collection of easy-to-use environments, tools for easily customizing environments, and tools to ensure the reproducibility and robustness of RL research. Through this unified framework, Gymnasium significantly streamlines the process of developing and testing RL algorithms, enabling researchers to focus more on innovation and less on implementation details. By providing a standardized platform for RL research, Gymnasium helps to drive forward the field of reinforcement learning and unlock its full potential. Gymnasium is available online at https://github.com/Farama-Foundation/Gymnasium
Authors: Wissal Hamhoum, Soumaya Cherkaoui
Abstract: Malicious attacks on vehicular networks pose a serious threat to road safety as well as communication reliability. A major source of these threats stems from misbehaving vehicles within the network. To address this challenge, we propose a Large Language Model (LLM)-empowered Misbehavior Detection System (MDS) within an edge-cloud detection framework. Specifically, we fine-tune Mistral-7B, a compact and high-performing LLM, to detect misbehavior based on Basic Safety Messages (BSM) sequences as the edge component for real-time detection, while a larger LLM deployed in the cloud validates and reinforces the edge model's detection through a more comprehensive analysis. By updating only 0.012% of the model parameters, our model, which we named MistralBSM, achieves 98% accuracy in binary classification and 96% in multiclass classification on a selected set of attacks from VeReMi dataset, outperforming LLAMA2-7B and RoBERTa. Our results validate the potential of LLMs in MDS, showing a significant promise in strengthening vehicular network security to better ensure the safety of road users.
Authors: Zhipeng Yin, Sribala Vidyadhari Chinta, Zichong Wang, Matthew Gonzalez, Wenbin Zhang
Abstract: The integration of AI in education holds immense potential for personalizing learning experiences and transforming instructional practices. However, AI systems can inadvertently encode and amplify biases present in educational data, leading to unfair or discriminatory outcomes. As researchers have sought to understand and mitigate these biases, a growing body of work has emerged examining fairness in educational AI. These studies, though expanding rapidly, remain fragmented due to differing assumptions, methodologies, and application contexts. Moreover, existing surveys either focus on algorithmic fairness without an educational setting or emphasize educational methods while overlooking fairness. To this end, this survey provides a comprehensive systematic review of algorithmic fairness within educational AI, explicitly bridging the gap between technical fairness research and educational applications. We integrate multiple dimensions, including bias sources, fairness definitions, mitigation strategies, evaluation resources, and ethical considerations, into a harmonized, education-centered framework. In addition, we explicitly examine practical challenges such as censored or partially observed learning outcomes and the persistent difficulty in quantifying and managing the trade-off between fairness and predictive utility, enhancing the applicability of fairness frameworks to real-world educational AI systems. Finally, we outline an emerging pathway toward fair AI-driven education and by situating these technologies and practical insights within broader educational and ethical contexts, this review establishes a comprehensive foundation for advancing fairness, accountability, and inclusivity in the field of AI education.
Authors: Jonathan Light, Yuanzhe Liu, Ziniu Hu
Abstract: Offline reinforcement learning often requires a quality dataset that we can train a policy on. However, in many situations, it is not possible to get such a dataset, nor is it easy to train a policy to perform well in the actual environment given the offline data. We propose using data distillation to train and distill a better dataset which can then be used for training a better policy model. We show that our method is able to synthesize a dataset where a model trained on it achieves similar performance to a model trained on the full dataset or a model trained using percentile behavioral cloning. Our project site is available at https://datasetdistillation4rl.github.io . We also provide our implementation at https://github.com/ggflow123/DDRL .
URLs: https://datasetdistillation4rl.github.io, https://github.com/ggflow123/DDRL
Authors: Joshua Hang Sai Ip, Georgios Makrygiorgos, Ali Mesbah
Abstract: Deep neural networks are increasingly used as an effective parameterization of control policies in various learning-based control paradigms. For continuous-time optimal control problems (OCPs), which are central to many decision-making tasks, control policy learning can be cast as a neural ordinary differential equation (NODE) problem wherein state and control constraints are naturally accommodated. This paper presents a NODE approach to solving continuous-time OCPs for the case of stabilizing a known constrained nonlinear system around a target state. The approach, termed Lyapunov-NODE control (L-NODEC), uses a novel Lyapunov loss formulation that incorporates an exponentially-stabilizing control Lyapunov function to learn a state-feedback neural control policy, bridging the gap of solving continuous-time OCPs via NODEs with stability guarantees. The proposed Lyapunov loss allows L-NODEC to guarantee exponential stability of the controlled system, as well as its adversarial robustness to perturbations to the initial state. The performance of L-NODEC is illustrated in two problems, including a dose delivery problem in plasma medicine. In both cases, L-NODEC effectively stabilizes the controlled system around the target state despite perturbations to the initial state and reduces the inference time necessary to reach the target.
Authors: Akhil Premkumar
Abstract: We explore the connection between deep learning and information theory through the paradigm of diffusion models. A diffusion model converts noise into structured data by reinstating, imperfectly, information that is erased when data was diffused to noise. This information is stored in a neural network during training. We quantify this information by introducing a measure called neural entropy, which is related to the total entropy produced by diffusion. Neural entropy is a function of not just the data distribution, but also the diffusive process itself. Measurements of neural entropy on a few simple image diffusion models reveal that they are extremely efficient at compressing large ensembles of structured data.
Authors: Mohamed Dhouioui, Jonathan Barnoud, Rhoslyn Roebuck Williams, Harry J. Stroud, Phil Bates, David R. Glowacki
Abstract: Molecular dynamics (MD) simulations are a crucial computational tool for researchers to understand and engineer molecular structure and function in areas such as drug discovery, protein engineering, and material design. Despite their utility, MD simulations are expensive, owing to the high dimensionality of molecular systems. Interactive molecular dynamics in virtual reality (iMD-VR) has recently emerged as a "human-in-the-loop" strategy for efficiently navigating hyper-dimensional molecular systems. By providing an immersive 3D environment that enables visualization and manipulation of real-time molecular simulations running on high-performance computing architectures, iMD-VR enables researchers to reach out and guide molecular conformational dynamics, in order to efficiently explore complex, high-dimensional molecular systems. Moreover, iMD-VR simulations generate rich datasets that capture human experts' spatial insight regarding molecular structure and function. This paper explores the use of researcher-generated iMD-VR datasets to train AI agents via imitation learning (IL). IL enables agents to mimic complex behaviours from expert demonstrations, circumventing the need for explicit programming or intricate reward design. In this article, we review IL across robotics and Multi-agents systems domains which are comparable to iMD-VR, and discuss how iMD-VR recordings could be used to train IL models to interact with MD simulations. We then illustrate the applications of these ideas through a proof-of-principle study where iMD-VR data was used to train a CNN network on a simple molecular manipulation task; namely, threading a small molecule through a nanotube pore. Finally, we outline future research directions and potential challenges of using AI agents to augment human expertise in navigating vast molecular conformational spaces.
Authors: Davy Darankoum, Manon Villalba, Clelia Allioux, Baptiste Caraballo, Carine Dumont, Eloise Gronlier, Corinne Roucard, Yann Roche, Chloe Habermacher, Sergei Grudinin, Julien Volle
Abstract: Epilepsy represents the most prevalent neurological disease in the world. One-third of people suffering from mesial temporal lobe epilepsy (MTLE) exhibit drug resistance, urging the need to develop new treatments. A key part in anti-seizure medication (ASM) development is the capability of detecting and quantifying epileptic seizures occurring in electroencephalogram (EEG) signals, which is crucial for treatment efficacy evaluation. In this study, we introduced a seizure detection pipeline based on deep learning models applied to raw EEG signals. This pipeline integrates: a new pre-processing technique which segments continuous raw EEG signals without prior distinction between seizure and seizure-free activities; a post-processing algorithm developed to reassemble EEG segments and allow the identification of seizures start/end; and finally, a new evaluation procedure based on a strict seizure events comparison between predicted and real labels. Models training have been performed using a data splitting strategy which addresses the potential for data leakage. We demonstrated the fundamental differences between a seizure classification and a seizure detection task and showed the differences in performance between the two tasks. Finally, we demonstrated the generalization capabilities across species of our best architecture, combining a Convolutional Neural Network and a Transformer encoder. The model was trained on animal EEGs and tested on human EEGs with a F1-score of 93% on a balanced Bonn dataset.
Authors: Jianqiao Mao, Max A. Little
Abstract: A major limitation of machine learning (ML) prediction models is that they recover associational, rather than causal, predictive relationships between variables. In high-stakes automation applications of ML this is problematic, as the model often learns spurious, non-causal associations. This paper proposes mechanism learning, a simple method which uses causally weighted Gaussian Mixture Models (CW-GMMs) to deconfound observational data such that any appropriate ML model is forced to learn predictive relationships between effects and their causes (reverse causal inference), despite the potential presence of multiple unknown and unmeasured confounding. Effect variables can be very high-dimensional, and the predictive relationship nonlinear, as is common in ML applications. This novel method is widely applicable, the only requirement is the existence of a set of mechanism variables mediating the cause (prediction target) and effect (feature data), which is independent of the (unmeasured) confounding variables. We test our method on fully synthetic, semi-synthetic and real-world datasets, demonstrating that it can discover reliable, unbiased, causal ML predictors where by contrast, the same ML predictor trained naively using classical supervised learning on the original observational data, is heavily biased by spurious associations. We provide code to implement the results in the paper, online.
Authors: Yingxu Wang, Nan Yin, Mingyan Xiao, Xinhao Yi, Siwei Liu, Shangsong Liang
Abstract: Graph Neural Networks (GNNs) with equivariant properties have achieved significant success in modeling complex dynamic systems and molecular properties. However, their expressiveness ability is limited by: (1) Existing methods often overlook the over-smoothing issue caused by traditional GNN models, as well as the gradient explosion or vanishing problems in deep GNNs. (2) Most models operate on first-order information, neglecting that the real world often consists of second-order systems, which further limits the model's representation capabilities. To address these issues, we propose the \textbf{Du}al \textbf{S}econd-order \textbf{E}quivariant \textbf{G}raph \textbf{O}rdinary Differential Equation (\method{}) for equivariant representation. Specifically, \method{} apply the dual second-order equivariant graph ordinary differential equations (Graph ODEs) on graph embeddings and node coordinates, simultaneously. Theoretically, we first prove that \method{} maintains the equivariant property. Furthermore, we provide theoretical insights showing that \method{} effectively alleviates the over-smoothing problem in both feature representation and coordinate update. Additionally, we demonstrate that the proposed \method{} mitigates the exploding and vanishing gradients problem, facilitating the training of deep multi-layer GNNs. Extensive experiments on benchmark datasets validate the superiority of the proposed \method{} compared to baselines.
Authors: Irina Bara\v{s}in, Bla\v{z} Bertalani\v{c}, Mihael Mohor\v{c}i\v{c}, Carolina Fortuna
Abstract: Time series classification is a relevant step supporting decision-making processes in various domains, and deep neural models have shown promising performance in this respect. Despite significant advancements in deep learning, the theoretical understanding of how and why complex architectures function remains limited, prompting the need for more interpretable models. Recently, the Kolmogorov-Arnold Networks (KANs) have been proposed as a more interpretable alternative to deep learning. While KAN-related research is significantly rising, to date, the study of KAN architectures for time series classification has been limited. In this paper, we aim to conduct a comprehensive and robust exploration of the KAN architecture for time series classification utilising 117 datasets from UCR benchmark archive, from multiple different domains. More specifically, we investigate a) the transferability of reference architectures designed for regression to classification tasks, b) identifying the hyperparameter and implementation configurations for an architecture that best generalizes across 117 datasets, c) the associated complexity trade-offs and d) evaluate KANs interpretability. Our results demonstrate that (1) the Efficient KAN outperforms MLPs in both performance and training times, showcasing its suitability for classification tasks. (2) Efficient KAN exhibits greater stability than the original KAN across grid sizes, depths, and layer configurations, especially when lower learning rates are employed. (3) KAN achieves competitive accuracy compared to state-of-the-art models such as HIVE-COTE2 and InceptionTime, while maintaining smaller architectures and faster training times, highlighting its favorable balance of performance and transparency. (4) The interpretability of the KAN model, as confirmed by SHAP analysis, reinforces its capacity for transparent decision-making.
Authors: A. Feder Cooper, Christopher A. Choquette-Choo, Miranda Bogen, Kevin Klyman, Matthew Jagielski, Katja Filippova, Ken Liu, Alexandra Chouldechova, Jamie Hayes, Yangsibo Huang, Eleni Triantafillou, Peter Kairouz, Nicole Elyse Mitchell, Niloofar Mireshghallah, Abigail Z. Jacobs, James Grimmelmann, Vitaly Shmatikov, Christopher De Sa, Ilia Shumailov, Andreas Terzis, Solon Barocas, Jennifer Wortman Vaughan, Danah Boyd, Yejin Choi, Sanmi Koyejo, Fernando Delgado, Percy Liang, Daniel E. Ho, Pamela Samuelson, Miles Brundage, David Bau, Seth Neel, Hanna Wallach, Amy B. Cyphert, Mark A. Lemley, Nicolas Papernot, Katherine Lee
Abstract: "Machine unlearning" is a popular proposed solution for mitigating the existence of content in an AI model that is problematic for legal or moral reasons, including privacy, copyright, safety, and more. For example, unlearning is often invoked as a solution for removing the effects of specific information from a generative-AI model's parameters, e.g., a particular individual's personal data or the inclusion of copyrighted content in the model's training data. Unlearning is also proposed as a way to prevent a model from generating targeted types of information in its outputs, e.g., generations that closely resemble a particular individual's data or reflect the concept of "Spiderman." Both of these goals--the targeted removal of information from a model and the targeted suppression of information from a model's outputs--present various technical and substantive challenges. We provide a framework for ML researchers and policymakers to think rigorously about these challenges, identifying several mismatches between the goals of unlearning and feasible implementations. These mismatches explain why unlearning is not a general-purpose solution for circumscribing generative-AI model behavior in service of broader positive impact.
Authors: Menglin Yang, Jialin Chen, Jinkai Tao, Yifei Zhang, Jiahong Liu, Jiasheng Zhang, Qiyao Ma, Harshit Verma, Regina Zhang, Min Zhou, Irwin King, Rex Ying
Abstract: The rapid advancement of foundation modelslarge-scale neural networks trained on diverse, extensive datasetshas revolutionized artificial intelligence, enabling unprecedented advancements across domains such as natural language processing, computer vision, and scientific discovery. However, the substantial parameter count of these models, often reaching billions or trillions, poses significant challenges in adapting them to specific downstream tasks. Low-Rank Adaptation (LoRA) has emerged as a highly promising approach for mitigating these challenges, offering a parameter-efficient mechanism to fine-tune foundation models with minimal computational overhead. This survey provides the first comprehensive review of LoRA techniques beyond large Language Models to general foundation models, including recent techniques foundations, emerging frontiers and applications of low-rank adaptation across multiple domains. Finally, this survey discusses key challenges and future research directions in theoretical understanding, scalability, and robustness. This survey serves as a valuable resource for researchers and practitioners working with efficient foundation model adaptation.
Authors: Yasmin Salehi, Dennis Giannacopoulos
Abstract: Graph clustering plays a crucial role in graph representation learning but often faces challenges in achieving feature-space diversity. While Deep Modularity Networks (DMoN) leverage modularity maximization and collapse regularization to ensure structural separation, they lack explicit mechanisms for feature-space separation, assignment dispersion, and assignment-confidence control. We address this limitation by proposing Deep Modularity Networks with Diversity-Preserving Regularization (DMoN-DPR), which introduces three novel regularization terms: distance-based for inter-cluster separation, variance-based for per-cluster assignment dispersion, and an assignment-entropy penalty with a small positive weight, encouraging more confident assignments gradually. Our method significantly enhances label-based clustering metrics on feature-rich benchmark datasets (paired two-tailed t-test, $p\leq0.05$), demonstrating the effectiveness of incorporating diversity-preserving regularizations in creating meaningful and interpretable clusters.
Authors: Yikun Hou, Suvrit Sra, Alp Yurtsever
Abstract: Gradient descent for matrix factorization exhibits an implicit bias toward approximately low-rank solutions. While existing theories often assume the boundedness of iterates, empirically the bias persists even with unbounded sequences. This reflects a dynamic where factors develop low-rank structure while their magnitudes increase, tending to align with certain directions. To capture this behavior in a stable way, we introduce a new factorization model: $X\approx UDV^\top$, where $U$ and $V$ are constrained within norm balls, while $D$ is a diagonal factor allowing the model to span the entire search space. Experiments show that this model consistently exhibits a strong implicit bias, yielding truly (rather than approximately) low-rank solutions. Extending the idea to neural networks, we introduce a new model featuring constrained layers and diagonal components that achieves competitive performance on various regression and classification tasks while producing lightweight, low-rank representations.
Authors: Yunyang Li, Lin Huang, Zhihao Ding, Chu Wang, Xinran Wei, Han Yang, Zun Wang, Chang Liu, Yu Shi, Peiran Jin, Tao Qin, Mark Gerstein, Jia Zhang
Abstract: Equivariant Graph Neural Networks (EGNNs) have demonstrated significant success in modeling microscale systems, including those in chemistry, biology and materials science. However, EGNNs face substantial computational challenges due to the high cost of constructing edge features via spherical tensor products, making them impractical for large-scale systems. To address this limitation, we introduce E2Former, an equivariant and efficient transformer architecture that incorporates the Wigner $6j$ convolution (Wigner $6j$ Conv). By shifting the computational burden from edges to nodes, the Wigner $6j$ Conv reduces the complexity from $O(|\mathcal{E}|)$ to $ O(| \mathcal{V}|)$ while preserving both the model's expressive power and rotational equivariance. We show that this approach achieves a 7x-30x speedup compared to conventional $\mathrm{SO}(3)$ convolutions. Furthermore, our empirical results demonstrate that the derived E2Former mitigates the computational challenges of existing approaches without compromising the ability to capture detailed geometric information. This development could suggest a promising direction for scalable and efficient molecular modeling.
Authors: Fan Wang, Pengtao Shao, Yiming Zhang, Bo Yu, Shaoshan Liu, Ning Ding, Yang Cao, Yu Kang, Haifeng Wang
Abstract: In-Context Reinforcement Learning (ICRL) enables agents to learn automatically and on-the-fly from their interactive experiences. However, a major challenge in scaling up ICRL is the lack of scalable task collections. To address this, we propose the procedurally generated tabular Markov Decision Processes, named AnyMDP. Through a carefully designed randomization process, AnyMDP is capable of generating high-quality tasks on a large scale while maintaining relatively low structural biases. To facilitate efficient meta-training at scale, we further introduce decoupled policy distillation and induce prior information in the ICRL framework. Our results demonstrate that, with a sufficiently large scale of AnyMDP tasks, the proposed model can generalize to tasks that were not considered in the training set through versatile in-context learning paradigms. The scalable task set provided by AnyMDP also enables a more thorough empirical investigation of the relationship between data distribution and ICRL performance. We further show that the generalization of ICRL potentially comes at the cost of increased task diversity and longer adaptation periods. This finding carries critical implications for scaling robust ICRL capabilities, highlighting the necessity of diverse and extensive task design, and prioritizing asymptotic performance over few-shot adaptation.
Authors: Qitao Tan, Jun Liu, Zheng Zhan, Caiwei Ding, Yanzhi Wang, Xiaolong Ma, Jaewoo Lee, Jin Lu, Geng Yuan
Abstract: Large language models (LLMs) excel across various tasks, but standard first-order (FO) fine-tuning demands considerable memory, significantly limiting real-world deployment. Recently, zeroth-order (ZO) optimization stood out as a promising memory-efficient training paradigm, avoiding backward passes and relying solely on forward passes for gradient estimation, making it attractive for resource-constrained scenarios. However, ZO method lags far behind FO method in both convergence speed and accuracy. To bridge the gap, we introduce a novel layer-wise divergence analysis that uncovers the distinct update pattern of FO and ZO optimization. Aiming to resemble the learning capacity of FO method from the findings, we propose Divergence-driven Zeroth-Order (DiZO) optimization. DiZO conducts divergence-driven layer adaptation by incorporating projections to ZO updates, generating diverse-magnitude updates precisely scaled to layer-wise individual optimization needs. Our results demonstrate that DiZO significantly reduces the needed iterations for convergence without sacrificing throughput, cutting training GPU hours by up to 48\% on various datasets. Moreover, DiZO consistently outperforms the representative ZO baselines in fine-tuning RoBERTa-large, OPT-series, and Llama-series on downstream tasks and, in some cases, even surpasses memory-intensive FO fine-tuning. Our code is released at https://github.com/Skilteee/DiZO.
Authors: Yuankai Luo, Lei Shi, Xiao-Ming Wu
Abstract: Message-passing Graph Neural Networks (GNNs) are often criticized for their limited expressiveness, issues like over-smoothing and over-squashing, and challenges in capturing long-range dependencies. Conversely, Graph Transformers (GTs) are regarded as superior due to their employment of global attention mechanisms, which potentially mitigate these challenges. Literature frequently suggests that GTs outperform GNNs in graph-level tasks, especially for graph classification and regression on small molecular graphs. In this study, we explore the untapped potential of GNNs through an enhanced framework, GNN+, which integrates six widely used techniques: edge feature integration, normalization, dropout, residual connections, feed-forward networks, and positional encoding, to effectively tackle graph-level tasks. We conduct a systematic re-evaluation of three classic GNNs (GCN, GIN, and GatedGCN) enhanced by the GNN+ framework across 14 well-known graph-level datasets. Our results reveal that, contrary to prevailing beliefs, these classic GNNs consistently match or surpass the performance of GTs, securing top-three rankings across all datasets and achieving first place in eight. Furthermore, they demonstrate greater efficiency, running several times faster than GTs on many datasets. This highlights the potential of simple GNN architectures, challenging the notion that complex mechanisms in GTs are essential for superior graph-level performance. Our source code is available at https://github.com/LUOyk1999/GNNPlus.
Authors: Luke Snow, Vikram Krishnamurthy
Abstract: A neural stochastic differential equation (SDE) is an SDE with drift and diffusion terms parametrized by neural networks. The training procedure for neural SDEs consists of optimizing the SDE vector field (neural network) parameters to minimize the expected value of an objective functional on infinite-dimensional path-space. Existing training techniques focus on methods to efficiently compute path-wise gradients of the objective functional with respect to these parameters, then pair this with Monte-Carlo simulation to estimate the gradient expectation. In this work we introduce a novel training technique which bypasses and improves upon this Monte-Carlo simulation; we extend results in the theory of Wiener space cubature to approximate the expected objective functional value by a weighted sum of functional evaluations of deterministic ODE solutions. Our main mathematical contribution enabling this approximation is an extension of cubature bounds to the setting of Lipschitz-nonlinear functionals acting on path-space. Our resulting constructive algorithm allows for more computationally efficient training along several lines. First, it circumvents Brownian motion simulation and enables the use of efficient parallel ODE solvers, thus decreasing the complexity of path-functional evaluation. Furthermore, and more surprisingly, we show that the number of paths required to achieve a given (expected loss functional oracle value) approximation can be reduced in this deterministic cubature regime. Specifically, we show that under reasonable regularity assumptions we can observe a O(1/n) convergence rate, where n is the number of path evaluations; in contrast with the standard O(1/sqrt(n)) rate of naive Monte-Carlo or the O(log(n)^d /n) rate of quasi-Monte-Carlo.
Authors: Xinyu Zhang, Zewei Zhou, Zhaoyi Wang, Yangjie Ji, Yanjun Huang, Hong Chen
Abstract: Vehicle-to-everything technologies (V2X) have become an ideal paradigm to extend the perception range and see through the occlusion. Exiting efforts focus on single-frame cooperative perception, however, how to capture the temporal cue between frames with V2X to facilitate the prediction task even the planning task is still underexplored. In this paper, we introduce the Co-MTP, a general cooperative trajectory prediction framework with multi-temporal fusion for autonomous driving, which leverages the V2X system to fully capture the interaction among agents in both history and future domains to benefit the planning. In the history domain, V2X can complement the incomplete history trajectory in single-vehicle perception, and we design a heterogeneous graph transformer to learn the fusion of the history feature from multiple agents and capture the history interaction. Moreover, the goal of prediction is to support future planning. Thus, in the future domain, V2X can provide the prediction results of surrounding objects, and we further extend the graph transformer to capture the future interaction among the ego planning and the other vehicles' intentions and obtain the final future scenario state under a certain planning action. We evaluate the Co-MTP framework on the real-world dataset V2X-Seq, and the results show that Co-MTP achieves state-of-the-art performance and that both history and future fusion can greatly benefit prediction.
Authors: Ratun Rahman, Pablo Moriano, Samee U. Khan, Dinh C. Nguyen
Abstract: Electric load forecasting is essential for power management and stability in smart grids. This is mainly achieved via advanced metering infrastructure, where smart meters (SMs) record household energy data. Traditional machine learning (ML) methods are often employed for load forecasting, but require data sharing, which raises data privacy concerns. Federated learning (FL) can address this issue by running distributed ML models at local SMs without data exchange. However, current FL-based approaches struggle to achieve efficient load forecasting due to imbalanced data distribution across heterogeneous SMs. This paper presents a novel personalized federated learning (PFL) method for high-quality load forecasting in metering networks. A meta-learning-based strategy is developed to address data heterogeneity at local SMs in the collaborative training of local load forecasting models. Moreover, to minimize the load forecasting delays in our PFL model, we study a new latency optimization problem based on optimal resource allocation at SMs. A theoretical convergence analysis is also conducted to provide insights into FL design for federated load forecasting. Extensive simulations from real-world datasets show that our method outperforms existing approaches regarding better load forecasting and reduced operational latency costs.
Authors: Wenda Chu, Zihui Wu, Yifan Chen, Yang Song, Yisong Yue
Abstract: We study the problem of posterior sampling in discrete-state spaces using discrete diffusion models. While posterior sampling methods for continuous diffusion models have achieved remarkable progress, analogous methods for discrete diffusion models remain challenging. In this work, we introduce a principled plug-and-play discrete diffusion posterior sampling algorithm based on split Gibbs sampling, which we call SGDD. Our algorithm enables reward-guided generation and solving inverse problems in discrete-state spaces. We demonstrate the convergence of SGDD to the target posterior distribution and verify this through controlled experiments on synthetic benchmarks. Our method enjoys state-of-the-art posterior sampling performance on a range of benchmarks for discrete data, including DNA sequence design, discrete image inverse problems, and music infilling, achieving more than 30% improved performance compared to existing baselines. Our code is available at https://github.com/chuwd19/Split-Gibbs-Discrete-Diffusion-Posterior-Sampling.
URLs: https://github.com/chuwd19/Split-Gibbs-Discrete-Diffusion-Posterior-Sampling.
Authors: Bo-Yi Liu, Zhi-Xuan Liu, Kuan Lun Chen, Shih-Yu Tsai, Jie Gao, Hao-Tsung Yang
Abstract: Deep learning models are widely used in decision-making and recommendation systems, where they typically rely on the assumption of a static data distribution between training and deployment. However, real-world deployment environments often violate this assumption. Users who receive negative outcomes may adapt their features to meet model criteria, i.e., recourse action. These adaptive behaviors create shifts in the data distribution and when models are retrained on this shifted data, a feedback loop emerges: user behavior influences the model, and the updated model in turn reshapes future user behavior. Despite its importance, this bidirectional interaction between users and models has received limited attention. In this work, we develop a general framework to model user strategic behaviors and their interactions with decision-making systems under resource constraints and competitive dynamics. Both the theoretical and empirical analyses show that user recourse behavior tends to push logistic and MLP models toward increasingly higher decision standards, resulting in higher recourse costs and less reliable recourse actions over time. To mitigate these challenges, we propose two methods--Fair-top-k and Dynamic Continual Learning (DCL)--which significantly reduce recourse cost and improve model robustness. Our findings draw connections to economic theories, highlighting how algorithmic decision-making can unintentionally reinforce a higher standard and generate endogenous barriers to entry.
Authors: Satyajeet Sahoo, Jhareswar Maiti
Abstract: An important aspect of text mining involves information retrieval in form of discovery of semantic themes (topics) from documents using topic modelling. While generative topic models like Latent Dirichlet Allocation (LDA) or Latent Semantic Analysis (LSA) elegantly model topics as probability distributions and are useful in identifying latent topics from large document corpora with minimal supervision, they suffer from difficulty in topic interpretability and reduced performance in shorter texts. Here we propose a novel Multivariate Gaussian Topic Model (MGTM). In this approach topics are presented as Multivariate Gaussian Distributions and documents as Gaussian Mixture Models. Applying EM algorithm on a document corpus, the various constituent Multivariate Gaussian distributions corresponding to the latent topics and their respective parameters are identified. Analysis of the parameters of each distribution helps identify the respective topic keywords, and from these key-words topic annotations are carried out. This approach is applied on 20 newsgroups dataset to demonstrate the interpretability benefits vis-`a-vis 4 other benchmark models. The effectiveness of this model in capturing the semantic theme of the topics with high interpretability is examined by calculating the topic coherence and comparing the coherence values with benchmark models. This model achieves a highest mean topic coherence (0.7) and median topic coherence (0.76) vis-`a-vis the benchmark models, demonstrating high effectiveness in identifying interpretable, semantically coherent topics.
Authors: Renjie Liu, Haitian Jiang, Xiao Yan, Bo Tang, Jinyang Li
Abstract: GraphRAG enhances large language models (LLMs) to generate quality answers for user questions by retrieving related facts from external knowledge graphs. However, current GraphRAG methods are primarily evaluated on and overly tailored for knowledge graph question answering (KGQA) benchmarks, which are biased towards a few specific question patterns and do not reflect the diversity of real-world questions. To better evaluate GraphRAG methods, we propose a complete four-class taxonomy to categorize the basic patterns of knowledge graph questions and use it to create PolyBench, a new GraphRAG benchmark encompassing a comprehensive set of graph questions. With the new benchmark, we find that existing GraphRAG methods fall short in effectiveness (i.e., quality of the generated answers) and/or efficiency (i.e., response time or token usage) because they adopt either a fixed graph traversal strategy or free-form exploration by LLMs for fact retrieval. However, different question patterns require distinct graph traversal strategies and context formation. To facilitate better retrieval, we propose PolyG, an adaptive GraphRAG approach by decomposing and categorizing the questions according to our proposed question taxonomy. Built on top of a unified interface and execution engine, PolyG dynamically prompts an LLM to generate a graph database query to retrieve the context for each decomposed basic question. Compared with SOTA GraphRAG methods, PolyG achieves a higher win rate in generation quality and has a low response latency and token cost. Our code and benchmark are open-source at https://github.com/Liu-rj/PolyG.
Authors: Peizhi Niu, Yu-Hsiang Wang, Vishal Rana, Chetan Rupakheti, Abhishek Pandey, Olgica Milenkovic
Abstract: We introduce a new graph diffusion model for small molecule generation, DMol, which outperforms the state-of-the-art DiGress model in terms of validity by roughly 1.5% across all benchmarking datasets while reducing the number of diffusion steps by at least 10-fold, and the running time to roughly one half. The performance improvements are a result of a careful change in the objective function and a graph noise scheduling approach which, at each diffusion step, allows one to only change a subset of nodes of varying size in the molecule graph. Another relevant property of the method is that it can be easily combined with junction-tree-like graph representations that arise by compressing a collection of relevant ring structures into supernodes. Unlike classical junction-tree techniques that involve VAEs and require complicated reconstruction steps, compressed DMol directly performs graph diffusion on a graph that compresses only a carefully selected set of frequent carbon rings into supernodes, which results in straightforward sample generation. This compressed DMol method offers additional validity improvements over generic DMol of roughly 2%, increases the novelty of the method, and further improves the running time due to reductions in the graph size.
Authors: Jesse C. Cresswell
Abstract: Trustworthy AI encompasses many aspirational aspects for aligning AI systems with human values, including fairness, privacy, robustness, explainability, and uncertainty quantification. Ultimately the goal of Trustworthy AI research is to achieve all aspects simultaneously. However, efforts to enhance one aspect often introduce unintended trade-offs that negatively impact others. In this position paper, we review notable approaches to these five aspects and systematically consider every pair, detailing the negative interactions that can arise. For example, applying differential privacy to model training can amplify biases, undermining fairness. Drawing on these findings, we take the position that current research practices of improving one or two aspects in isolation are insufficient. Instead, research on Trustworthy AI must account for interactions between aspects and adopt a holistic view across all relevant axes at once. To illustrate our perspective, we provide guidance on how practitioners can work towards integrated trust, examples of how interactions affect the financial industry, and alternative views.
Authors: Wei Chen, Xin Yan, Bin Wen, Fan Yang, Tingting Gao, Di Zhang, Long Chen
Abstract: Although multimodal large language models (MLLMs) exhibit remarkable reasoning capabilities on complex multimodal understanding tasks, they still suffer from the notorious hallucination issue: generating outputs misaligned with obvious visual or factual evidence. Currently, training-based solutions, like direct preference optimization (DPO), leverage paired preference data to suppress hallucinations. However, they risk sacrificing general reasoning capabilities due to the likelihood displacement. Meanwhile, training-free solutions, like contrastive decoding, achieve this goal by subtracting the estimated hallucination pattern from a distorted input. Yet, these handcrafted perturbations (e.g., add noise to images) may poorly capture authentic hallucination patterns. To avoid these weaknesses of existing methods, and realize robust hallucination mitigation (i.e., maintaining general reasoning performance), we propose a novel framework: Decoupling Contrastive Decoding (DCD). Specifically, DCD decouples the learning of positive and negative samples in preference datasets, and trains separate positive and negative image projections within the MLLM. The negative projection implicitly models real hallucination patterns, which enables vision-aware negative images in the contrastive decoding inference stage. Our DCD alleviates likelihood displacement by avoiding pairwise optimization and generalizes robustly without handcrafted degradation. Extensive ablations across hallucination benchmarks and general reasoning tasks demonstrate the effectiveness of DCD, i.e., it matches DPO's hallucination suppression while preserving general capabilities and outperforms the handcrafted contrastive decoding methods.
Authors: Nathan Lambert
Abstract: Reinforcement learning from human feedback (RLHF) has become an important technical and storytelling tool to deploy the latest machine learning systems. In this book, we hope to give a gentle introduction to the core methods for people with some level of quantitative background. The book starts with the origins of RLHF -- both in recent literature and in a convergence of disparate fields of science in economics, philosophy, and optimal control. We then set the stage with definitions, problem formulation, data collection, and other common math used in the literature. The core of the book details every optimization stage in using RLHF, from starting with instruction tuning to training a reward model and finally all of rejection sampling, reinforcement learning, and direct alignment algorithms. The book concludes with advanced topics -- understudied research questions in synthetic data and evaluation -- and open questions for the field.
Authors: Kazuhisa Fujita
Abstract: This paper presents a comprehensive formulation of Kaneko's Error Diffusion Learning Algorithm (EDLA) and evaluates its effectiveness across parity check, regression, and image classification tasks. EDLA is a biologically inspired learning algorithm that provides an alternative to conventional backpropagation for training artificial neural networks. EDLA employs a single global error signal that diffuses across networks composed of paired positive and negative sublayers, eliminating traditional layer-wise error backpropagation. This study evaluates EDLA's effectiveness using benchmark tasks, such as parity check, regression, and image classification, by systematically varying the neuron count, network depth, and learning rates to assess its performance comprehensively. The experimental results demonstrate that EDLA achieves consistently high accuracy across multiple benchmarks, highlighting its effectiveness as a learning algorithm for neural networks. The choice of learning rate, neuron count, and network depth significantly influences EDLA's efficiency and convergence speed. Analysis of internal network representations reveals meaningful feature extraction capabilities, and the network's overall performance is found to be competitive with networks trained via conventional backpropagation, especially in shallow architectures. This study introduces EDLA, a biologically plausible alternative to traditional backpropagation previously underrecognized due to language barriers. By reformulating EDLA, systematically evaluating its performance, and presenting empirical evidence of its effectiveness, this study increases the visibility and accessibility of EDLA and contributes to biologically inspired training methodologies.
Authors: Ignacio Peis, Batuhan Koyuncu, Isabel Valera, Jes Frellsen
Abstract: We introduce a novel generative framework for functions by integrating Implicit Neural Representations (INRs) and Transformer-based hypernetworks into latent variable models. Unlike prior approaches that rely on MLP-based hypernetworks with scalability limitations, our method employs a Transformer-based decoder to generate INR parameters from latent variables, addressing both representation capacity and computational efficiency. Our framework extends latent diffusion models (LDMs) to INR generation by replacing standard decoders with a Transformer-based hypernetwork, which can be trained either from scratch or via hyper-transforming: a strategy that fine-tunes only the decoder while freezing the pre-trained latent space. This enables efficient adaptation of existing generative models to INR-based representations without requiring full retraining. We validate our approach across multiple modalities, demonstrating improved scalability, expressiveness, and generalization over existing INR-based generative models. Our findings establish a unified and flexible framework for learning structured function representations.
Authors: Nuojin Cheng, Alireza Doostan, Stephen Becker
Abstract: Efficient optimization remains a fundamental challenge across numerous scientific and engineering domains, especially when objective function and gradient evaluations are computationally expensive. While zeroth-order optimization methods offer effective approaches when gradients are inaccessible, their practical performance can be limited by the high cost associated with function queries. This work introduces the bi-fidelity stochastic subspace descent (BF-SSD) algorithm, a novel zeroth-order optimization method designed to reduce this computational burden. BF-SSD leverages a bi-fidelity framework, constructing a surrogate model from a combination of computationally inexpensive low-fidelity (LF) and accurate high-fidelity (HF) function evaluations. This surrogate model facilitates an efficient backtracking line search for step size selection, for which we provide theoretical convergence guarantees under standard assumptions. We perform a comprehensive empirical evaluation of BF-SSD across four distinct problems: a synthetic optimization benchmark, dual-form kernel ridge regression, black-box adversarial attacks on machine learning models, and transformer-based black-box language model fine-tuning. Numerical results demonstrate that BF-SSD consistently achieves superior optimization performance while requiring significantly fewer HF function evaluations compared to relevant baseline methods. This study highlights the efficacy of integrating bi-fidelity strategies within zeroth-order optimization, positioning BF-SSD as a promising and computationally efficient approach for tackling large-scale, high-dimensional problems encountered in various real-world applications.
Authors: Houda Belhad, Asmae Bourbia, Salma Boughanja
Abstract: Chronic diseases, such as cardiovascular disease, diabetes, chronic kidney disease, and thyroid disorders, are the leading causes of premature mortality worldwide. Early detection and intervention are crucial for improving patient outcomes, yet traditional diagnostic methods often fail due to the complex nature of these conditions. This study explores the application of machine learning (ML) and deep learning (DL) techniques to predict chronic disease and thyroid disorders. We used a variety of models, including Logistic Regression (LR), Random Forest (RF), Gradient Boosted Trees (GBT), Neural Networks (NN), Decision Trees (DT) and Native Bayes (NB), to analyze and predict disease outcomes. Our methodology involved comprehensive data pre-processing, including handling missing values, categorical encoding, and feature aggregation, followed by model training and evaluation. Performance metrics such ad precision, recall, accuracy, F1-score, and Area Under the Curve (AUC) were used to assess the effectiveness of each model. The results demonstrated that ensemble methods like Random Forest and Gradient Boosted Trees consistently outperformed. Neutral Networks also showed superior performance, particularly in capturing complex data patterns. The findings highlight the potential of ML and DL in revolutionizing chronic disease prediction, enabling early diagnosis and personalized treatment strategies. However, challenges such as data quality, model interpretability, and the need for advanced computational techniques in healthcare to improve patient outcomes and reduce the burden of chronic diseases. This study was conducted as part of Big Data class project under the supervision of our professors Mr. Abderrahmane EZ-ZAHOUT and Mr. Abdessamad ESSAIDI.
Authors: Thibault de Surrel, Florian Yger, Fabien Lotte, Sylvain Chevallier
Abstract: The goal of this paper is to show how different machine learning tools on the Riemannian manifold $\mathcal{P}_d$ of Symmetric Positive Definite (SPD) matrices can be united under a probabilistic framework. For this, we will need several Gaussian distributions defined on $\mathcal{P}_d$. We will show how popular classifiers on $\mathcal{P}_d$ can be reinterpreted as Bayes Classifiers using these Gaussian distributions. These distributions will also be used for outlier detection and dimension reduction. By showing that those distributions are pervasive in the tools used on $\mathcal{P}_d$, we allow for other machine learning tools to be extended to $\mathcal{P}_d$.
Authors: Wei Chen, Shigui Li, Jiacheng Li, Junmei Yang, John Paisley, Delu Zeng
Abstract: Density ratio estimation is fundamental to tasks involving $f$-divergences, yet existing methods often fail under significantly different distributions or inadequately overlapping supports -- the density-chasm and the support-chasm problems. Additionally, prior approaches yield divergent time scores near boundaries, leading to instability. We design $\textbf{D}^3\textbf{RE}$, a unified framework for \textbf{robust}, \textbf{stable} and \textbf{efficient} density ratio estimation. We propose the dequantified diffusion bridge interpolant (DDBI), which expands support coverage and stabilizes time scores via diffusion bridges and Gaussian dequantization. Building on DDBI, the proposed dequantified Schr{\"o}dinger bridge interpolant (DSBI) incorporates optimal transport to solve the Schr{\"o}dinger bridge problem, enhancing accuracy and efficiency. Our method offers uniform approximation and bounded time scores in theory, and outperforms baselines empirically in mutual information and density estimation tasks.
Authors: Dorit Hochbaum, Torpong Nitayanont
Abstract: In many scenarios of binary classification, only positive instances are provided in the training data, leaving the rest of the data unlabeled. This setup, known as positive-unlabeled (PU) learning, is addressed here with a network flow-based method which utilizes pairwise similarities between samples. The method we propose here, 2-HNC, leverages Hochbaum's Normalized Cut (HNC) and the set of solutions it provides by solving a parametric minimum cut problem. The set of solutions, that are nested partitions of the samples into two sets, correspond to varying tradeoff values between the two goals: high intra-similarity inside the sets and low inter-similarity between the two sets. This nested sequence is utilized here to deliver a ranking of unlabeled samples by their likelihood of being negative. Building on this insight, our method, 2-HNC, proceeds in two stages. The first stage generates this ranking without assuming any negative labels, using a problem formulation that is constrained only on positive labeled samples. The second stage augments the positive set with likely-negative samples and recomputes the classification. The final label prediction selects among all generated partitions in both stages, the one that delivers a positive class proportion, closest to a prior estimate of this quantity, which is assumed to be given. Extensive experiments across synthetic and real datasets show that 2-HNC yields strong performance and often surpasses existing state-of-the-art algorithms.
Authors: Brigt H{\aa}vardstun, Felix Marti-Perez, C\`esar Ferri, Jan Arne Telle
Abstract: In this work, we introduce metrics to evaluate the use of simplified time series in the context of interpretability of a TSC - a Time Series Classifier. Such simplifications are important because time series data, in contrast to text and image data, are not intuitively under- standable to humans. These metrics are related to the complexity of the simplifications - how many segments they contain - and to their loyalty - how likely they are to maintain the classification of the original time series. We focus on simplifications that select a subset of the original data points, and show that these typically have high Shapley value, thereby aiding interpretability. We employ these metrics to experimentally evaluate four distinct simplification algorithms, across several TSC algorithms and across datasets of varying characteristics, from seasonal or stationary to short or long. We subsequently perform a human-grounded evaluation with forward simulation, that confirms also the practical utility of the introduced metrics to evaluate the use of simplifications in the context of inter- pretability of TSC. Our findings are summarized in a framework for deciding, for a given TSC, if the various simplifications are likely to aid in its interpretability.
Authors: Zijun Chen, Shengbo Wang, Nian Si
Abstract: Motivated by practical applications where stable long-term performance is critical-such as robotics, operations research, and healthcare-we study the problem of distributionally robust (DR) average-reward reinforcement learning. We propose two algorithms that achieve near-optimal sample complexity. The first reduces the problem to a DR discounted Markov decision process (MDP), while the second, Anchored DR Average-Reward MDP, introduces an anchoring state to stabilize the controlled transition kernels within the uncertainty set. Assuming the nominal MDP is uniformly ergodic, we prove that both algorithms attain a sample complexity of $\widetilde{O}\left(|\mathbf{S}||\mathbf{A}| t_{\mathrm{mix}}^2\varepsilon^{-2}\right)$ for estimating the optimal policy as well as the robust average reward under KL and $f_k$-divergence-based uncertainty sets, provided the uncertainty radius is sufficiently small. Here, $\varepsilon$ is the target accuracy, $|\mathbf{S}|$ and $|\mathbf{A}|$ denote the sizes of the state and action spaces, and $t_{\mathrm{mix}}$ is the mixing time of the nominal MDP. This represents the first finite-sample convergence guarantee for DR average-reward reinforcement learning. We further validate the convergence rates of our algorithms through numerical experiments.
Authors: Yihan Zhu, Gang Liu, Eric Inae, Tengfei Luo, Meng Jiang
Abstract: Polymers are large macromolecules composed of repeating structural units known as monomers and are widely applied in fields such as energy storage, construction, medicine, and aerospace. However, existing graph neural network methods, though effective for small molecules, only model the single unit of polymers and fail to produce consistent vector representations for the true polymer structure with varying numbers of units. To address this challenge, we introduce Graph Repetition Invariance (GRIN), a novel method to learn polymer representations that are invariant to the number of repeating units in their graph representations. GRIN integrates a graph-based maximum spanning tree alignment with repeat-unit augmentation to ensure structural consistency. We provide theoretical guarantees for repetition-invariance from both model and data perspectives, demonstrating that three repeating units are the minimal augmentation required for optimal invariant representation learning. GRIN outperforms state-of-the-art baselines on both homopolymer and copolymer benchmarks, learning stable, repetition-invariant representations that generalize effectively to polymer chains of unseen sizes.
Authors: Xiaotian Zhang, Yue Shang, Entao Yang, Ge Zhang
Abstract: Understanding neural network's (NN) generalizability remains a central question in deep learning research. The special phenomenon of grokking, where NNs abruptly generalize long after the training performance reaches a near-perfect level, offers a unique window to investigate the underlying mechanisms of NNs' generalizability. Here we propose an interpretation for grokking by framing it as a computational glass relaxation: viewing NNs as a physical system where parameters are the degrees of freedom and train loss is the system energy, we find memorization process resembles a rapid cooling of liquid into non-equilibrium glassy state at low temperature and the later generalization is like a slow relaxation towards a more stable configuration. This mapping enables us to sample NNs' Boltzmann entropy (states of density) landscape as a function of training loss and test accuracy. Our experiments in transformers on arithmetic tasks suggests that there is NO entropy barrier in the memorization-to-generalization transition of grokking, challenging previous theory that defines grokking as a first-order phase transition. We identify a high-entropy advantage under grokking, an extension of prior work linking entropy to generalizability but much more significant. Inspired by grokking's far-from-equilibrium nature, we develop a toy optimizer WanD based on Wang-landau molecular dynamics, which can eliminate grokking without any constraints and find high-norm generalizing solutions. This provides strictly-defined counterexamples to theory attributing grokking solely to weight norm evolution towards the Goldilocks zone and also suggests new potential ways for optimizer design.
Authors: Hanlin Zhu, Shibo Hao, Zhiting Hu, Jiantao Jiao, Stuart Russell, Yuandong Tian
Abstract: Large Language Models (LLMs) have demonstrated remarkable performance in many applications, including challenging reasoning problems via chain-of-thoughts (CoTs) techniques that generate ``thinking tokens'' before answering the questions. While existing theoretical works demonstrate that CoTs with discrete tokens boost the capability of LLMs, recent work on continuous CoTs lacks a theoretical understanding of why it outperforms discrete counterparts in various reasoning tasks such as directed graph reachability, a fundamental graph reasoning problem that includes many practical domain applications as special cases. In this paper, we prove that a two-layer transformer with $D$ steps of continuous CoTs can solve the directed graph reachability problem, where $D$ is the diameter of the graph, while the best known result of constant-depth transformers with discrete CoTs requires $O(n^2)$ decoding steps where $n$ is the number of vertices ($D
Authors: Keqi Deng, Philip C. Woodland
Abstract: While Transformer self-attention offers strong parallelism, the Key-Value (KV) cache grows linearly with sequence length and becomes a bottleneck for inference efficiency. Multi-head latent attention was recently developed to compress the KV cache into a low-rank latent space. This paper proposes Multi-head Temporal Latent Attention (MTLA), which further reduces the KV cache size along the temporal dimension, greatly lowering the memory footprint of self-attention inference. MTLA employs a hyper-network to dynamically merge temporally adjacent KV cache vectors. To address the mismatch between the compressed KV cache and processed sequence lengths, a stride-aware causal mask is proposed to ensure efficient parallel training and consistency with inference behaviour. Experiments across tasks, including speech translation, speech recognition, speech understanding and text summarisation, demonstrate that MTLA achieves competitive performance compared to standard Multi-Head Attention (MHA), while greatly improving inference speed and GPU memory usage. For example, on a English-German speech translation task, MTLA achieves a 5.3x speedup and a reduction in GPU memory usage by a factor of 8.3 compared to MHA, while maintaining translation quality.
Authors: Louis B\'ethune, David Vigouroux, Yilun Du, Rufin VanRullen, Thomas Serre, Victor Boutin
Abstract: What is the shortest path between two data points lying in a high-dimensional space? While the answer is trivial in Euclidean geometry, it becomes significantly more complex when the data lies on a curved manifold -- requiring a Riemannian metric to describe the space's local curvature. Estimating such a metric, however, remains a major challenge in high dimensions. In this work, we propose a method for deriving Riemannian metrics directly from pretrained Energy-Based Models (EBMs) -- a class of generative models that assign low energy to high-density regions. These metrics define spatially varying distances, enabling the computation of geodesics -- shortest paths that follow the data manifold's intrinsic geometry. We introduce two novel metrics derived from EBMs and show that they produce geodesics that remain closer to the data manifold and exhibit lower curvature distortion, as measured by alignment with ground-truth trajectories. We evaluate our approach on increasingly complex datasets: synthetic datasets with known data density, rotated character images with interpretable geometry, and high-resolution natural images embedded in a pretrained VAE latent space. Our results show that EBM-derived metrics consistently outperform established baselines, especially in high-dimensional settings. Our work is the first to derive Riemannian metrics from EBMs, enabling data-aware geodesics and unlocking scalable, geometry-driven learning for generative modeling and simulation.
Authors: Xiaoyan Hu, Lauren Pick, Ho-fung Leung, Farzan Farnia
Abstract: The rapid advancement of generative AI has provided users with a wide range of well-trained models to address diverse prompts. When selecting a model for a given prompt, users should weigh not only its performance but also its service cost. However, existing model-selection methods typically emphasize performance while overlooking cost differences. In this paper, we introduce PromptWise, an online learning framework that assigns prompts to generative models in a cost-aware manner. PromptWise estimates prompt-model compatibility to select the least expensive model expected to deliver satisfactory outputs. Unlike standard contextual bandits that make a one-shot decision per prompt, PromptWise employs a cost-aware bandit structure that allows sequential model assignments per prompt to reduce total service cost. Through numerical experiments on tasks such as code generation and translation, we demonstrate that PromptWise can achieve performance comparable to baseline selection methods while incurring substantially lower costs. The code is available at: github.com/yannxiaoyanhu/PromptWise.
Authors: Felix Chalumeau, Daniel Rajaonarivonivelomanantsoa, Ruan de Kock, Claude Formanek, Sasha Abramowitz, Oumayma Mahjoub, Wiem Khlifi, Simon Du Toit, Louay Ben Nessir, Refiloe Shabe, Arnol Fokam, Siddarth Singh, Ulrich Mbou Sob, Arnu Pretorius
Abstract: Reinforcement learning (RL) systems have countless applications, from energy-grid management to protein design. However, such real-world scenarios are often extremely difficult, combinatorial in nature, and require complex coordination between multiple agents. This level of complexity can cause even state-of-the-art RL systems, trained until convergence, to hit a performance ceiling which they are unable to break out of with zero-shot inference. Meanwhile, many digital or simulation-based applications allow for an inference phase that utilises a specific time and compute budget to explore multiple attempts before outputting a final solution. In this work, we show that such an inference phase employed at execution time, and the choice of a corresponding inference strategy, are key to breaking the performance ceiling observed in complex multi-agent RL problems. Our main result is striking: we can obtain up to a 126% and, on average, a 45% improvement over the previous state-of-the-art across 17 tasks, using only a couple seconds of extra wall-clock time during execution. We also demonstrate promising compute scaling properties, supported by over 60k experiments, making it the largest study on inference strategies for complex RL to date. Our experimental data and code are available at https://sites.google.com/view/inference-strategies-rl.
URLs: https://sites.google.com/view/inference-strategies-rl.
Authors: Laetitia Chapel, Romain Tavenard, Samuel Vaiter
Abstract: Optimal Transport (OT) has attracted significant interest in the machine learning community, not only for its ability to define meaningful distances between probability distributions -- such as the Wasserstein distance -- but also for its formulation of OT plans. Its computational complexity remains a bottleneck, though, and slicing techniques have been developed to scale OT to large datasets. Recently, a novel slicing scheme, dubbed min-SWGG, lifts a single one-dimensional plan back to the original multidimensional space, finally selecting the slice that yields the lowest Wasserstein distance as an approximation of the full OT plan. Despite its computational and theoretical advantages, min-SWGG inherits typical limitations of slicing methods: (i) the number of required slices grows exponentially with the data dimension, and (ii) it is constrained to linear projections. Here, we reformulate min-SWGG as a bilevel optimization problem and propose a differentiable approximation scheme to efficiently identify the optimal slice, even in high-dimensional settings. We furthermore define its generalized extension for accommodating to data living on manifolds. Finally, we demonstrate the practical value of our approach in various applications, including gradient flows on manifolds and high-dimensional spaces, as well as a novel sliced OT-based conditional flow matching for image generation -- where fast computation of transport plans is essential.
Authors: Jian Yao, Ran Cheng, Xingyu Wu, Jibin Wu, Kay Chen Tan
Abstract: The reasoning capabilities of large language models (LLMs) have advanced rapidly, particularly following the release of DeepSeek R1, which has inspired a surge of research into data quality and reinforcement learning (RL) algorithms. Despite the pivotal role diversity plays in RL, its influence on LLM reasoning remains largely underexplored. To bridge this gap, this work presents a systematic investigation into the impact of diversity in RL-based training for LLM reasoning, and proposes a novel diversity-aware policy optimization method. Across evaluations on 12 LLMs, we observe a strong positive correlation between the solution diversity and Potential at k (a novel metric quantifying an LLM's reasoning potential) in high-performing models. This finding motivates our method to explicitly promote diversity during RL training. Specifically, we design a token-level diversity and reformulate it into a practical objective, then we selectively apply it to positive samples. Integrated into the R1-zero training framework, our method achieves a 3.5 percent average improvement across four mathematical reasoning benchmarks, while generating more diverse and robust solutions.
Authors: Andreas Auer, Patrick Podest, Daniel Klotz, Sebastian B\"ock, G\"unter Klambauer, Sepp Hochreiter
Abstract: In-context learning, the ability of large language models to perform tasks using only examples provided in the prompt, has recently been adapted for time series forecasting. This paradigm enables zero-shot prediction, where past values serve as context for forecasting future values, making powerful forecasting tools accessible to non-experts and increasing the performance when training data are scarce. Most existing zero-shot forecasting approaches rely on transformer architectures, which, despite their success in language, often fall short of expectations in time series forecasting, where recurrent models like LSTMs frequently have the edge. Conversely, while LSTMs are well-suited for time series modeling due to their state-tracking capabilities, they lack strong in-context learning abilities. We introduce TiRex that closes this gap by leveraging xLSTM, an enhanced LSTM with competitive in-context learning skills. Unlike transformers, state-space models, or parallelizable RNNs such as RWKV, TiRex retains state-tracking, a critical property for long-horizon forecasting. To further facilitate its state-tracking ability, we propose a training-time masking strategy called CPM. TiRex sets a new state of the art in zero-shot time series forecasting on the HuggingFace benchmarks GiftEval and Chronos-ZS, outperforming significantly larger models including TabPFN-TS (Prior Labs), Chronos Bolt (Amazon), TimesFM (Google), and Moirai (Salesforce) across both short- and long-term forecasts.
Authors: YuQing Xie, Tess Smidt
Abstract: Equivariant neural networks have proven to be effective for tasks with known underlying symmetries. However, optimizing equivariant networks can be tricky and best training practices are less established than for standard networks. In particular, recent works have found small training benefits from relaxing equivariance constraints. This raises the question: do equivariance constraints introduce fundamental obstacles to optimization? Or do they simply require different hyperparameter tuning? In this work, we investigate this question through a theoretical analysis of the loss landscape geometry. We focus on networks built using permutation representations, which we can view as a subset of unconstrained MLPs. Importantly, we show that the parameter symmetries of the unconstrained model has nontrivial effects on the loss landscape of the equivariant subspace and under certain conditions can provably prevent learning of the global minima. Further, we empirically demonstrate in such cases, relaxing to an unconstrained MLP can sometimes solve the issue. Interestingly, the weights eventually found via relaxation corresponds to a different choice of group representation in the hidden layer. From this, we draw 3 key takeaways. (1) By viewing the unconstrained version of an architecture, we can uncover hidden parameter symmetries which were broken by choice of constraint enforcement (2) Hidden symmetries give important insights on loss landscapes and can induce critical points and even minima (3) Hidden symmetry induced minima can sometimes be escaped by constraint relaxation and we observe the network jumps to a different choice of constraint enforcement. Effective equivariance relaxation may require rethinking the fixed choice of group representation in the hidden layers.
Authors: Daniel Berg Thomsen, Adrien Taylor, Aymeric Dieuleveut
Abstract: Communication between agents often constitutes a major computational bottleneck in distributed learning. One of the most common mitigation strategies is to compress the information exchanged, thereby reducing communication overhead. To counteract the degradation in convergence associated with compressed communication, error feedback schemes -- most notably $\mathrm{EF}$ and $\mathrm{EF}^{21}$ -- were introduced. In this work, we provide a tight analysis of both of these methods. Specifically, we find the Lyapunov function that yields the best possible convergence rate for each method -- with matching lower bounds. This principled approach yields sharp performance guarantees and enables a rigorous, apples-to-apples comparison between $\mathrm{EF}$, $\mathrm{EF}^{21}$, and compressed gradient descent. Our analysis is carried out in the simplified single-agent setting, which allows for clean theoretical insights and fair comparison of the underlying mechanisms.
Authors: Shubham Parashar, Shurui Gui, Xiner Li, Hongyi Ling, Sushil Vemuri, Blake Olson, Eric Li, Yu Zhang, James Caverlee, Dileep Kalathil, Shuiwang Ji
Abstract: We aim to improve the reasoning capabilities of language models via reinforcement learning (RL). Recent RL post-trained models like DeepSeek-R1 have demonstrated reasoning abilities on mathematical and coding tasks. However, prior studies suggest that using RL alone to improve reasoning on inherently difficult tasks is less effective. Here, we draw inspiration from curriculum learning and propose to schedule tasks from easy to hard (E2H), allowing LLMs to build reasoning skills gradually. Our method is termed E2H Reasoner. Empirically, we observe that, although easy tasks are important initially, fading them out through appropriate scheduling is essential in preventing overfitting. Theoretically, we establish convergence guarantees for E2H Reasoner within an approximate policy iteration framework. We derive finite-sample complexity bounds and show that when tasks are appropriately decomposed and conditioned, learning through curriculum stages requires fewer total samples than direct learning. Experiments across multiple domains show that E2H Reasoner significantly improves the reasoning ability of small LLMs (1.5B to 3B), which otherwise struggle when trained with vanilla RL alone, highlighting the effectiveness of our method. Our code can be found on https://github.com/divelab/E2H-Reasoning.
Authors: Jin Huang, Yuchao Jin, Le An, Josh Park
Abstract: This paper introduces an efficient Vision-Language Model (VLM) pipeline specifically optimized for deployment on embedded devices, such as those used in robotics and autonomous driving. The pipeline significantly reduces the computational overhead by jointly leveraging patch selection to filter irrelevant camera views, a token selection module to reduce input sequence length for the LLM, and speculative decoding to accelerate token generation. Evaluation on the NVIDIA DRIVE Thor platform for automonous driving application, our pipeline achieves $2.5\times$ end-to-end latency reduction without compromising task accuracy. The speed-up further increases to $3.2\times$ when applying FP8 post-training quantization. These results demonstrate our pipeline as a viable solution for enabling real-time VLM deployment in resource-constrained environments.
Authors: Meir Feder, Ruediger Urbanke, Yaniv Fogel
Abstract: We introduce an information-theoretic framework that views learning as universal prediction under log loss, characterized through regret bounds. Central to the framework is an effective notion of architecture-based model complexity, defined by the probability mass or volume of models in the vicinity of the data-generating process, or its projection on the model class. This volume is related to spectral properties of the expected Hessian or the Fisher Information Matrix, leading to tractable approximations. We argue that successful architectures possess a broad complexity range, enabling learning in highly over-parameterized model classes. The framework sheds light on the role of inductive biases, the effectiveness of stochastic gradient descent, and phenomena such as flat minima. It unifies online, batch, supervised, and generative settings, and applies across the stochastic-realizable and agnostic regimes. Moreover, it provides insights into the success of modern machine-learning architectures, such as deep neural networks and transformers, suggesting that their broad complexity range naturally arises from their layered structure. These insights open the door to the design of alternative architectures with potentially comparable or even superior performance.
Authors: Marco Spinaci, Marek Polewczyk, Maximilian Schambach, Sam Thelin
Abstract: Tabular in-context learning (ICL) has recently achieved state-of-the-art (SOTA) performance on several tabular prediction tasks. Previously restricted to classification problems on small tables, recent advances such as TabPFN and TabICL have extended its use to larger datasets. Although current table-native ICL architectures are architecturally efficient and well-adapted to tabular data structures, their exclusive training on synthetic data limits their ability to fully leverage the rich semantics and world knowledge contained in real-world tabular data. At the other end of the spectrum, tabular ICL models based on pretrained large language models such as TabuLa-8B integrate deep semantic understanding and world knowledge but are only able to make use of a small amount of context due to inherent architectural limitations. With the aim to combine the best of both these worlds, we introduce ConTextTab, integrating semantic understanding and alignment into a table-native ICL framework. By employing specialized embeddings for different data modalities and by training on large-scale real-world tabular data, our model is competitive with SOTA across a broad set of benchmarks while setting a new standard on the semantically rich CARTE benchmark. Code and model checkpoints are available at: https://github.com/SAP-samples/sap-rpt-1-oss.
Authors: Flavio Martinelli, Alexander Van Meegen, Berfin \c{S}im\c{s}ek, Wulfram Gerstner, Johanni Brea
Abstract: The loss landscapes of neural networks contain minima and saddle points that may be connected in flat regions or appear in isolation. We identify and characterize a special structure in the loss landscape: channels along which the loss decreases extremely slowly, while the output weights of at least two neurons, $a_i$ and $a_j$, diverge to $\pm$infinity, and their input weight vectors, $\mathbf{w_i}$ and $\mathbf{w_j}$, become equal to each other. At convergence, the two neurons implement a gated linear unit: $a_i\sigma(\mathbf{w_i} \cdot \mathbf{x}) + a_j\sigma(\mathbf{w_j} \cdot \mathbf{x}) \rightarrow \sigma(\mathbf{w} \cdot \mathbf{x}) + (\mathbf{v} \cdot \mathbf{x}) \sigma'(\mathbf{w} \cdot \mathbf{x})$. Geometrically, these channels to infinity are asymptotically parallel to symmetry-induced lines of critical points. Gradient flow solvers, and related optimization methods like SGD or ADAM, reach the channels with high probability in diverse regression settings, but without careful inspection they look like flat local minima with finite parameter values. Our characterization provides a comprehensive picture of these quasi-flat regions in terms of gradient dynamics, geometry, and functional interpretation. The emergence of gated linear units at the end of the channels highlights a surprising aspect of the computational capabilities of fully connected layers.
Authors: Ivan Marisca, Jacob Bamberger, Cesare Alippi, Michael M. Bronstein
Abstract: Graph Neural Networks (GNNs) have achieved remarkable success across various domains. However, recent theoretical advances have identified fundamental limitations in their information propagation capabilities, such as over-squashing, where distant nodes fail to effectively exchange information. While extensively studied in static contexts, this issue remains unexplored in Spatiotemporal GNNs (STGNNs), which process sequences associated with graph nodes. Nonetheless, the temporal dimension amplifies this challenge by increasing the information that must be propagated. In this work, we formalize the spatiotemporal over-squashing problem and demonstrate its distinct characteristics compared to the static case. Our analysis reveals that, counterintuitively, convolutional STGNNs favor information propagation from points temporally distant rather than close in time. Moreover, we prove that architectures that follow either time-and-space or time-then-space processing paradigms are equally affected by this phenomenon, providing theoretical justification for computationally efficient implementations. We validate our findings on synthetic and real-world datasets, providing deeper insights into their operational dynamics and principled guidance for more effective designs.
Authors: Nick Erickson, Lennart Purucker, Andrej Tschalzev, David Holzm\"uller, Prateek Mutalik Desai, David Salinas, Frank Hutter
Abstract: With the growing popularity of deep learning and foundation models for tabular data, the need for standardized and reliable benchmarks is higher than ever. However, current benchmarks are static. Their design is not updated even if flaws are discovered, model versions are updated, or new models are released. To address this, we introduce TabArena, the first continuously maintained living tabular benchmarking system. To launch TabArena, we manually curate a representative collection of datasets and well-implemented models, conduct a large-scale benchmarking study to initialize a public leaderboard, and assemble a team of experienced maintainers. Our results highlight the influence of validation method and ensembling of hyperparameter configurations to benchmark models at their full potential. While gradient-boosted trees are still strong contenders on practical tabular datasets, we observe that deep learning methods have caught up under larger time budgets with ensembling. At the same time, foundation models excel on smaller datasets. Finally, we show that ensembles across models advance the state-of-the-art in tabular machine learning. We observe that some deep learning models are overrepresented in cross-model ensembles due to validation set overfitting, and we encourage model developers to address this issue. We launch TabArena with a public leaderboard, reproducible code, and maintenance protocols to create a living benchmark available at https://tabarena.ai.
URLs: https://tabarena.ai.
Authors: Fran\c{c}ois Rozet, Ruben Ohana, Michael McCabe, Gilles Louppe, Fran\c{c}ois Lanusse, Shirley Ho
Abstract: The steep computational cost of diffusion models at inference hinders their use as fast physics emulators. In the context of image and video generation, this computational drawback has been addressed by generating in the latent space of an autoencoder instead of the pixel space. In this work, we investigate whether a similar strategy can be effectively applied to the emulation of dynamical systems and at what cost. We find that the accuracy of latent-space emulation is surprisingly robust to a wide range of compression rates (up to 1000x). We also show that diffusion-based emulators are consistently more accurate than non-generative counterparts and compensate for uncertainty in their predictions with greater diversity. Finally, we cover practical design choices, spanning from architectures to optimizers, that we found critical to train latent-space emulators.
Authors: Eloi Campagne, Yvenn Amara-Ouali, Yannig Goude, Itai Zehavi, Argyris Kalogeratos
Abstract: Forecasting electricity demand is increasingly challenging as energy systems become more decentralized and intertwined with renewable sources. Graph Neural Networks (GNNs) have recently emerged as a powerful paradigm to model spatial dependencies in load data while accommodating complex non-stationarities. This paper introduces a comprehensive framework that integrates graph-based forecasting with attention mechanisms and ensemble aggregation strategies to enhance both predictive accuracy and interpretability. Several GNN architectures -- including Graph Convolutional Networks, GraphSAGE, APPNP, and Graph Attention Networks -- are systematically evaluated on synthetic, regional (France), and fine-grained (UK) datasets. Empirical results demonstrate that graph-aware models consistently outperform conventional baselines such as Feed Forward Neural Networks and foundation models like TiREX. Furthermore, attention layers provide valuable insights into evolving spatial interactions driven by meteorological and seasonal dynamics. Ensemble aggregation, particularly through bottom-up expert combination, further improves robustness under heterogeneous data conditions. Overall, the study highlights the complementarity between structural modeling, interpretability, and robustness, and discusses the trade-offs between accuracy, model complexity, and transparency in graph-based electricity load forecasting.
Authors: Ningyuan Huang, Richard Stiskalek, Jun-Young Lee, Adrian E. Bayer, Charles C. Margossian, Christian Kragh Jespersen, Lucia A. Perez, Lawrence K. Saul, Francisco Villaescusa-Navarro
Abstract: Cosmological simulations provide a wealth of data in the form of point clouds and directed trees. A crucial goal is to extract insights from this data that shed light on the nature and composition of the Universe. In this paper we introduce CosmoBench, a benchmark dataset curated from state-of-the-art cosmological simulations whose runs required more than 41 million core-hours and generated over two petabytes of data. CosmoBench is the largest dataset of its kind: it contains 34 thousand point clouds from simulations of dark matter halos and galaxies at three different length scales, as well as 25 thousand directed trees that record the formation history of halos on two different time scales. The data in CosmoBench can be used for multiple tasks -- to predict cosmological parameters from point clouds and merger trees, to predict the velocities of individual halos and galaxies from their collective positions, and to reconstruct merger trees on finer time scales from those on coarser time scales. We provide several baselines on these tasks, some based on established approaches from cosmological modeling and others rooted in machine learning. For the latter, we study different approaches -- from simple linear models that are minimally constrained by symmetries to much larger and more computationally-demanding models in deep learning, such as graph neural networks. We find that least-squares fits with a handful of invariant features sometimes outperform deep architectures with many more parameters and far longer training time. Still there remains tremendous potential to improve these baselines by combining machine learning and cosmology to fully exploit the data. CosmoBench sets the stage for bridging cosmology and geometric deep learning at scale. We invite the community to push the frontier of scientific discovery by engaging with this dataset, available at https://cosmobench.streamlit.app
Authors: Minhak Song, Beomhan Baek, Kwangjun Ahn, Chulhee Yun
Abstract: As both model and dataset sizes continue to scale rapidly, conventional pretraining strategies with fixed compute budgets-such as cosine learning rate schedules-are increasingly inadequate for large-scale training. Recent alternatives, including warmup-stable-decay (WSD) schedules and weight averaging, offer greater flexibility. However, WSD relies on explicit decay phases to track progress, while weight averaging addresses this limitation at the cost of additional memory. In search of a more principled and scalable alternative, we revisit the Schedule-Free (SF) method [Defazio et al., 2024], which has shown strong empirical performance across diverse settings. We show that SF-AdamW effectively navigates the "river" structure of the loss landscape without decay phases or auxiliary averaging, making it particularly suitable for continuously scaling training workloads. To understand this behavior, we conduct a theoretical and empirical analysis of SF dynamics, revealing that it implicitly performs weight averaging without memory overhead. Guided by this analysis, we propose a refined variant of SF that improves robustness to momentum and performs better under large batch sizes, addressing key limitations of the original method. Together, these results establish SF as a practical, scalable, and theoretically grounded approach for language model training.
Authors: Shuyao Li, Ilias Diakonikolas, Jelena Diakonikolas
Abstract: Distributionally Robust Optimization (DRO) provides a framework for decision-making under distributional uncertainty, yet its effectiveness can be compromised by outliers in the training data. This paper introduces a principled approach to simultaneously address both challenges. We focus on optimizing Wasserstein-1 DRO objectives for generalized linear models with convex Lipschitz loss functions, where an $\epsilon$-fraction of the training data is adversarially corrupted. Our primary contribution lies in a novel modeling framework that integrates robustness against training data contamination with robustness against distributional shifts, alongside an efficient algorithm inspired by robust statistics to solve the resulting optimization problem. We prove that our method achieves an estimation error of $O(\sqrt{\epsilon})$ for the true DRO objective value using only the contaminated data under the bounded covariance assumption. This work establishes the first rigorous guarantees, supported by efficient computation, for learning under the dual challenges of data contamination and distributional shifts.
Authors: Zijie Guo, Jiong Wang, Fenghua Ling, Wangxu Wei, Xiaoyu Yue, Zhe Jiang, Wanghan Xu, Jing-Jia Luo, Lijing Cheng, Yoo-Geun Ham, Fengfei Song, Pierre Gentine, Toshio Yamagata, Ben Fei, Wenlong Zhang, Xinyu Gu, Chao Li, Yaqiang Wang, Tao Chen, Wanli Ouyang, Bowen Zhou, Lei Bai
Abstract: Scientific progress in Earth science depends on integrating data across the planet's interconnected spheres. However, the accelerating volume and fragmentation of multi-sphere knowledge and data have surpassed human analytical capacity. This creates a major bottleneck for discovery, especially in climate science. To address this challenge, we introduce EarthLink, the first self-evolving AI agent system designed as an interactive "copilot" for Earth scientists. Through natural language interaction, EarthLink automates the entire research workflow by integrating planning, code execution, data analysis, and physical reasoning into a unified process that directly addresses this limitation. Beyond efficiency, it exhibits human-like cross-disciplinary analytical ability and achieves proficiency comparable to a junior researcher in expert evaluations on core large-scale climate tasks, including model-observation comparison and climate change understanding. When tasked with an open scientific problem, specifically the discovery of precursors of the Atlantic Ni\~no, EarthLink autonomously developed a research strategy, identified sources of predictability, verified its hypotheses with available data, and proposed a physically consistent mechanism. These emerging capabilities enable a new human-AI research paradigm. Scientists can focus on value and result judgments, while AI systems handle complex data analysis and knowledge integration. This accelerates the pace and breadth of discovery in Earth sciences. The system is accessible at our website https://earthlink.intern-ai.org.cn.
Authors: Salah A. Faroughi, Farinaz Mostajeran, Amin Hamed Mashhadzadeh, Shirko Faroughi
Abstract: The field of scientific machine learning, which originally utilized multilayer perceptrons (MLPs), is increasingly adopting Kolmogorov-Arnold Networks (KANs) for data encoding. This shift is driven by the limitations of MLPs, including poor interpretability, fixed activation functions, and difficulty capturing localized or high-frequency features. KANs address these issues with enhanced interpretability and flexibility, enabling more efficient modeling of complex nonlinear interactions and effectively overcoming the constraints associated with conventional MLP architectures. This review categorizes recent progress in KAN-based models across three distinct perspectives: (i) data-driven learning, (ii) physics-informed modeling, and (iii) deep-operator learning. Each perspective is examined through the lens of architectural design, training strategies, application efficacy, and comparative evaluation against MLP-based counterparts. By benchmarking KANs against MLPs, we highlight consistent improvements in accuracy, convergence, and spectral representation, clarifying KANs' advantages in capturing complex dynamics while learning more effectively. In addition to reviewing recent literature, this work also presents several comparative evaluations that clarify central characteristics of KAN modeling and hint at their potential implications for real-world applications. Finally, this review identifies critical challenges and open research questions in KAN development, particularly regarding computational efficiency, theoretical guarantees, hyperparameter tuning, and algorithm complexity. We also outline future research directions aimed at improving the robustness, scalability, and physical consistency of KAN-based frameworks.
Authors: Biyi Fang, Truong Vo, Jean Utke, Diego Klabjan
Abstract: Transformer architectures dominate modern NLP but often demand heavy computational resources and intricate hyperparameter tuning. To mitigate these challenges, we propose a novel framework, BoostTransformer, that augments transformers with boosting principles through subgrid token selection and importance-weighted sampling. Our method incorporates a least square boosting objective directly into the transformer pipeline, enabling more efficient training and improved performance. Across multiple fine-grained text classification benchmarks, BoostTransformer demonstrates both faster convergence and higher accuracy, surpassing standard transformers while minimizing architectural search overhead.
Authors: Claudiu Leoveanu-Condrei
Abstract: Generative models, particularly Large Language Models (LLMs), produce fluent outputs yet lack verifiable guarantees. We adapt Design by Contract (DbC) and type-theoretic principles to introduce a contract layer that mediates every LLM call. Contracts stipulate semantic and type requirements on inputs and outputs, coupled with probabilistic remediation to steer generation toward compliance. The layer exposes the dual view of LLMs as semantic parsers and probabilistic black-box components. Contract satisfaction is probabilistic and semantic validation is operationally defined through programmer-specified conditions on well-typed data structures. More broadly, this work postulates that any two agents satisfying the same contracts are \emph{functionally equivalent} with respect to those contracts.
Authors: Neng Kai Nigel Neo, Lim Jing, Ngoui Yong Zhau Preston, Koh Xue Ting Serene, Bingquan Shen
Abstract: A common approach to the de novo molecular generation problem from mass spectra involves a two-stage pipeline: (1) encoding mass spectra into molecular fingerprints, followed by (2) decoding these fingerprints into molecular structures. In our work, we adopt MIST (Goldman et. al., 2023) as the encoder and MolForge (Ucak et. al., 2023) as the decoder, leveraging additional training data to enhance performance. We also threshold the probabilities of each fingerprint bit to focus on the presence of substructures. This results in a tenfold improvement over previous state-of-the-art methods, generating top-1 31% / top-10 40% of molecular structures correctly from mass spectra in MassSpecGym (Bushuiev et. al., 2024). We position this as a strong baseline for future research in de novo molecule elucidation from mass spectra.
Authors: Hao Zhang, Aining Jia, Weifeng Bu, Yushu Cai, Kai Sheng, Hao Chen, Xin He
Abstract: Large Language Models (LLMs) demonstrate exceptional performance but entail significant memory and computational costs, restricting their practical deployment. While existing INT4/INT8 quantization reduces these costs, they often degrade accuracy or lack optimal efficiency. INT6 quantization offers a superior trade-off between model accuracy and inference efficiency, but lacks hardware support in modern GPUs, forcing emulation via higher-precision arithmetic units that limit acceleration. In this paper, we propose FlexQ, a novel post-training INT6 quantization framework combining algorithmic innovation with system-level optimizations. FlexQ employs uniform 6-bit weight quantization across all layers, with adaptive retention of 8-bit activations in layers identified through layer-wise sensitivity analysis. To maximize hardware efficiency, we develop a specialized high-performance GPU kernel supporting matrix multiplication for W6A6 and W6A8 representations via Binary Tensor Core (BTC) equivalents, effectively bypassing the lack of native INT6 tensor cores. Evaluations on LLaMA family models show FlexQ maintains near-FP16 accuracy, with perplexity increases of no more than 0.1 on WikiText2. The proposed kernel achieves an average 1.39$\times$ speedup over ABQ-LLM on LLaMA-2-70B linear layers. End-to-end, FlexQ delivers 1.33$\times$ inference acceleration and 1.21$\times$ memory savings over SmoothQuant. Code is released at https://github.com/FlyFoxPlayer/FlexQ.
Authors: Mojtaba Fayaz-Bakhsh, Danial Ataee, MohammadAmin Fazli
Abstract: Active preference learning offers an efficient approach to modeling preferences, but it is hindered by the cold-start problem, which leads to a marked decline in performance when no initial labeled data are available. While cold-start solutions have been proposed for domains such as vision and text, the cold-start problem in active preference learning remains largely unexplored, underscoring the need for practical, effective methods. Drawing inspiration from established practices in social and economic research, the proposed method initiates learning with a self-supervised phase that employs Principal Component Analysis (PCA) to generate initial pseudo-labels. This process produces a \say{warmed-up} model based solely on the data's intrinsic structure, without requiring expert input. The model is then refined through an active learning loop that strategically queries a simulated noisy oracle for labels. Experiments conducted on various socio-economic datasets, including those related to financial credibility, career success rate, and socio-economic status, consistently show that the PCA-driven approach outperforms standard active learning strategies that start without prior information. This work thus provides a computationally efficient and straightforward solution that effectively addresses the cold-start problem.
Authors: Will Fein, Ryan J. Horwitz, John E. Brown III, Amit Misra, Felipe Oviedo, Kevin White, Juan M. Lavista Ferres, Samuel K. Wasser
Abstract: The transnational ivory trade continues to drive the decline of elephant populations across Africa, and trafficking networks remain difficult to disrupt. Tusks seized by law enforcement officials carry forensic information on the traffickers responsible for their export, including DNA evidence and handwritten markings made by traffickers. For 20 years, analyses of tusk DNA have identified where elephants were poached and established connections among shipments of ivory. While the links established using genetic evidence are extremely conclusive, genetic data is expensive and sometimes impossible to obtain. But though handwritten markings are easy to photograph, they are rarely documented or analyzed. Here, we present an AI-driven pipeline for extracting and analyzing handwritten markings on seized elephant tusks, offering a novel, scalable, and low-cost source of forensic evidence. Having collected 6,085 photographs from eight large seizures of ivory over a 6-year period (2014-2019), we used an object detection model to extract over 17,000 individual markings, which were then labeled and described using state-of-the-art AI tools. We identified 184 recurring "signature markings" that connect the tusks on which they appear. 20 signature markings were observed in multiple seizures, establishing forensic links between these seizures through traffickers involved in both shipments. This work complements other investigative techniques by filling in gaps where other data sources are unavailable. The study demonstrates the transformative potential of AI in wildlife forensics and highlights practical steps for integrating handwriting analysis into efforts to disrupt organized wildlife crime.
Authors: Weijian Mai, Jiamin Wu, Yu Zhu, Zhouheng Yao, Dongzhan Zhou, Andrew F. Luo, Qihao Zheng, Wanli Ouyang, Chunfeng Song
Abstract: Deciphering how visual stimuli are transformed into cortical responses is a fundamental challenge in computational neuroscience. This visual-to-neural mapping is inherently a one-to-many relationship, as identical visual inputs reliably evoke variable hemodynamic responses across trials, contexts, and subjects. However, existing deterministic methods struggle to simultaneously model this biological variability while capturing the underlying functional consistency that encodes stimulus information. To address these limitations, we propose SynBrain, a generative framework that simulates the transformation from visual semantics to neural responses in a probabilistic and biologically interpretable manner. SynBrain introduces two key components: (i) BrainVAE models neural representations as continuous probability distributions via probabilistic learning while maintaining functional consistency through visual semantic constraints; (ii) A Semantic-to-Neural Mapper acts as a semantic transmission pathway, projecting visual semantics into the neural response manifold to facilitate high-fidelity fMRI synthesis. Experimental results demonstrate that SynBrain surpasses state-of-the-art methods in subject-specific visual-to-fMRI encoding performance. Furthermore, SynBrain adapts efficiently to new subjects with few-shot data and synthesizes high-quality fMRI signals that are effective in improving data-limited fMRI-to-image decoding performance. Beyond that, SynBrain reveals functional consistency across trials and subjects, with synthesized signals capturing interpretable patterns shaped by biological neural variability. Our code is available at https://github.com/MichaelMaiii/SynBrain.
Authors: Jayneel Parekh, Pegah Khayatan, Mustafa Shukor, Arnaud Dapogny, Alasdair Newson, Matthieu Cord
Abstract: Steering has emerged as a practical approach to enable post-hoc guidance of LLMs towards enforcing a specific behavior. However, it remains largely underexplored for multimodal LLMs (MLLMs); furthermore, existing steering techniques, such as mean steering, rely on a single steering vector, applied independently of the input query. This paradigm faces limitations when the desired behavior is dependent on the example at hand. For example, a safe answer may consist in abstaining from answering when asked for an illegal activity, or may point to external resources or consultation with an expert when asked about medical advice. In this paper, we investigate a fine-grained steering that uses an input-specific linear shift. This shift is computed using contrastive input-specific prompting. However, the input-specific prompts required for this approach are not known at test time. Therefore, we propose to train a small auxiliary module to predict the input-specific steering vector. Our approach, dubbed as L2S (Learn-to-Steer), demonstrates that it reduces hallucinations and enforces safety in MLLMs, outperforming other static baselines. Our code is publicly available at https://jayneelparekh.github.io/learn-to-steer/
Authors: Jiahua Lu, Huaxiao Liu, Shuotong Bai, Junjie Xu, Renqiang Luo, Enyan Dai
Abstract: Graph Neural Networks (GNNs) have achieved remarkable success across diverse applications. However, due to the biases in the graph structures, graph neural networks face significant challenges in fairness. Although the original user graph structure is generally biased, it is promising to guide these existing structures toward unbiased ones by introducing new links. The fairness guidance via new links could foster unbiased communities, thereby enhancing fairness in downstream applications. To address this issue, we propose a novel framework named FairGuide. Specifically, to ensure fairness in downstream tasks trained on fairness-guided graphs, we introduce a differentiable community detection task as a pseudo downstream task. Our theoretical analysis further demonstrates that optimizing fairness within this pseudo task effectively enhances structural fairness, promoting fairness generalization across diverse downstream applications. Moreover, FairGuide employs an effective strategy which leverages meta-gradients derived from the fairness-guidance objective to identify new links that significantly enhance structural fairness. Extensive experimental results demonstrate the effectiveness and generalizability of our proposed method across a variety of graph-based fairness tasks.
Authors: Sadman Mohammad Nasif, Md Abrar Jahin, M. F. Mridha
Abstract: The Lottery Ticket Hypothesis (LTH) suggests that within large neural networks, there exist sparse, trainable "winning tickets" capable of matching the performance of the full model, but identifying them through Iterative Magnitude Pruning (IMP) is computationally expensive. Recent work introduced COLT, an accelerator that discovers a "consensus" subnetwork by intersecting masks from models trained on disjoint data partitions; however, this approach discards all non-overlapping weights, assuming they are unimportant. This paper challenges that assumption and proposes the Disentangled Lottery Ticket (DiLT) Hypothesis, which posits that the intersection mask represents a universal, task-agnostic "core" subnetwork, while the non-overlapping difference masks capture specialized, task-specific "specialist" subnetworks. A framework is developed to identify and analyze these components using the Gromov-Wasserstein (GW) distance to quantify functional similarity between layer representations and reveal modular structures through spectral clustering. Experiments on ImageNet and fine-grained datasets such as Stanford Cars, using ResNet and Vision Transformer architectures, show that the "core" ticket provides superior transfer learning performance, the "specialist" tickets retain domain-specific features enabling modular assembly, and the full re-assembled "union" ticket outperforms COLT - demonstrating that non-consensus weights play a critical functional role. This work reframes pruning as a process for discovering modular, disentangled subnetworks rather than merely compressing models.
Authors: Diksha Gupta, Antonio Honsell, Chuan Xu, Nirupam Gupta, Giovanni Neglia
Abstract: Distributed learning enables scalable model training over decentralized data, but remains hindered by Byzantine faults and high communication costs. While both challenges have been studied extensively in isolation, their interplay has received limited attention. Prior work has shown that naively combining communication compression with Byzantine-robust aggregation can severely weaken resilience to faulty nodes. The current state-of-the-art, Byz-DASHA-PAGE, leverages a momentum-based variance reduction scheme to counteract the negative effect of compression noise on Byzantine robustness. In this work, we introduce RoSDHB, a new algorithm that integrates classical Polyak momentum with a coordinated compression strategy. Theoretically, RoSDHB matches the convergence guarantees of Byz-DASHA-PAGE under the standard $(G,B)$-gradient dissimilarity model, while relying on milder assumptions and requiring less memory and communication per client. Empirically, RoSDHB demonstrates stronger robustness while achieving substantial communication savings compared to Byz-DASHA-PAGE.
Authors: Rodrigo Tertulino, Ricardo Almeida
Abstract: This study proposes and validates a Federated Learning (FL) framework to proactively identify at-risk students while preserving data privacy. Persistently high dropout rates in distance education remain a pressing institutional challenge. Using the large-scale OULAD dataset, we simulate a privacy-centric scenario where models are trained on early academic performance and digital engagement patterns. Our work investigates the practical trade-offs between model complexity (Logistic Regression vs. a Deep Neural Network) and the impact of local data balancing. The resulting federated model achieves strong predictive power (ROC AUC approximately 85%), demonstrating that FL is a practical and scalable solution for early-warning systems that inherently respects student data sovereignty.
Authors: Marc Boull\'e, Nicolas Voisine, Bruno Guerraz, Carine Hue, Felipe Olmos, Vladimir Popescu, St\'ephane Gouache, St\'ephane Bouget, Alexis Bondu, Luc Aurelien Gauthier, Yassine Nair Benrekia, Fabrice Cl\'erot, Vincent Lemaire
Abstract: Khiops is an open source machine learning tool designed for mining large multi-table databases. Khiops is based on a unique Bayesian approach that has attracted academic interest with more than 20 publications on topics such as variable selection, classification, decision trees and co-clustering. It provides a predictive measure of variable importance using discretisation models for numerical data and value clustering for categorical data. The proposed classification/regression model is a naive Bayesian classifier incorporating variable selection and weight learning. In the case of multi-table databases, it provides propositionalisation by automatically constructing aggregates. Khiops is adapted to the analysis of large databases with millions of individuals, tens of thousands of variables and hundreds of millions of records in secondary tables. It is available on many environments, both from a Python library and via a user interface.
Authors: Yilin Li, Guozhu Meng, Mingyang Sun, Yanzhong Wang, Kun Sun, Hailong Chang, Yuekang Li
Abstract: On-device deep learning models have extensive real world demands. Deep learning compilers efficiently compile models into executables for deployment on edge devices, but these executables may face the threat of reverse engineering. Previous studies have attempted to decompile DNN executables, but they face challenges in handling compilation optimizations and analyzing quantized compiled models. In this paper, we present NeuroDeX to unlock diverse support in decompiling DNN executables. NeuroDeX leverages the semantic understanding capabilities of LLMs along with dynamic analysis to accurately and efficiently perform operator type recognition, operator attribute recovery and model reconstruction. NeuroDeX can recover DNN executables into high-level models towards compilation optimizations, different architectures and quantized compiled models. We conduct experiments on 96 DNN executables across 12 common DNN models. Extensive experimental results demonstrate that NeuroDeX can decompile non-quantized executables into nearly identical high-level models. NeuroDeX can recover functionally similar high-level models for quantized executables, achieving an average top-1 accuracy of 72%. NeuroDeX offers a more comprehensive and effective solution compared to previous DNN executables decompilers.
Authors: Hangzhan Jin, Sitao Luan, Sicheng Lyu, Guillaume Rabusseau, Reihaneh Rabbany, Doina Precup, Mohammad Hamdaqa
Abstract: The two-stage fine-tuning paradigm of Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (RL) has empirically shown better reasoning performance than one-stage SFT for the post-training of Large Language Models (LLMs). However, the evolution and mechanism behind the synergy of SFT and RL are still under-explored and inconclusive. In our study, we find the well-known claim "SFT memorizes, RL generalizes" is over-simplified, and discover that: (1) OOD performance peaks at the early stage of SFT and then declines (OOD forgetting), the best SFT checkpoint cannot be captured by training/test loss; (2) the subsequent RL stage does not generate fundamentally better OOD capability, instead it plays an \textbf{OOD restoration} role, recovering the lost reasoning ability during SFT; (3) The recovery ability has boundaries, \ie{} \textbf{if SFT trains for too short or too long, RL cannot recover the lost OOD ability;} (4) To uncover the underlying mechanisms behind the forgetting and restoration process, we employ SVD analysis on parameter matrices, manually edit them, and observe their impacts on model performance. Unlike the common belief that the shift of model capacity mainly results from the changes of singular values, we find that they are actually quite stable throughout fine-tuning. Instead, the OOD behavior strongly correlates with the \textbf{rotation of singular vectors}. Our findings re-identify the roles of SFT and RL in the two-stage fine-tuning and discover the rotation of singular vectors as the key mechanism. %reversing the rotations induced by SFT, which shows recovery from forgetting, whereas imposing the SFT parameter directions onto a RL-tuned model results in performance degradation. Code is available at https://github.com/xiaodanguoguo/RL_Heals_SFT
Authors: Zhenyu Tao, Wei Xu, Xiaohu You
Abstract: The bisimulation metric (BSM) is a powerful tool for computing state similarities within a Markov decision process (MDP), revealing that states closer in BSM have more similar optimal value functions. While BSM has been successfully utilized in reinforcement learning (RL) for tasks like state representation learning and policy exploration, its application to multiple-MDP scenarios, such as policy transfer, remains challenging. Prior work has attempted to generalize BSM to pairs of MDPs, but a lack of rigorous analysis of its mathematical properties has limited further theoretical progress. In this work, we formally establish a generalized bisimulation metric (GBSM) between pairs of MDPs, which is rigorously proven with the three fundamental properties: GBSM symmetry, inter-MDP triangle inequality, and the distance bound on identical state spaces. Leveraging these properties, we theoretically analyse policy transfer, state aggregation, and sampling-based estimation in MDPs, obtaining explicit bounds that are strictly tighter than those derived from the standard BSM. Additionally, GBSM provides a closed-form sample complexity for estimation, improving upon existing asymptotic results based on BSM. Numerical results validate our theoretical findings and demonstrate the effectiveness of GBSM in multi-MDP scenarios.
Authors: Binghui Li, Fengling Chen, Zixun Huang, Lean Wang, Lei Wu
Abstract: Scaling laws have emerged as a unifying lens for understanding and guiding the training of large language models (LLMs). However, existing studies predominantly focus on the final-step loss, leaving open whether the entire loss dynamics obey similar laws and, crucially, how the learning rate schedule (LRS) shapes them. We address these gaps in a controlled theoretical setting by analyzing stochastic gradient descent (SGD) on a power-law kernel regression model. The key insight is a novel intrinsic-time viewpoint, which captures the training progress more faithfully than iteration count. We then establish a Functional Scaling Law (FSL) that captures the full loss trajectory under arbitrary LRSs, with the schedule's influence entering through a simple convolutional functional. We further instantiate the theory for three representative LRSs -- constant, exponential decay, and warmup-stable-decay (WSD) -- and derive explicit scaling relations in both data- and compute-limited regimes. These comparisons explain key empirical phenomena: (i) higher-capacity models are more data- and compute-efficient; (ii) learning-rate decay improves training efficiency; and (iii) WSD-type schedules outperform pure decay. Finally, experiments on LLMs ranging from 0.1B to 1B parameters demonstrate the practical relevance of FSL as a surrogate model for fitting and predicting loss trajectories in large-scale pre-training.
Authors: Xin Yu, Cong Xie, Ziyu Zhao, Tiantian Fan, Lingzhou Xue, Zhi Zhang
Abstract: Low-rank adaptation (LoRA) has become a widely used paradigm for parameter-efficient fine-tuning of large language models, yet its representational capacity often lags behind full fine-tuning. Within the context of LoRA, a key open question is how to obtain expressive low-rank adapters from over-parameterized spaces. We propose \textit{PrunedLoRA}, a new framework that leverages structured pruning to obtain highly representative low-rank adapters from an over-parameterized initialization. Unlike prior approaches that impose a fixed low-rank budget, PrunedLoRA dynamically prunes less important components during fine-tuning and prevents their reactivation, enabling flexible and adaptive rank allocation. For structured pruning, by minimizing the pruning error for overall loss, we provide fine-grained pruning and recovery updates in a gradient-based pruning strategy with grounded interpretation. We provide the first theoretical analysis of the robustness of structured pruning and provably show that under the impact of weight perturbation, gradient-based pruning is more robust than activation-based pruning with respect to overall loss. Empirically, PrunedLoRA consistently outperforms LoRA and its variants across supervised fine-tuning tasks in mathematical reasoning, code generation, and natural language understanding, and it also demonstrates advantages over existing structured pruning methods across diverse sparsity levels.
Authors: Shutong Wu, Jiawei Zhang
Abstract: Diffusion Large Language Models (DLLMs) have emerged as a new paradigm of language modeling beyond autoregressive next-token prediction. Thanks to their bidirectional attention mechanism, DLLMs are more capable of capturing the connection of context, and thus show unique advantages in challenges like the famous "reversal curse" or learning under data-constrained scenarios. In addition, taking advantage of their inherent modeling foundations, DLLMs have the great potential of efficient inference with parallel decoding algorithms, which enable multi-token prediction per step. However, the high generation quality often requires the number of decoding steps equal to the sequence length, which performs a one-token-per-step decoding, and existing parallel decoding algorithms, which yield suboptimal decoding paths, bring inference speedup at the cost of non-negligible performance degradation. To overcome this challenge, we introduce Free Draft-and-Verification (FreeDave), a novel fast decoding algorithm tailored for DLLMs that achieves lossless parallel decoding without any model modification or extra modules. Specifically, we propose an algorithm of parallel-decoded candidate generation and verification, which is theoretically guaranteed to use the fewest model forward calls to reproduce the same sequence generated by static decoding when enough computation and memory budget is provided. By extensive evaluations on math reasoning and code generation benchmarks across different DLLMs, FreeDave is proven to boost the inference throughput up to $3.78\times$ without performance degradation.
Authors: Gautham Bekal, Ashish Pujari, Scott David Kelly
Abstract: Continual learning involves learning from a stream of data without repetition of data points, a scenario that is inherently complex due to distributional shift across tasks. We propose a query-only attention mechanism that discards keys and values, yet preserves the core inductive bias of transformer architectures. In continual learning scenarios, this simplified mechanism significantly mitigates both loss of plasticity and catastrophic forgetting, outperforming baselines such as selective re-initialization. We establish a conceptual link between query-only attention, full transformer attention, and model agnostic meta-learning, framing them as instances of meta-learning. We further provide intuition for why query-based models and attention networks help preserve plasticity in continual settings. Finally, through preliminary Hessian spectrum analysis, we observe that models maintaining higher curvature rank across tasks tend to retain plasticity. Our findings suggest that full attention may not be essential for capturing the benefits of meta-learning in continual learning.
Authors: Chaoran Cheng, Yusong Wang, Yuxin Chen, Xiangxin Zhou, Nanning Zheng, Ge Liu
Abstract: Consistency models are a class of generative models that enable few-step generation for diffusion and flow matching models. While consistency models have achieved promising results on Euclidean domains like images, their applications to Riemannian manifolds remain challenging due to the curved geometry. In this work, we propose the Riemannian Consistency Model (RCM), which, for the first time, enables few-step consistency modeling while respecting the intrinsic manifold constraint imposed by the Riemannian geometry. Leveraging the covariant derivative and exponential-map-based parameterization, we derive the closed-form solutions for both discrete- and continuous-time training objectives for RCM. We then demonstrate theoretical equivalence between the two variants of RCM: Riemannian consistency distillation (RCD) that relies on a teacher model to approximate the marginal vector field, and Riemannian consistency training (RCT) that utilizes the conditional vector field for training. We further propose a simplified training objective that eliminates the need for the complicated differential calculation. Finally, we provide a unique kinematics perspective for interpreting the RCM objective, offering new theoretical angles. Through extensive experiments, we manifest the superior generative quality of RCM in few-step generation on various non-Euclidean manifolds, including flat-tori, spheres, and the 3D rotation group SO(3).
Authors: Junyi Wu, Guang Lin
Abstract: We propose PO-CKAN, a physics-informed deep operator framework based on Chunkwise Rational Kolmogorov--Arnold Networks (KANs), for approximating the solution operators of partial differential equations. This framework leverages a Deep Operator Network (DeepONet) architecture that incorporates Chunkwise Rational Kolmogorov-Arnold Network (CKAN) sub-networks for enhanced function approximation. The principles of Physics-Informed Neural Networks (PINNs) are integrated into the operator learning framework to enforce physical consistency. This design enables the efficient learning of physically consistent spatio-temporal solution operators and allows for rapid prediction for parametric time-dependent PDEs with varying inputs (e.g., parameters, initial/boundary conditions) after training. Validated on challenging benchmark problems, PO-CKAN demonstrates accurate operator learning with results closely matching high-fidelity solutions. PO-CKAN adopts a DeepONet-style branch--trunk architecture with its sub-networks instantiated as rational KAN modules, and enforces physical consistency via a PDE residual (PINN-style) loss. On Burgers' equation with $\nu=0.01$, PO-CKAN reduces the mean relative $L^2$ error by approximately 48\% compared to PI-DeepONet, and achieves competitive accuracy on the Eikonal and diffusion--reaction benchmarks.
Authors: Zixuan Gong, Jiaye Teng, Yong Liu
Abstract: While looped transformers (termed as Looped-Attn) often outperform standard transformers (termed as Single-Attn) on complex reasoning tasks, the theoretical basis for this advantage remains underexplored. In this paper, we explain this phenomenon through the lens of loss landscape geometry, inspired by empirical observations of their distinct dynamics at both sample and Hessian levels. To formalize this, we extend the River-Valley landscape model by distinguishing between U-shaped valleys (flat) and V-shaped valleys (steep). Based on empirical observations, we conjecture that the recursive architecture of Looped-Attn induces a landscape-level inductive bias towards River-V-Valley. Theoretical derivations based on this inductive bias guarantee a better loss convergence along the river due to valley hopping, and further encourage learning about complex patterns compared to the River-U-Valley induced by Single-Attn. Building on this insight, we propose SHIFT (Staged HIerarchical Framework for Progressive Training), a staged training framework that accelerates the training process of Looped-Attn while achieving comparable performances.
Authors: Ali Atiah Alzahrani
Abstract: We examine whether regime-conditioned generative scenarios combined with a convex CVaR allocator improve portfolio decisions under regime shifts. We present MARCD, a generative-to-decision framework with: (i) a Gaussian HMM to infer latent regimes; (ii) a diffusion generator that produces regime-conditioned scenarios; (iii) signal extraction via blended, shrunk moments; and (iv) a governed CVaR epigraph quadratic program. Contributions: Within the Scenario stage we introduce a tail-weighted diffusion objective that up-weights low-quantile outcomes relevant for drawdowns and a regime-expert (MoE) denoiser whose gate increases with crisis posteriors; both are evaluated end-to-end through the allocator. Under strict walk-forward on liquid multi-asset ETFs (2005-2025), MARCD exhibits stronger scenario calibration and materially smaller drawdowns: MaxDD 9.3% versus 14.1% for BL (a 34% reduction) over 2020-2025 out-of-sample. The framework provides an auditable pipeline with explicit budget, box, and turnover constraints, demonstrating the value of decision-aware generative modeling in finance.
Authors: Mohammad Amin Nabian, Sudeep Chavare, Deepak Akhare, Rishikesh Ranade, Ram Cherukuri, Srinivas Tadepalli
Abstract: Crashworthiness assessment is a critical aspect of automotive design, traditionally relying on high-fidelity finite element (FE) simulations that are computationally expensive and time-consuming. This work presents an exploratory comparative study on developing machine learning-based surrogate models for efficient prediction of structural deformation in crash scenarios using the NVIDIA PhysicsNeMo framework. Given the limited prior work applying machine learning to structural crash dynamics, the primary contribution lies in demonstrating the feasibility and engineering utility of the various modeling approaches explored in this work. We investigate two state-of-the-art neural network architectures for modeling crash dynamics: MeshGraphNet, and Transolver. Additionally, we examine three strategies for modeling transient dynamics: time-conditional, the standard Autoregressive approach, and a stability-enhanced Autoregressive scheme incorporating rollout-based training. The models are evaluated on a comprehensive Body-in-White (BIW) crash dataset comprising 150 detailed FE simulations using LS-DYNA. The dataset represents a structurally rich vehicle assembly with over 200 components, including 38 key components featuring variable thickness distributions to capture realistic manufacturing variability. Each model utilizes the undeformed mesh geometry and component characteristics as inputs to predict the spatiotemporal evolution of the deformed mesh during the crash sequence. Evaluation results show that the models capture the overall deformation trends with reasonable fidelity, demonstrating the feasibility of applying machine learning to structural crash dynamics. Although not yet matching full FE accuracy, the models achieve orders-of-magnitude reductions in computational cost, enabling rapid design exploration and early-stage optimization in crashworthiness evaluation.
Authors: Longwei Wang, Ifrat Ikhtear Uddin, KC Santosh, Chaowei Zhang, Xiao Qin, Yang Zhou
Abstract: Adversarial examples reveal critical vulnerabilities in deep neural networks by exploiting their sensitivity to imperceptible input perturbations. While adversarial training remains the predominant defense strategy, it often incurs significant computational cost and may compromise clean-data accuracy. In this work, we investigate an architectural approach to adversarial robustness by embedding group-equivariant convolutions-specifically, rotation- and scale-equivariant layers-into standard convolutional neural networks (CNNs). These layers encode symmetry priors that align model behavior with structured transformations in the input space, promoting smoother decision boundaries and greater resilience to adversarial attacks. We propose and evaluate two symmetry-aware architectures: a parallel design that processes standard and equivariant features independently before fusion, and a cascaded design that applies equivariant operations sequentially. Theoretically, we demonstrate that such models reduce hypothesis space complexity, regularize gradients, and yield tighter certified robustness bounds under the CLEVER (Cross Lipschitz Extreme Value for nEtwork Robustness) framework. Empirically, our models consistently improve adversarial robustness and generalization across CIFAR-10, CIFAR-100, and CIFAR-10C under both FGSM and PGD attacks, without requiring adversarial training. These findings underscore the potential of symmetry-enforcing architectures as efficient and principled alternatives to data augmentation-based defenses.
Authors: Weilin Wan, Weizhong Zhang, Cheng Jin
Abstract: Data selection improves computational efficiency by choosing informative subsets of training samples. However, existing methods ignore the compute budget, treating data selection and importance evaluation independently of compute budget constraints. Yet empirical studies show no algorithm can consistently outperform others (or even random selection) across varying budgets. We therefore argue that compute budget must be integral to data-selection strategies, since different budgets impose distinct requirements on data quantity, quality, and distribution for effective training. To this end, we propose a novel Computational budget-Aware Data Selection (CADS) method and naturally formulate it into a bilevel optimization framework, where the inner loop trains the model within the constraints of the computational budget on some selected subset of training data, while the outer loop optimizes data selection based on model evaluation. Our technical contributions lie in addressing two main challenges in solving this bilevel optimization problem: the expensive Hessian matrix estimation for outer-loop gradients and the computational burden of achieving inner-loop optimality during iterations. To solve the first issue, we propose a probabilistic reparameterization strategy and compute the gradient using a Hessian-free policy gradient estimator. To address the second challenge, we transform the inner optimization problem into a penalty term in the outer objective, further discovering that we only need to estimate the minimum of a one-dimensional loss to calculate the gradient, significantly improving efficiency. Extensive experiments show that our method achieves performance gains of up to 14.42% over baselines in vision and language benchmarks.
Authors: Adeel Safder
Abstract: Deep neural networks (DNNs) achieve remarkable performance but often suffer from overfitting due to their high capacity. We introduce Momentum-Adaptive Gradient Dropout (MAGDrop), a novel regularization method that dynamically adjusts dropout rates on activations based on current gradients and accumulated momentum, enhancing stability in non-convex optimization landscapes. To theoretically justify MAGDrop's effectiveness, we derive a non-asymptotic, computable PAC-Bayes generalization bound that accounts for its adaptive nature, achieving up to 29.2\% tighter bounds compared to standard approaches by leveraging momentum-driven perturbation control. Empirically, the activation-based MAGDrop achieves competitive performance on MNIST (99.52\%) and CIFAR-10 (92.03\%), with generalization gaps of 0.48\% and 6.52\%, respectively. We provide fully reproducible code and numerical computation of our bounds to validate our theoretical claims. Our work bridges theoretical insights and practical advancements, offering a robust framework for enhancing DNN generalization, making it suitable for high-stakes applications.
Authors: Wang Zixian
Abstract: Direct Preference Optimization (DPO) has become a standard for aligning models with human feedback, yet its reliance on hard, pairwise preferences makes it brittle to annotator noise and distribution shift. We propose Anchored Direct Preference Optimization (ADPO), a generalized framework that learns from soft, listwise supervision by anchoring policy updates to a reference model. Our key theoretical contribution is to show that this anchoring mechanism imposes an implicit trust region on the policy update, enforced by the softmax Fisher information metric. This provides a robust geometric interpretation for both fixed and dynamic anchor strategies. Our central empirical finding is a task-dependent tradeoff between anchor update strategies. Through controlled experiments across twelve scenarios and two MuJoCo environments, we demonstrate that (1) for online exploration in noisy environments, a dynamic anchor that tracks the learning policy is superior, improving performance by 5 to 11 percent over a fixed anchor; and (2) for offline distillation, a fixed anchor pointing to the teacher policy is dramatically more effective, achieving returns of 206.7 on HalfCheetah-v5 (387 percent of teacher) and 65.4 on Hopper-v5 (61 percent of teacher), while reducing KL divergence to the teacher by up to 5000 times compared with standard knowledge distillation. These findings offer clear, practical guidance for selecting anchor strategies and establish ADPO as a robust, unified framework for preference learning. Larger models further amplify ADPO's benefits (0.718 vs. 0.416 at hidden dimension 256), suggesting that anchoring acts as an effective trust-region regularizer. We release code and configurations to facilitate reproducibility.
Authors: Jiacheng Liu, Xinyu Wang, Yuqi Lin, Zhikai Wang, Peiru Wang, Peiliang Cai, Qinming Zhou, Zhengan Yan, Zexuan Yan, Zhengyi Shi, Chang Zou, Yue Ma, Linfeng Zhang
Abstract: Diffusion Models have become a cornerstone of modern generative AI for their exceptional generation quality and controllability. However, their inherent \textit{multi-step iterations} and \textit{complex backbone networks} lead to prohibitive computational overhead and generation latency, forming a major bottleneck for real-time applications. Although existing acceleration techniques have made progress, they still face challenges such as limited applicability, high training costs, or quality degradation. Against this backdrop, \textbf{Diffusion Caching} offers a promising training-free, architecture-agnostic, and efficient inference paradigm. Its core mechanism identifies and reuses intrinsic computational redundancies in the diffusion process. By enabling feature-level cross-step reuse and inter-layer scheduling, it reduces computation without modifying model parameters. This paper systematically reviews the theoretical foundations and evolution of Diffusion Caching and proposes a unified framework for its classification and analysis. Through comparative analysis of representative methods, we show that Diffusion Caching evolves from \textit{static reuse} to \textit{dynamic prediction}. This trend enhances caching flexibility across diverse tasks and enables integration with other acceleration techniques such as sampling optimization and model distillation, paving the way for a unified, efficient inference framework for future multimodal and interactive applications. We argue that this paradigm will become a key enabler of real-time and efficient generative AI, injecting new vitality into both theory and practice of \textit{Efficient Generative Intelligence}.
Authors: Liang Ye, Shengqin Chen, Jiazhu Dai
Abstract: The rapid progress of graph generation has raised new security concerns, particularly regarding backdoor vulnerabilities. While prior work has explored backdoor attacks in image diffusion and unconditional graph generation, conditional, especially text-guided graph generation remains largely unexamined. This paper proposes BadGraph, a backdoor attack method against latent diffusion models for text-guided graph generation. BadGraph leverages textual triggers to poison training data, covertly implanting backdoors that induce attacker-specified subgraphs during inference when triggers appear, while preserving normal performance on clean inputs. Extensive experiments on four benchmark datasets (PubChem, ChEBI-20, PCDes, MoMu) demonstrate the effectiveness and stealth of the attack: less than 10% poisoning rate can achieves 50% attack success rate, while 24% suffices for over 80% success rate, with negligible performance degradation on benign samples. Ablation studies further reveal that the backdoor is implanted during VAE and diffusion training rather than pretraining. These findings reveal the security vulnerabilities in latent diffusion models of text-guided graph generation, highlight the serious risks in models' applications such as drug discovery and underscore the need for robust defenses against the backdoor attack in such diffusion models.
Authors: Daniel M. Steinberg, Asiri Wijesinghe, Rafael Oliveira, Piotr Koniusz, Cheng Soon Ong, Edwin V. Bonilla
Abstract: We introduce active generation of Pareto sets (A-GPS), a new framework for online discrete black-box multi-objective optimization (MOO). A-GPS learns a generative model of the Pareto set that supports a-posteriori conditioning on user preferences. The method employs a class probability estimator (CPE) to predict non-dominance relations and to condition the generative model toward high-performing regions of the search space. We also show that this non-dominance CPE implicitly estimates the probability of hypervolume improvement (PHVI). To incorporate subjective trade-offs, A-GPS introduces preference direction vectors that encode user-specified preferences in objective space. At each iteration, the model is updated using both Pareto membership and alignment with these preference directions, producing an amortized generative model capable of sampling across the Pareto front without retraining. The result is a simple yet powerful approach that achieves high-quality Pareto set approximations, avoids explicit hypervolume computation, and flexibly captures user preferences. Empirical results on synthetic benchmarks and protein design tasks demonstrate strong sample efficiency and effective preference incorporation.
Authors: Federico Danieli, Pau Rodriguez, Miguel Sarabia, Xavier Suau, Luca Zappella
Abstract: Recurrent Neural Networks (RNNs) laid the foundation for sequence modeling, but their intrinsic sequential nature restricts parallel computation, creating a fundamental barrier to scaling. This has led to the dominance of parallelizable architectures like Transformers and, more recently, State Space Models (SSMs). While SSMs achieve efficient parallelization through structured linear recurrences, this linearity constraint limits their expressive power and precludes modeling complex, nonlinear sequence-wise dependencies. To address this, we present ParaRNN, a framework that breaks the sequence-parallelization barrier for nonlinear RNNs. Building on prior work, we cast the sequence of nonlinear recurrence relationships as a single system of equations, which we solve in parallel using Newton's iterations combined with custom parallel reductions. Our implementation achieves speedups of up to 665x over naive sequential application, allowing training nonlinear RNNs at unprecedented scales. To showcase this, we apply ParaRNN to adaptations of LSTM and GRU architectures, successfully training models of 7B parameters that attain perplexity comparable to similarly-sized Transformers and Mamba2 architectures. To accelerate research in efficient sequence modeling, we release the ParaRNN codebase as an open-source framework for automatic training-parallelization of nonlinear RNNs, enabling researchers and practitioners to explore new nonlinear RNN models at scale.
Authors: Mahavir Dabas, Tran Huynh, Nikhil Reddy Billa, Jiachen T. Wang, Peng Gao, Charith Peris, Yao Ma, Rahul Gupta, Ming Jin, Prateek Mittal, Ruoxi Jia
Abstract: Large language models remain vulnerable to jailbreak attacks that bypass safety guardrails to elicit harmful outputs. Defending against novel jailbreaks represents a critical challenge in AI safety. Adversarial training -- designed to make models robust against worst-case perturbations -- has been the dominant paradigm for adversarial robustness. However, due to optimization challenges and difficulties in defining realistic threat models, adversarial training methods often fail on newly developed jailbreaks in practice. This paper proposes a new paradigm for improving robustness against unseen jailbreaks, centered on the Adversarial D\'ej\`a Vu hypothesis: novel jailbreaks are not fundamentally new, but largely recombinations of adversarial skills from previous attacks. We study this hypothesis through a large-scale analysis of 32 attack papers published over two years. Using an automated pipeline, we extract and compress adversarial skills into a sparse dictionary of primitives, with LLMs generating human-readable descriptions. Our analysis reveals that unseen attacks can be effectively explained as sparse compositions of earlier skills, with explanatory power increasing monotonically as skill coverage grows. Guided by this insight, we introduce Adversarial Skill Compositional Training (ASCoT), which trains on diverse compositions of skill primitives rather than isolated attack instances. ASCoT substantially improves robustness to unseen attacks, including multi-turn jailbreaks, while maintaining low over-refusal rates. We also demonstrate that expanding adversarial skill coverage, not just data scale, is key to defending against novel attacks. \textcolor{red}{\textbf{Warning: This paper contains content that may be harmful or offensive in nature.
Authors: Yasas Senarath, Hemant Purohit
Abstract: User behavior on online platforms is evolving, reflecting real-world changes in how people post, whether it's helpful messages or hate speech. Models that learn to capture this content can experience a decrease in performance over time due to data drift, which can lead to ineffective behavioral analytics systems. However, fine-tuning such a model over time with new data can be detrimental due to catastrophic forgetting. Replay-based approaches in continual learning offer a simple yet efficient method to update such models, minimizing forgetting by maintaining a buffer of important training instances from past learned tasks. However, the main limitation of this approach is the fixed size of the buffer. External knowledge bases can be utilized to overcome this limitation through data augmentation. We propose a novel augmentation-based approach to incorporate external knowledge in the replay-based continual learning framework. We evaluate several strategies with three datasets from prior studies related to deviant behavior classification to assess the integration of external knowledge in continual learning and demonstrate that augmentation helps outperform baseline replay-based approaches.
Authors: Pritish Chakraborty, Indradyumna Roy, Soumen Chakrabarti, Abir De
Abstract: Retrieving graphs from a large corpus, that contain a subgraph isomorphic to a given query graph, is a core operation in many real-world applications. While recent multi-vector graph representations and scores based on set alignment and containment can provide accurate subgraph isomorphism tests, their use in retrieval remains limited by their need to score corpus graphs exhaustively. We introduce CORGII (Contextual Representation of Graphs for Inverted Indexing), a graph indexing framework in which, starting with a contextual dense graph representation, a differentiable discretization module computes sparse binary codes over a learned latent vocabulary. This text document-like representation allows us to leverage classic, highly optimized inverted indices, while supporting soft (vector) set containment scores. Pushing this paradigm further, we replace the classical, fixed impact weight of a `token' on a graph (such as TFIDF or BM25) with a data-driven, trainable impact weight. Finally, we explore token expansion to support multi-probing the index for smoother accuracy-efficiency tradeoffs. To our knowledge, CORGII is the first indexer of dense graph representations using discrete tokens mapping to efficient inverted lists. Extensive experiments show that CORGII provides better trade-offs between accuracy and efficiency, compared to several baselines.
Authors: Daniel Sin, Milad Toutounchian
Abstract: In our article, we describe a method for generating counterfactual explanations in high-dimensional spaces using four steps that involve fitting our dataset to a model, finding the decision boundary, determining constraints on the problem, and computing the closest point (counterfactual explanation) from that boundary. We propose a discretized approach where we find many discrete points on the boundary and then identify the closest feasible counterfactual explanation. This method, which we later call $\textit{Optimal Point for Boundary Approximation}$ (OPBA), applies binary search to find decision boundary points and then searches for the closest boundary point. Across four datasets of varying dimensionality, we show that our method can outperform current methods for counterfactual generation with reductions in distance between $5\%$ to $50\%$ in terms of the $L_2$ norm. Our method can also handle real-world constraints by restricting changes to immutable and categorical features, such as age, gender, sex, height, and other related characteristics such as the case for a health-based dataset. In terms of runtime, the OPBA algorithm generates decision boundary points on multiple orders of magnitude in the same given time when we compare to a grid-based approach. In general, our method provides a simple and effective model-agnostic method that can compute nearest feasible (i.e. realistic with constraints) counterfactual explanations. All of our results and code are available at: https://github.com/dsin85691/OPBA_For_Counterfactuals
Authors: Thaweerath Phisannupawong, Joshua Julian Damanik, Han-Lim Choi
Abstract: Flight delay prediction has become a key focus in air traffic management, as delays highlight inefficiencies that impact overall network performance. This paper presents a lightweight large language model-based multimodal flight delay prediction, formulated from the perspective of air traffic controllers monitoring aircraft delay after entering the terminal area. The approach integrates trajectory representations with textual aeronautical information, including flight information, weather reports, and aerodrome notices, by adapting trajectory data into the language modality to capture airspace conditions. The experiments show that the model consistently achieves sub-minute prediction error by effectively leveraging contextual information related to the sources of delay, fulfilling the operational standard for minute-level precision. The framework demonstrates that linguistic understanding, when combined with cross-modality adaptation of trajectory data, enhances delay prediction. Moreover, the approach shows practicality and potential scalability for real-world operations, supporting real-time updates that refine predictions upon receiving new operational information.
Authors: Marko Karbevski, Antonij Mijoski
Abstract: The Query, Key, Value weight triplet is a building block of current attention mechanisms in state-of-the-art LLMs. We theoretically investigate whether this triplet can be reduced, proving under simplifying assumptions that the Query weights are redundant, thereby reducing the number of non-embedding/lm-head parameters by over 8%. We validate the theory on full-complexity GPT-3 small architectures (with layer normalization, skip connections, and weight decay) trained from scratch, demonstrating that the reduced model achieves comparable validation loss to standard baselines. These findings motivate the investigation of the Query weight redundancy at scale.
Authors: Akira Tamamori
Abstract: This paper presents Two-Stage LKPLO, a novel multi-stage outlier detection framework that overcomes the coexisting limitations of conventional projection-based methods: their reliance on a fixed statistical metric and their assumption of a single data structure. Our framework uniquely synthesizes three key concepts: (1) a generalized loss-based outlyingness measure (PLO) that replaces the fixed metric with flexible, adaptive loss functions like our proposed SVM-like loss; (2) a global kernel PCA stage to linearize non-linear data structures; and (3) a subsequent local clustering stage to handle multi-modal distributions. Comprehensive 5-fold cross-validation experiments on 10 benchmark datasets, with automated hyperparameter optimization, demonstrate that Two-Stage LKPLO achieves state-of-the-art performance. It significantly outperforms strong baselines on datasets with challenging structures where existing methods fail, most notably on multi-cluster data (Optdigits) and complex, high-dimensional data (Arrhythmia). Furthermore, an ablation study empirically confirms that the synergistic combination of both the kernelization and localization stages is indispensable for its superior performance. This work contributes a powerful new tool for a significant class of outlier detection problems and underscores the importance of hybrid, multi-stage architectures.
Authors: He Yang, Fei Ren, Francesco Calabro, Hai-Sui Yu, Xiaohui Chen, Pei-Zhi Zhuang
Abstract: We are delighted to see the recent development of physics-informed extreme learning machine (PIELM) for its higher computational efficiency and accuracy compared to other physics-informed machine learning (PIML) paradigms. Since a comprehensive summary or review of PIELM is currently unavailable, we would like to take this opportunity to share our perspectives and experiences on this promising research direction. We can see that many efforts have been made to solve ordinary/partial differential equations (ODEs/PDEs) characterized by sharp gradients, nonlinearities, high-frequency behavior, hard constraints, uncertainty, multiphysics coupling, and interpretability. Despite these encouraging successes, many pressing challenges remain to be tackled, which also provides opportunities to develop more robust, interpretable, and generalizable PIELM frameworks for scientific and engineering applications.
Authors: Biyi Fang, Truong Vo, Kripa Rajshekhar, Diego Klabjan
Abstract: Traditional topic models such as Latent Dirichlet Allocation (LDA) have been widely used to uncover latent structures in text corpora, but they often struggle to integrate auxiliary information such as metadata, user attributes, or document labels. These limitations restrict their expressiveness, personalization, and interpretability. To address this, we propose nnLDA, a neural-augmented probabilistic topic model that dynamically incorporates side information through a neural prior mechanism. nnLDA models each document as a mixture of latent topics, where the prior over topic proportions is generated by a neural network conditioned on auxiliary features. This design allows the model to capture complex nonlinear interactions between side information and topic distributions that static Dirichlet priors cannot represent. We develop a stochastic variational Expectation-Maximization algorithm to jointly optimize the neural and probabilistic components. Across multiple benchmark datasets, nnLDA consistently outperforms LDA and Dirichlet-Multinomial Regression in topic coherence, perplexity, and downstream classification. These results highlight the benefits of combining neural representation learning with probabilistic topic modeling in settings where side information is available.
Authors: Beomhan Baek, Minhak Song, Chulhee Yun
Abstract: Adam [Kingma and Ba, 2015] is the de facto optimizer in deep learning, yet its theoretical understanding remains limited. Prior analyses show that Adam favors solutions aligned with $\ell_\infty$-geometry, but these results are restricted to the full-batch regime. In this work, we study the implicit bias of incremental Adam (using one sample per step) for logistic regression on linearly separable data, and we show that its bias can deviate from the full-batch behavior. To illustrate this, we construct a class of structured datasets where incremental Adam provably converges to the $\ell_2$-max-margin classifier, in contrast to the $\ell_\infty$-max-margin bias of full-batch Adam. For general datasets, we develop a proxy algorithm that captures the limiting behavior of incremental Adam as $\beta_2 \to 1$ and we characterize its convergence direction via a data-dependent dual fixed-point formulation. Finally, we prove that, unlike Adam, Signum [Bernstein et al., 2018] converges to the $\ell_\infty$-max-margin classifier for any batch size by taking $\beta$ close enough to 1. Overall, our results highlight that the implicit bias of Adam crucially depends on both the batching scheme and the dataset, while Signum remains invariant.
Authors: Viktor Nilsson, Pierre Nyquist
Abstract: Finding equilibrium points in continuous minmax games has become a key problem within machine learning, in part due to its connection to the training of generative adversarial networks and reinforcement learning. Because of existence and robustness issues, recent developments have shifted from pure equilibria to focusing on mixed equilibrium points. In this work we consider a method for finding mixed equilibria in two-layer zero-sum games based on entropic regularisation, where the two competing strategies are represented by two sets of interacting particles. We show that the sequence of empirical measures of the particle system satisfies a large deviation principle as the number of particles grows to infinity, and how this implies convergence of the empirical measure and the associated Nikaid\^o-Isoda error, complementing existing law of large numbers results.
Authors: Antonia A. L. Dos Santos, Danilo A. Sarti, Rafael A. Moral, Andrew C. Parnell
Abstract: We propose a Bayesian tensor regression model to accommodate the effect of multiple factors on phenotype prediction. We adopt a set of prior distributions that resolve identifiability issues that may arise between the parameters in the model. Further, we incorporate a spike-and-slab structure that identifies which interactions are relevant for inclusion in the linear predictor, even when they form a subset of the available variables. Simulation experiments show that our method outperforms previous related models and machine learning algorithms under different sample sizes and degrees of complexity. We further explore the applicability of our model by analysing real-world data related to wheat production across Ireland from 2010 to 2019. Our model performs competitively and overcomes key limitations found in other analogous approaches. Finally, we adapt a set of visualisations for the posterior distribution of the tensor effects that facilitate the identification of optimal interactions between the tensor variables, whilst accounting for the uncertainty in the posterior distribution.
Authors: Xavier Daull, Patrice Bellot, Emmanuel Bruno, Vincent Martin, Elisabeth Murisasco
Abstract: This paper reviews the state-of-the-art of large language models (LLM) architectures and strategies for "complex" question-answering with a focus on hybrid architectures. LLM based chatbot services have allowed anyone to grasp the potential of LLM to solve many common problems, but soon discovered their limitations for complex questions. Addressing more specific, complex questions (e.g., "What is the best mix of power-generation methods to reduce climate change ?") often requires specialized architectures, domain knowledge, new skills, decomposition and multi-step resolution, deep reasoning, sensitive data protection, explainability, and human-in-the-loop processes. Therefore, we review: (1) necessary skills and tasks for handling complex questions and common LLM limits to overcome; (2) dataset, cost functions and evaluation metrics for measuring and improving (e.g. accuracy, explainability, fairness, robustness, groundedness, faithfulness, toxicity...); (3) family of solutions to overcome LLM limitations by (a) training and reinforcement (b) hybridization, (c) prompting, (d) agentic-architectures (agents, tools) and extended reasoning.
Authors: Gil Kur, Eli Putterman, Alexander Rakhlin
Abstract: It is well known that Empirical Risk Minimization (ERM) may attain minimax suboptimal rates in terms of the mean squared error (Birg\'e and Massart, 1993). In this paper, we prove that, under relatively mild assumptions, the suboptimality of ERM must be due to its large bias. Namely, the variance error term of ERM is bounded by the minimax rate. In the fixed design setting, we provide an elementary proof of this result using the probabilistic method. Then, we extend our proof to the random design setting for various models. In addition, we provide a simple proof of Chatterjee's admissibility theorem (Chatterjee, 2014, Theorem 1.4), which states that in the fixed design setting, ERM cannot be ruled out as an optimal method, and then we extend this result to the random design setting. We also show that our estimates imply the stability of ERM, complementing the main result of Caponnetto and Rakhlin (2006) for non-Donsker classes. Finally, we highlight the somewhat irregular nature of the loss landscape of ERM in the non-Donsker regime, by showing that functions can be close to ERM, in terms of $L_2$ distance, while still being far from almost-minimizers of the empirical loss.
Authors: Yuhang Hu, Judah Goldfeder, Zhizhuo Zhang, Xinyue Zhu, Ruibo Liu, Philippe Wyder, Jiong Lin, Hod Lipson
Abstract: For robots to truly collaborate and assist humans, they must understand not only logic and instructions, but also the subtle emotions, aesthetics, and feelings that define our humanity. Human art and aesthetics are among the most elusive concepts-often difficult even for people to articulate-and without grasping these fundamentals, robots will be unable to help in many spheres of daily life. Consider the long-promised robotic butler: automating domestic chores demands more than motion planning. It requires an internal model of cleanliness and tidiness-a challenge largely unexplored by AI. To bridge this gap, we propose an approach that equips domestic robots to perform simple tidying tasks via knolling, the practice of arranging scattered items into neat, space-efficient layouts. Unlike the uniformity of industrial settings, household environments feature diverse objects and highly subjective notions of tidiness. Drawing inspiration from NLP, we treat knolling as a sequential prediction problem and employ a transformer based model to forecast each object's placement. Our method learns a generalizable concept of tidiness, generates diverse solutions adaptable to varying object sets, and incorporates human preferences for personalized arrangements. This work represents a step forward in building robots that internalize human aesthetic sense and can genuinely co-create in our living spaces.
Authors: Saurav Agarwal, Ramya Muthukrishnan, Walker Gosrich, Vijay Kumar, Alejandro Ribeiro
Abstract: Coverage control is the problem of navigating a robot swarm to collaboratively monitor features or a phenomenon of interest not known a priori. The problem is challenging in decentralized settings with robots that have limited communication and sensing capabilities. We propose a learnable Perception-Action-Communication (LPAC) architecture for the problem, wherein a convolutional neural network (CNN) processes localized perception; a graph neural network (GNN) facilitates robot communications; finally, a shallow multi-layer perceptron (MLP) computes robot actions. The GNN enables collaboration in the robot swarm by computing what information to communicate with nearby robots and how to incorporate received information. Evaluations show that the LPAC models -- trained using imitation learning -- outperform standard decentralized and centralized coverage control algorithms. The learned policy generalizes to environments different from the training dataset, transfers to larger environments with more robots, and is robust to noisy position estimates. The results indicate the suitability of LPAC architectures for decentralized navigation in robot swarms to achieve collaborative behavior.
Authors: Xin Hao, Changyang She, Phee Lep Yeoh, Yuhong Liu, Branka Vucetic, Yonghui Li
Abstract: In this paper, we develop a deep learning-based bandwidth allocation policy that is: 1) scalable with the number of users and 2) transferable to different communication scenarios, such as non-stationary wireless channels, different quality-of-service (QoS) requirements, and dynamically available resources. To support scalability, the bandwidth allocation policy is represented by a graph neural network (GNN), with which the number of training parameters does not change with the number of users. To enable the generalization of the GNN, we develop a hybrid-task meta-learning (HML) algorithm that trains the initial parameters of the GNN with different communication scenarios during meta-training. Next, during meta-testing, a few samples are used to fine-tune the GNN with unseen communication scenarios. Simulation results demonstrate that our HML approach can improve the initial performance by 8.79%, and sample efficiency by 73%, compared with existing benchmarks. After fine-tuning, our near-optimal GNN-based policy can achieve close to the same reward with much lower inference complexity compared to the optimal policy obtained using iterative optimization. Numerical results validate that our HML can reduce the computation time by approximately 200 to 2000 times than the optimal iterative algorithm.
Authors: Ziliang Zhang, Huaming Yu, Danqin Ren, Chenyu Zhang, Minghua Sun, Xin Qi
Abstract: This study presents OceanCastNet (OCN), a machine learning approach for wave forecasting that incorporates wind and wave fields to predict significant wave height, mean wave period, and mean wave direction.We evaluate OCN's performance against the operational ECWAM model using two independent datasets: NDBC buoy and Jason-3 satellite observations. NDBC station validation indicates OCN performs better at 24 stations compared to ECWAM's 10 stations, and Jason-3 satellite validation confirms similar accuracy across 228-hour forecasts. OCN successfully captures wave patterns during extreme weather conditions, demonstrated through Typhoon Goni with prediction errors typically within $\pm$0.5 m. The approach also offers computational efficiency advantages. The results suggest that machine learning approaches can achieve performance comparable to conventional wave forecasting systems for operational wave prediction applications.
Authors: Felix Chalumeau, Refiloe Shabe, Noah De Nicola, Arnu Pretorius, Thomas D. Barrett, Nathan Grinsztajn
Abstract: Routing Problems are central to many real-world applications, yet remain challenging due to their (NP-)hard nature. Amongst existing approaches, heuristics often offer the best trade-off between quality and scalability, making them suitable for industrial use. While Reinforcement Learning (RL) offers a flexible framework for designing heuristics, its adoption over handcrafted heuristics remains incomplete. Existing learned methods still lack the ability to adapt to specific instances and fully leverage the available computational budget. Current best methods either rely on a collection of pre-trained policies, or on RL fine-tuning; hence failing to fully utilize newly available information within the constraints of the budget. In response, we present MEMENTO, an approach that leverages memory to improve the search of neural solvers at inference. MEMENTO leverages online data collected across repeated attempts to dynamically adjust the action distribution based on the outcome of previous decisions. We validate its effectiveness on the Traveling Salesman and Capacitated Vehicle Routing problems, demonstrating its superiority over tree-search and policy-gradient fine-tuning; and showing that it can be zero-shot combined with diversity-based solvers. We successfully train all RL auto-regressive solvers on large instances, and verify MEMENTO's scalability and data-efficiency: pushing the state-of-the-art on 11 out of 12 evaluated tasks.
Authors: Yahya Saleh, Armin Iske
Abstract: Let $C_h$ be a composition operator mapping $L^2(\Omega_1)$ into $L^2(\Omega_2)$ for some open sets $\Omega_1, \Omega_2 \subseteq \mathbb{R}^n$. We characterize the mappings $h$ that transform Riesz bases of $L^2(\Omega_1)$ into Riesz bases of $L^2(\Omega_2)$. Restricting our analysis to differentiable mappings, we demonstrate that mappings $h$ that preserve Riesz bases have Jacobian determinants that are bounded away from zero and infinity. We discuss implications of these results for approximation theory, highlighting the potential of using bijective neural networks to construct Riesz bases with favorable approximation properties.
Authors: Alejandro Rodriguez-Garcia, Anindya Ghosh, Jie Mei, Srikanth Ramaswamy
Abstract: Recent progress in artificial intelligence (AI) has been driven by insights from physics and neuroscience, particularly through the development of artificial neural networks (ANNs) capable of complex cognitive tasks such as vision and language processing. Despite these advances, they struggle with continual learning, adaptable knowledge transfer, robustness, and resource efficiency -- capabilities that biological systems handle seamlessly. Specifically, neuromorphic systems and artificial neural networks often overlook two key biophysical properties of neural circuits: neuronal diversity and cell-specific neuromodulation. These mechanisms, essential for regulating dynamic learning across brain scales, allow neuromodulators to introduce degeneracy in biological neural networks, ensuring stability and adaptability under changing conditions. In this article, we summarize recent bioinspired models, learning rules, and architectures, and propose a framework for augmenting ANNs, which has the potential to bridge the gap between neuroscience and AI through neurobiological first principles. Our proposed dual-framework approach leverages spiking neural networks to emulate diverse spiking behaviors and dendritic compartmental dynamics, thereby simulating the morphological and functional diversity of neuronal computations. Finally, we outline how integrating these biophysical principles into task-driven spiking neural networks and neuromorphic systems provides scalable solutions for continual learning, adaptability, robustness, and resource-efficiency. Additionally, this approach will not only provide insights into how emergent behaviors arise in neural networks but also catalyze the development of more efficient, reliable, and intelligent neuromorphic systems and robotic agents.
Authors: Yehonathan Refael, Adam Hakim, Lev Greenberg, Satya Lokam, Tal Aviv, Ben Fishman, Shachar Seidman, Racchit Jain, Jay Tenenbaum
Abstract: Large language models (LLMs) have recently seen widespread adoption in both academia and industry. As these models grow, they become valuable intellectual property (IP), reflecting substantial investments by their owners. The high cost of cloud-based deployment has spurred interest in running models on edge devices, but this risks exposing parameters to theft and unauthorized use. Existing approaches to protect model IP on the edge trade off practicality, accuracy, or deployment requirements. We introduce SLIP, a hybrid inference algorithm designed to protect edge-deployed models from theft. SLIP is, to our knowledge, the first hybrid protocol that is both practical for real-world applications and provably secure, while incurring zero accuracy degradation and minimal latency overhead. It partitions the model across two computing resources: one secure but expensive, and one cost-effective but vulnerable. Using matrix decomposition, the secure resource retains the most sensitive portion of the model's IP while performing only a small fraction of the computation; the vulnerable resource executes the remainder. The protocol includes security guarantees that prevent attackers from using the partition to infer the protected information. Finally, we present experimental results that demonstrate the robustness and effectiveness of our method, positioning it as a compelling solution for protecting LLMs.
Authors: Aske Plaat, Annie Wong, Suzan Verberne, Joost Broekens, Niki van Stein, Thomas Back
Abstract: Large language models (LLMs) with billions of parameters exhibit in-context learning abilities, enabling few-shot learning on tasks that the model was not specifically trained for. Traditional models achieve breakthrough performance on language tasks, but do not perform well on basic reasoning benchmarks. However, a new in-context learning approach, Chain-of-thought, has demonstrated strong multi-step reasoning abilities on these benchmarks. The research on LLM reasoning abilities started with the question whether LLMs can solve grade school math word problems, and has expanded to other tasks in the past few years. This article reviews the field of multi-step reasoning with LLMs. We propose a taxonomy that identifies different ways to generate, evaluate, and control multi-step reasoning. We provide an in-depth coverage of core approaches and open problems, and we propose a research agenda for the near future. We find that multi-step reasoning approaches have progressed beyond math word problems, and can now successfully solve challenges in logic, combinatorial games, and robotics, sometimes by first generating code that is then executed by external tools. Many studies in multi-step methods use reinforcement learning for finetuning, external optimization loops, in-context reinforcement learning, and self-reflection.
Authors: Jooyoung Lee, Se Yoon Jeong, Munchurl Kim
Abstract: Unlike fixed- or variable-rate image coding, progressive image coding (PIC) aims to compress various qualities of images into a single bitstream, increasing the versatility of bitstream utilization and providing high compression efficiency compared to simulcast compression. Research on neural network (NN)-based PIC is in its early stages, mainly focusing on applying varying quantization step sizes to the transformed latent representations in a hierarchical manner. These approaches are designed to compress only the progressively added information as the quality improves, considering that a wider quantization interval for lower-quality compression includes multiple narrower sub-intervals for higher-quality compression. However, the existing methods are based on handcrafted quantization hierarchies, resulting in sub-optimal compression efficiency. In this paper, we propose an NN-based progressive coding method that firstly utilizes learned quantization step sizes via learning for each quantization layer. We also incorporate selective compression with which only the essential representation components are compressed for each quantization layer. We demonstrate that our method achieves significantly higher coding efficiency than the existing approaches with decreased decoding time and reduced model size. The source code is publicly available at https://github.com/JooyoungLeeETRI/DeepHQ
Authors: Oleksandr Borysenko, Mykhailo Bratchenko, Ilya Lukin, Mykola Luhanko, Ihor Omelchenko, Andrii Sotnikov, Alessandro Lomi
Abstract: A Quantum Natural Gradient (QNG) algorithm for optimization of variational quantum circuits has been proposed recently. In this study, we employ the Langevin equation with a QNG stochastic force to demonstrate that its discrete-time solution gives a generalized form of the above-specified algorithm, which we call Momentum-QNG. Similar to other optimization algorithms with the momentum term, such as the Stochastic Gradient Descent with momentum, RMSProp with momentum and Adam, Momentum-QNG is more effective to escape local minima and plateaus in the variational parameter space and, therefore, demonstrates an improved performance compared to the basic QNG. In this paper we benchmark Momentum-QNG together with the basic QNG, Adam and Momentum optimizers and explore its convergence behaviour. Among the benchmarking problems studied, the best result is obtained for the quantum Sherrington-Kirkpatrick model in the strong spin glass regime. Our open-source code is available at https://github.com/borbysh/Momentum-QNG
Authors: Zhizhuo Kou, Holam Yu, Junyu Luo, Jingshu Peng, Xujia Li, Chengzhong Liu, Juntao Dai, Lei Chen, Sirui Han, Yike Guo
Abstract: We present a novel three-stage framework leveraging Large Language Models (LLMs) within a risk-aware multi-agent system for automate strategy finding in quantitative finance. Our approach addresses the brittleness of traditional deep learning models in financial applications by: employing prompt-engineered LLMs to generate executable alpha factor candidates across diverse financial data, implementing multimodal agent-based evaluation that filters factors based on market status, predictive quality while maintaining category balance, and deploying dynamic weight optimization that adapts to market conditions. Experimental results demonstrate the robust performance of the strategy in Chinese & US market regimes compared to established benchmarks. Our work extends LLMs capabilities to quantitative trading, providing a scalable architecture for financial signal extraction and portfolio construction. The overall framework significantly outperforms all benchmarks with 53.17% cumulative return on SSE50 (Jan 2023 to Jan 2024), demonstrating superior risk-adjusted performance and downside protection on the market.
Authors: Md Tanvirul Alam, Dipkamal Bhusal, Nidhi Rastogi
Abstract: Static feature-based Android malware detection using machine learning (ML) remains critical due to its scalability and efficiency. However, existing approaches often overlook security-critical reproducibility concerns, such as dataset duplication, inadequate hyperparameter tuning, and variance from random initialization. This can significantly compromise the practical effectiveness of these systems. In this paper, we systematically investigate these challenges by proposing a more rigorous methodology for model selection and evaluation. Using two widely used datasets, Drebin and APIGraph, we evaluate six ML models of varying complexity under both offline and continuous active learning settings. Our analysis demonstrates that, contrary to popular belief, well-tuned, simpler models, particularly tree-based methods like XGBoost, consistently outperform more complex neural networks, especially when duplicates are removed. To promote transparency and reproducibility, we open-source our codebase, which is extensible for integrating new models and datasets, facilitating reproducible security research.
Authors: Akhilesh Aravapalli, Mounika Marreddy, Radhika Mamidi, Manish Gupta, Subba Reddy Oota
Abstract: Transformer-based models have revolutionized the field of natural language processing. To understand why they perform so well and to assess their reliability, several studies have focused on questions such as: Which linguistic properties are encoded by these models, and to what extent? How robust are these models in encoding linguistic properties when faced with perturbations in the input text? However, these studies have mainly focused on BERT and the English language. In this paper, we investigate similar questions regarding encoding capability and robustness for 8 linguistic properties across 13 different perturbations in 6 Indic languages, using 9 multilingual Transformer models (7 universal and 2 Indic-specific). To conduct this study, we introduce a novel multilingual benchmark dataset, IndicSentEval, containing approximately $\sim$47K sentences. Surprisingly, our probing analysis of surface, syntactic, and semantic properties reveals that while almost all multilingual models demonstrate consistent encoding performance for English, they show mixed results for Indic languages. As expected, Indic-specific multilingual models capture linguistic properties in Indic languages better than universal models. Intriguingly, universal models broadly exhibit better robustness compared to Indic-specific models, particularly under perturbations such as dropping both nouns and verbs, dropping only verbs, or keeping only nouns. Overall, this study provides valuable insights into probing and perturbation-specific strengths and weaknesses of popular multilingual Transformer-based models for different Indic languages. We make our code and dataset publicly available [https://github.com/aforakhilesh/IndicBertology].
Authors: Zhiwei Li, Guodong Long, Jing Jiang, Chengqi Zhang, Qiang Yang
Abstract: Applying large pre-trained Vision-Language Models to recommendation is a burgeoning field, a direction we term Vision-Language-Recommendation (VLR). Bringing VLR to user-oriented on-device intelligence within a federated learning framework is a crucial step for enhancing user privacy and delivering personalized experiences. This paper introduces FedVLR, a federated VLR framework specially designed for user-specific personalized fusion of vision-language representations. At its core is a novel bi-level fusion mechanism: The server-side multi-view fusion module first generates a diverse set of pre-fused multimodal views. Subsequently, each client employs a user-specific mixture-of-expert mechanism to adaptively integrate these views based on individual user interaction history. This designed lightweight personalized fusion module provides an efficient solution to implement a federated VLR system. The effectiveness of our proposed FedVLR has been validated on seven benchmark datasets.
Authors: Shashank Kirtania, Naman Gupta, Priyanshu Gupta, Krishna Kariya, Sumit Gulwani, Arun Iyer, Suresh Parthasarathy, Arjun Radhakrishna, Sriram K. Rajamani, Gustavo Soares
Abstract: Large Language Models (LLMs) often generate incorrect or outdated information, especially in low-resource settings or when dealing with private data. To address this, Retrieval-Augmented Generation (RAG) uses external knowledge bases (KBs), but these can also suffer from inaccuracies. We introduce STACKFEED, a novel Structured Textual Actor-Critic Knowledge base editing with FEEDback approach that iteratively refines the KB based on expert feedback using a multi-actor, centralized critic reinforcement learning framework. STACKFEED defines a ReACT actor agent on each document to perform structured edits based on document specific targeted instructions. Experimental results showcase that STACKFEED significantly improves KB quality and performance of the RAG system. We evaluate STACKFEED on low-resource programming problems, modified python packaged and factual question-answering tasks.
Authors: Charles C. Margossian, Lawrence K. Saul
Abstract: Given an intractable target density $p$, variational inference (VI) attempts to find the best approximation $q$ from a tractable family $Q$. This is typically done by minimizing the exclusive Kullback-Leibler divergence, $\text{KL}(q||p)$. In practice, $Q$ is not rich enough to contain $p$, and the approximation is misspecified even when it is a unique global minimizer of $\text{KL}(q||p)$. In this paper, we analyze the robustness of VI to these misspecifications when $p$ exhibits certain symmetries and $Q$ is a location-scale family that shares these symmetries. We prove strong guarantees for VI not only under mild regularity conditions but also in the face of severe misspecifications. Namely, we show that (i) VI recovers the mean of $p$ when $p$ exhibits an \textit{even} symmetry, and (ii) it recovers the correlation matrix of $p$ when in addition~$p$ exhibits an \textit{elliptical} symmetry. These guarantees hold for the mean even when $q$ is factorized and $p$ is not, and for the correlation matrix even when~$q$ and~$p$ behave differently in their tails. We analyze various regimes of Bayesian inference where these symmetries are useful idealizations, and we also investigate experimentally how VI behaves in their absence.
Authors: Baiyue Wang, Anthony Bloch
Abstract: We consider learning nonholonomic dynamical systems while discovering the constraints, and describe in detail the case of the rolling disk. A nonholonomic system is a system subject to nonholonomic constraints. Unlike holonomic constraints, nonholonomic constraints do not define a sub-manifold on the configuration space. Therefore, the inverse problem of finding the constraints has to involve the tangent bundle. This paper discusses a general procedure to learn the dynamics of a nonholonomic system through Hamel's formalism, while discovering the system constraint by parameterizing it, given the data set of discrete trajectories on the tangent bundle $TQ$. We prove that there is a local minimum for convergence of the network. We also preserve symmetry of the system by reducing the Lagrangian to the Lie algebra of the selected group.
Authors: Kadek Hendrawan Palgunadi, Andreas Bergmeister, Andrea Bosisio, Laura Ermert, Maria Koroni, Nathana\"el Perraudin, Simon Dirmeier, Men-Andrin Meier
Abstract: Accurate prediction and synthesis of seismic waveforms are crucial for seismic-hazard assessment and earthquake-resistant infrastructure design. Existing prediction methods, such as ground-motion models and physics-based wave-field simulations, often fail to capture the full complexity of seismic wavefields, particularly at higher frequencies. This study introduces HighFEM, a novel, computationally efficient, and scalable (i.e., capable of generating many seismograms simultaneously) generative model for high-frequency seismic-waveform generation. Our approach leverages a spectrogram representation of the seismic-waveform data, which is reduced to a lower-dimensional manifold via an autoencoder. A state-of-the-art diffusion model is trained to generate this latent representation conditioned on key input parameters: earthquake magnitude, recording distance, site conditions, hypocenter depth, and azimuthal gap. The model generates waveforms with frequency content up to 50 Hz. Any scalar ground-motion statistic, such as peak ground-motion amplitudes and spectral accelerations, can be readily derived from the synthesized waveforms. We validate our model using commonly employed seismological metrics and performance metrics from image-generation studies. Our results demonstrate that the openly available model can generate realistic high-frequency seismic waveforms across a wide range of input parameters, even in data-sparse regions. For the scalar ground-motion statistics commonly used in seismic-hazard and earthquake-engineering studies, we show that our model accurately reproduces both the median trends of the real data and their variability. To evaluate and compare the growing number of these and similar Generative Waveform Models (GWMs), we argue that they should be openly available and included in community ground-motion-model evaluation efforts.
Authors: Mou\"in Ben Ammar, David Brellmann, Arturo Mendoza, Antoine Manzanera, Gianni Franchi
Abstract: Out-of-distribution (OOD) detection is essential for ensuring the reliability and safety of machine learning systems. In recent years, it has received increasing attention, particularly through post-hoc detection and training-based methods. In this paper, we focus on post-hoc OOD detection, which enables identifying OOD samples without altering the model's training procedure or objective. Our primary goal is to investigate the relationship between model capacity and its OOD detection performance. Specifically, we aim to answer the following question: Does the Double Descent phenomenon manifest in post-hoc OOD detection? This question is crucial, as it can reveal whether overparameterization, which is already known to benefit generalization, can also enhance OOD detection. Despite the growing interest in these topics by the classic supervised machine learning community, this intersection remains unexplored for OOD detection. We empirically demonstrate that the Double Descent effect does indeed appear in post-hoc OOD detection. Furthermore, we provide theoretical insights to explain why this phenomenon emerges in such setting. Finally, we show that the overparameterized regime does not yield superior results consistently, and we propose a method to identify the optimal regime for OOD detection based on our observations.
Authors: Yadh Hafsi, Edoardo Vittori
Abstract: This study investigates the development of an optimal execution strategy through reinforcement learning, aiming to determine the most effective approach for traders to buy and sell inventory within a finite time horizon. Our proposed model leverages input features derived from the current state of the limit order book and operates at a high frequency to maximize control. To simulate this environment and overcome the limitations associated with relying on historical data, we utilize the multi-agent market simulator ABIDES, which provides a diverse range of depth levels within the limit order book. We present a custom MDP formulation followed by the results of our methodology and benchmark the performance against standard execution strategies. Results show that the reinforcement learning agent outperforms standard strategies and offers a practical foundation for real-world trading applications.
Authors: Sanath Budakegowdanadoddi Nagaraju, Brian Bernhard Moser, Tobias Christian Nauen, Stanislav Frolov, Federico Raue, Andreas Dengel
Abstract: Transformer-based architectures have recently advanced the image reconstruction quality of super-resolution (SR) models. Yet, their scalability remains limited by quadratic attention costs and coarse patch embeddings that weaken pixel-level fidelity. We propose TaylorIR, a plug-and-play framework that enforces 1x1 patch embeddings for true pixel-wise reasoning and replaces conventional self-attention with TaylorShift, a Taylor-series-based attention mechanism enabling full token interactions with near-linear complexity. Across multiple SR benchmarks, TaylorIR delivers state-of-the-art performance while reducing memory consumption by up to 60%, effectively bridging the gap between fine-grained detail restoration and efficient transformer scaling.
Authors: Quang P. M. Pham, Khoi T. N. Nguyen, Lan C. Ngo, Truong Do, Dezhen Song, Truong-Son Hy
Abstract: Scene graphs have proven to be highly effective for various scene understanding tasks due to their compact and explicit representation of relational information. However, current methods often overlook the critical importance of preserving symmetry when generating scene graphs from 3D point clouds, which can lead to reduced accuracy and robustness, particularly when dealing with noisy, multi-view data. Furthermore, a major limitation of prior approaches is the lack of temporal modeling to capture time-dependent relationships among dynamically evolving entities in a scene. To address these challenges, we propose Temporal Equivariant Scene Graph Neural Network (TESGNN), consisting of two key components: (1) an Equivariant Scene Graph Neural Network (ESGNN), which extracts information from 3D point clouds to generate scene graph while preserving crucial symmetry properties, and (2) a Temporal Graph Matching Network, which fuses scene graphs generated by ESGNN across multiple time sequences into a unified global representation using an approximate graph-matching algorithm. Our combined architecture TESGNN shown to be effective compared to existing methods in scene graph generation, achieving higher accuracy and faster training convergence. Moreover, we show that leveraging the symmetry-preserving property produces a more stable and accurate global scene representation compared to existing approaches. Finally, it is computationally efficient and easily implementable using existing frameworks, making it well-suited for real-time applications in robotics and computer vision. This approach paves the way for more robust and scalable solutions to complex multi-view scene understanding challenges. Our source code is publicly available at: https://github.com/HySonLab/TESGraph
Authors: Hiep Vo Dang, Phong C. H. Nguyen
Abstract: Reconstructing high-fidelity fluid flow fields from sparse sensor measurements is vital for many science and engineering applications but remains challenging because of dimensional disparities between state and observational spaces. Due to such dimensional differences, the measurement operator becomes ill-conditioned and non-invertible, making the reconstruction of flow fields from sensor measurements extremely difficult. Although sparse optimization and machine learning address the above problems to some extent, questions about their generalization and efficiency remain, particularly regarding the discretization dependence of these models. In this context, deep operator learning offers a better solution as this approach models mappings between infinite-dimensional functional spaces, enabling superior generalization and discretization-independent reconstruction. We introduce FLRONet, a deep operator learning framework that is trained to reconstruct fluid flow fields from sparse sensor measurements. FLRONet employs a branch-trunk network architecture to represent the inverse measurement operator that maps sensor observations to the original flow field, a continuous function of both space and time. Validation performed on the CFDBench dataset has demonstrated that FLRONet consistently achieves high levels of reconstruction accuracy and robustness, even in scenarios where sensor measurements are inaccurate or missing. Furthermore, the operator learning approach endows FLRONet with the capability to perform zero-shot super-resolution in both spatial and temporal domains, offering a solution for rapid reconstruction of high-fidelity flow fields.
Authors: Jingruo Sun, Wenzhi Gao, Ellen Vitercik, Yinyu Ye
Abstract: Online linear programming (OLP) has found broad applications in revenue management and resource allocation. State-of-the-art OLP algorithms achieve low regret by repeatedly solving linear programming (LP) subproblems that incorporate updated resource information. However, LP-based methods are computationally expensive and often inefficient for large-scale applications. In contrast, recent first-order OLP algorithms are more computationally efficient but typically suffer from worse regret guarantees. To address these shortcomings, we propose a new algorithm that combines the strengths of LP-based and first-order OLP methods. The algorithm re-solves the LP subproblems periodically at a predefined frequency $f$ and uses the latest dual prices to guide online decision-making. In addition, a first-order method runs in parallel during each interval between LP re-solves, smoothing resource consumption. Our algorithm achieves $\mathscr{O}(\log (T/f) + \sqrt{f})$ regret, delivering a "wait-less" online decision-making process that balances the computational efficiency of first-order methods and the superior regret guarantee of LP-based methods.
Authors: Jiawei Zhang
Abstract: This research applies Harold Demsetz's concept of the nirvana approach to the realm of AI governance and debunks three common fallacies in various AI policy proposals--"the grass is always greener on the other side," "free lunch," and "the people could be different." Through this, I expose fundamental flaws in the current AI regulatory proposal. First, some commentators intuitively believe that people are more reliable than machines and that government works better in risk control than companies' self-regulation, but they do not fully compare the differences between the status quo and the proposed replacements. Second, when proposing some regulatory tools, some policymakers and researchers do not realize and even gloss over the fact that harms and costs are also inherent in their proposals. Third, some policy proposals are initiated based on a false comparison between the AI-driven world, where AI does lead to some risks, and an entirely idealized world, where no risk exists at all. However, the appropriate approach is to compare the world where AI causes risks to the real world where risks are everywhere, but people can live well with these risks. The prevalence of these fallacies in AI governance underscores a broader issue: the tendency to idealize potential solutions without fully considering their real-world implications. This idealization can lead to regulatory proposals that are not only impractical but potentially harmful to innovation and societal progress.
Authors: Junan Zhang, Jing Yang, Zihao Fang, Yuancheng Wang, Zehua Zhang, Zhuo Wang, Fan Fan, Zhizheng Wu
Abstract: We introduce AnyEnhance, a unified generative model for voice enhancement that processes both speech and singing voices. Based on a masked generative model, AnyEnhance is capable of handling both speech and singing voices, supporting a wide range of enhancement tasks including denoising, dereverberation, declipping, super-resolution, and target speaker extraction, all simultaneously and without fine-tuning. AnyEnhance introduces a prompt-guidance mechanism for in-context learning, which allows the model to natively accept a reference speaker's timbre. In this way, it could boost enhancement performance when a reference audio is available and enable the target speaker extraction task without altering the underlying architecture. Moreover, we also introduce a self-critic mechanism into the generative process for masked generative models, yielding higher-quality outputs through iterative self-assessment and refinement. Extensive experiments on various enhancement tasks demonstrate AnyEnhance outperforms existing methods in terms of both objective metrics and subjective listening tests. Demo audios are publicly available at https://amphionspace.github.io/anyenhance. An open-source implementation is provided at https://github.com/viewfinder-annn/anyenhance-v1-ccf-aatc.
URLs: https://amphionspace.github.io/anyenhance., https://github.com/viewfinder-annn/anyenhance-v1-ccf-aatc.
Authors: Tobias Dietz, Brian B. Moser, Tobias Nauen, Federico Raue, Stanislav Frolov, Andreas Dengel
Abstract: Dataset distillation aims to compress large datasets into compact yet highly informative subsets that preserve the training behavior of the original data. While this concept has gained traction in classification, its potential for image Super-Resolution (SR) remains largely untapped. In this work, we conduct the first systematic study of dataset distillation for SR, evaluating both pixel- and latent-space formulations. We show that a distilled dataset, occupying only 8.88% of the original size, can train SR models that retain nearly the same reconstruction fidelity as those trained on full datasets. Furthermore, we analyze how initialization strategies and distillation objectives affect efficiency, convergence, and visual quality. Our findings highlight the feasibility of SR dataset distillation and establish foundational insights for memory- and compute-efficient generative restoration models.
Authors: Anh-Tien Nguyen, Duy Minh Ho Nguyen, Nghiem Tuong Diep, Trung Quoc Nguyen, Nhat Ho, Jacqueline Michelle Metsch, Miriam Cindy Maurer, Daniel Sonntag, Hanibal Bohnenberger, Anne-Christin Hauschild
Abstract: Whole slide pathology image classification presents challenges due to gigapixel image sizes and limited annotation labels, hindering model generalization. This paper introduces a prompt learning method to adapt large vision-language models for few-shot pathology classification. We first extend the Prov-GigaPath vision foundation model, pre-trained on 1.3 billion pathology image tiles, into a vision-language model by adding adaptors and aligning it with medical text encoders via contrastive learning on 923K image-text pairs. The model is then used to extract visual features and text embeddings from few-shot annotations and fine-tunes with learnable prompt embeddings. Unlike prior methods that combine prompts with frozen features using prefix embeddings or self-attention, we propose multi-granular attention that compares interactions between learnable prompts with individual image patches and groups of them. This approach improves the model's ability to capture both fine-grained details and broader context, enhancing its recognition of complex patterns across sub-regions. To further improve accuracy, we leverage (unbalanced) optimal transport-based visual-text distance to secure model robustness by mitigating perturbations that might occur during the data augmentation process. Empirical experiments on lung, kidney, and breast pathology modalities validate the effectiveness of our approach; thereby, we surpass several of the latest competitors and consistently improve performance across diverse architectures, including CLIP, PLIP, and Prov-GigaPath integrated PLIP.
Authors: Dengdeng Huang, Shikui Tu
Abstract: Deep generative models provide a promising approach to de novo 3D peptide design. Most of them jointly model the distributions of peptide's position, orientation, and conformation, attempting to simultaneously converge to the target pocket. However, in the early stage of docking, optimizing conformation-only modalities such as rotation and torsion can be physically meaningless, as the peptide is initialized far from the protein pocket and no interaction field is present. We define this problem as the multimodal temporal inconsistency problem and claim it is a key factor contributing to low binding affinity in generated peptides. To address this challenge, we propose THFlow, a novel flow matching-based multimodal generative model that explicitly models the temporal hierarchy between peptide position and conformation. It employs a polynomial based conditional flow to accelerate positional convergence early on, and later aligns it with rotation and torsion for coordinated conformation refinement under the emerging interaction field. Additionally, we incorporate interaction-related features, such as polarity, to further enhance the model's understanding of peptide-protein binding. Extensive experiments demonstrate that THFlow outperforms existing methods in generating peptides with superior stability, affinity, and diversity, offering an effective and accurate solution for advancing peptide-based therapeutic development.
Authors: Vivianna Fang He, Sihan Li, Phanish Puranam, Feng Lin
Abstract: Numerous countries globally face shortages of medical experts, deepening inequalities in access to healthcare. Artificial Intelligence (AI)-based diagnostic tools hold considerable promise to tackle this challenge by enabling even novices to deliver expert-level medical services. However, reliance on AI for task completion may hinder the learning required for novices to develop expertise. We thus explore whether AI-based diagnostic tools can be used to enhance not only performance but also learning in the context of lung cancer diagnosis. We examine the distinct effects of AI input during training (i.e., learning how to diagnose) versus in practice (i.e., completing diagnostic tasks) on novice medical professionals' performance. In two field experiments, 576 medical students were randomly assigned across conditions, manipulating the access to AI input during their training, during a test of their diagnostic capabilities, or both. During practice, participants diagnosed potential lung cancer cases using chest CT scans, and their diagnoses were evaluated against the ground truth obtained through histopathological examinations. Study 1 (N = 336) revealed that AI input in training alone improved human diagnostic accuracy by 3.2 percentage points over the control, while AI input during practice alone increased human accuracy by 7.9 percentage points. Combined deployment in both training and practice yielded an improvement of 13.7 percentage points--significantly exceeding either approach alone. Study 2 (N = 240) showed that AI input in practice alone improved accuracy in subsequent practice, unaided by AI, by 9.9 percentage points over the control. Even minimally informative AI input in training improved diagnostic accuracy by 5.3 percentage points over the control. These results reveal AI's dual role: As a tool, it could rapidly improve novices' performance.
Authors: Andrea Bertazzi, Tim Johnston, Gareth O. Roberts, Alain Durmus
Abstract: This paper aims to provide differential privacy (DP) guarantees for Markov chain Monte Carlo (MCMC) algorithms. In a first part, we establish DP guarantees on samples output by MCMC algorithms as well as Monte Carlo estimators associated with these methods under assumptions on the convergence properties of the underlying Markov chain. In particular, our results highlight the critical condition of ensuring the target distribution is differentially private itself. In a second part, we specialise our analysis to the unadjusted Langevin algorithm and stochastic gradient Langevin dynamics and establish guarantees on their (R\'enyi) DP. To this end, we develop a novel methodology based on Girsanov's theorem combined with a perturbation trick to obtain bounds for an unbounded domain and in a non-convex setting. We establish: (i) uniform in $n$ privacy guarantees when the state of the chain after $n$ iterations is released, (ii) bounds on the privacy of the entire chain trajectory. These findings provide concrete guidelines for privacy-preserving MCMC.
Authors: Jingru Jia, Zehua Yuan, Junhao Pan, Paul E. McNamara, Deming Chen
Abstract: Strategic decision-making involves interactive reasoning where agents adapt their choices in response to others, yet existing evaluations of large language models (LLMs) often emphasize Nash Equilibrium (NE) approximation, overlooking the mechanisms driving their strategic choices. To bridge this gap, we introduce an evaluation framework grounded in behavioral game theory, disentangling reasoning capability from contextual effects. Testing 22 state-of-the-art LLMs, we find that GPT-o3-mini, GPT-o1, and DeepSeek-R1 dominate most games yet also demonstrate that the model scale alone does not determine performance. In terms of prompting enhancement, Chain-of-Thought (CoT) prompting is not universally effective, as it increases strategic reasoning only for models at certain levels while providing limited gains elsewhere. Additionally, we investigate the impact of encoded demographic features on the models, observing that certain assignments impact the decision-making pattern. For instance, GPT-4o shows stronger strategic reasoning with female traits than males, while Gemma assigns higher reasoning levels to heterosexual identities compared to other sexual orientations, indicating inherent biases. These findings underscore the need for ethical standards and contextual alignment to balance improved reasoning with fairness.
Authors: Tianyi Zeng, Tianyi Wang, Zimo Zeng, Feiyang Zhang, Jiseop Byeon, Yujin Wang, Yajie Zou, Yangyang Wang, Junfeng Jiao, Christian Claudel, Xinbo Chen
Abstract: Accurate state estimation is fundamental to intelligent vehicles. Wheel load, one of the most important chassis states, serves as an essential input for advanced driver assistance systems (ADAS) and exerts a direct influence on vehicle stability and safety. However, wheel load estimation remains challenging due to the complexity of chassis modeling and the susceptibility of nonlinear systems to noise. To address these issues, this paper first introduces a refined suspension linkage-level modeling approach that constructs a nonlinear instantaneous dynamic model by explicitly considering the complex geometric structure of the suspension. Building upon this, we propose a damper characteristics-based Bayesian physics-informed neural network (Damper-B-PINN) framework to estimate dynamic wheel load, which leverages the suspension dynamics as physical guidance of PINN while employing Bayesian inference to mitigate the effects of system noise and uncertainty. Moreover, a damper-characteristic physics conditioning (DPC) module is designed for embedding physical prior. The proposed Damper-B-PINN is evaluated using both high-fidelity simulation datasets generated by CarSim software and real-world datasets collected from a Formula Student race car. Experimental results demonstrate that our Damper-B-PINN consistently outperforms existing methods across various test conditions, particularly extreme ones. These findings highlight the potential of the proposed Damper-B-PINN framework to enhance the accuracy and robustness of dynamic wheel load estimation, thereby improving the reliability and safety of ADAS applications.
Authors: Anja Sheppard, Katherine A. Skinner
Abstract: In this work, we propose the use of Ground Penetrating Radar (GPR) for rover localization on Mars. Precise pose estimation is an important task for mobile robots exploring planetary surfaces, as they operate in GPS-denied environments. Although visual odometry provides accurate localization, it is computationally expensive and can fail in dim or high-contrast lighting. Wheel encoders can also provide odometry estimation, but are prone to slipping on the sandy terrain encountered on Mars. Although traditionally a scientific surveying sensor, GPR has been used on Earth for terrain classification and localization through subsurface feature matching. The Perseverance rover and the upcoming ExoMars rover have GPR sensors already equipped to aid in the search of water and mineral resources. We propose to leverage GPR to aid in Mars rover localization. Specifically, we develop a novel GPR-based deep learning model that predicts 1D relative pose translation. We fuse our GPR pose prediction method with inertial and wheel encoder data in a filtering framework to output rover localization. We perform experiments in a Mars analog environment and demonstrate that our GPR-based displacement predictions both outperform wheel encoders and improve multi-modal filtering estimates in high-slip environments. Lastly, we present the first dataset aimed at GPR-based localization in Mars analog environments, which will be made publicly available at https://umfieldrobotics.github.io/marslgpr.
Authors: Yuhao Huang, Ao Chang, Haoran Dou, Xing Tao, Xinrui Zhou, Yan Cao, Ruobing Huang, Alejandro F Frangi, Lingyun Bao, Xin Yang, Dong Ni
Abstract: Accurate segmentation of nodules in both 2D breast ultrasound (BUS) and 3D automated breast ultrasound (ABUS) is crucial for clinical diagnosis and treatment planning. Therefore, developing an automated system for nodule segmentation can enhance user independence and expedite clinical analysis. Unlike fully-supervised learning, weakly-supervised segmentation (WSS) can streamline the laborious and intricate annotation process. However, current WSS methods face challenges in achieving precise nodule segmentation, as many of them depend on inaccurate activation maps or inefficient pseudo-mask generation algorithms. In this study, we introduce a novel multi-agent reinforcement learning-based WSS framework called Flip Learning, which relies solely on 2D/3D boxes for accurate segmentation. Specifically, multiple agents are employed to erase the target from the box to facilitate classification tag flipping, with the erased region serving as the predicted segmentation mask. The key contributions of this research are as follows: (1) Adoption of a superpixel/supervoxel-based approach to encode the standardized environment, capturing boundary priors and expediting the learning process. (2) Introduction of three meticulously designed rewards, comprising a classification score reward and two intensity distribution rewards, to steer the agents' erasing process precisely, thereby avoiding both under- and over-segmentation. (3) Implementation of a progressive curriculum learning strategy to enable agents to interact with the environment in a progressively challenging manner, thereby enhancing learning efficiency. Extensively validated on the large in-house BUS and ABUS datasets, our Flip Learning method outperforms state-of-the-art WSS methods and foundation models, and achieves comparable performance as fully-supervised learning algorithms.
Authors: Leon Keller, Daniel Tanneberg, Jan Peters
Abstract: Imitation learning is a popular method for teaching robots new behaviors. However, most existing methods focus on teaching short, isolated skills rather than long, multi-step tasks. To bridge this gap, imitation learning algorithms must not only learn individual skills but also an abstract understanding of how to sequence these skills to perform extended tasks effectively. This paper addresses this challenge by proposing a neuro-symbolic imitation learning framework. Using task demonstrations, the system first learns a symbolic representation that abstracts the low-level state-action space. The learned representation decomposes a task into easier subtasks and allows the system to leverage symbolic planning to generate abstract plans. Subsequently, the system utilizes this task decomposition to learn a set of neural skills capable of refining abstract plans into actionable robot commands. Experimental results in three simulated robotic environments demonstrate that, compared to baselines, our neuro-symbolic approach increases data efficiency, improves generalization capabilities, and facilitates interpretability.
Authors: Kai Yan, Yufei Xu, Zhengyin Du, Xuesong Yao, Zheyu Wang, Xiaowen Guo, Jiecao Chen
Abstract: The rapid escalation from elementary school-level to frontier problems of the difficulty for LLM benchmarks in recent years have weaved a miracle for researchers that we are only inches away from surpassing human intelligence. However, is the LLMs' remarkable reasoning ability indeed comes from true intelligence by human standards, or are they simply reciting solutions witnessed during training at an Internet level? To study this problem, we propose RoR-Bench, a novel, multi-modal benchmark for detecting LLM's recitation behavior when asked simple reasoning problems but with conditions subtly shifted, and conduct empirical analysis on our benchmark. Surprisingly, we found existing cutting-edge LLMs unanimously exhibits extremely severe recitation behavior; by changing one phrase in the condition, top models such as OpenAI-o1 and DeepSeek-R1 can suffer 60 percent performance loss on elementary school-level arithmetic and reasoning problems. Such findings are a wake-up call to the LLM community that compels us to re-evaluate the true intelligence level of cutting-edge LLMs.
Authors: Khai Le-Duc, Tuyen Tran, Bach Phan Tat, Nguyen Kim Hai Bui, Quan Dang, Hung-Phong Tran, Thanh-Thuy Nguyen, Ly Nguyen, Tuan-Minh Phan, Thi Thu Phuong Tran, Chris Ngo, Nguyen X. Khanh, Thanh Nguyen-Tang
Abstract: Multilingual speech translation (ST) and machine translation (MT) in the medical domain enhances patient care by enabling efficient communication across language barriers, alleviating specialized workforce shortages, and facilitating improved diagnosis and treatment, particularly during pandemics. In this work, we present the first systematic study on medical ST, to our best knowledge, by releasing MultiMed-ST, a large-scale ST dataset for the medical domain, spanning all translation directions in five languages: Vietnamese, English, German, French, and Simplified/Traditional Chinese, together with the models. With 290,000 samples, this is the largest medical MT dataset and the largest many-to-many multilingual ST among all domains. Secondly, we present the most comprehensive ST analysis in the field's history, to our best knowledge, including: empirical baselines, bilingual-multilingual comparative study, end-to-end vs. cascaded comparative study, task-specific vs. multi-task sequence-to-sequence comparative study, code-switch analysis, and quantitative-qualitative error analysis. All code, data, and models are available online: https://github.com/leduckhai/MultiMed-ST
Authors: Ji Ma
Abstract: Large language models (LLMs) increasingly serve as human-like decision-making agents in social science and applied settings. These LLM-agents are typically assigned human-like characters and placed in real-life contexts. However, how these characters and contexts shape an LLM's behavior remains underexplored. This study proposes and tests methods for probing, quantifying, and modifying an LLM's internal representations in a Dictator Game -- a classic behavioral experiment on fairness and prosocial behavior. We extract "vectors of variable variations" (e.g., "male" to "female") from the LLM's internal state. Manipulating these vectors during the model's inference can substantially alter how those variables relate to the model's decision-making. This approach offers a principled way to study and regulate how social concepts can be encoded and engineered within transformer-based models, with implications for alignment, debiasing, and designing AI agents for social simulations in both academic and commercial applications, strengthening sociological theory and measurement.
Authors: Ole-Christian Galbo Engstr{\o}m, Michela Albano-Gaglio, Erik Schou Dreier, Yamine Bouzembrak, Maria Font-i-Furnols, Puneet Mishra, Kim Steenstrup Pedersen
Abstract: Current approaches to chemical map generation from hyperspectral images are based on models such as partial least squares (PLS) regression, generating pixel-wise predictions that do not consider spatial context and suffer from a high degree of noise. This study proposes an end-to-end deep learning approach using a modified version of U-Net and a custom loss function to directly obtain chemical maps from hyperspectral images, skipping all intermediate steps required for traditional pixel-wise analysis. This study compares the U-Net with the traditional PLS regression on a real dataset of pork belly samples with associated mean fat reference values. The U-Net obtains a test set root mean squared error that is 7% lower than that of PLS regression on the task of mean fat prediction. At the same time, U-Net generates fine detail chemical maps where 99.91% of the variance is spatially correlated. Conversely, only 2.37% of the variance in the PLS-generated chemical maps is spatially correlated, indicating that each pixel-wise prediction is largely independent of neighboring pixels. Additionally, while the PLS-generated chemical maps contain predictions far beyond the physically possible range of 0-100%, U-Net learns to stay inside this range. Thus, the findings of this study indicate that U-Net is superior to PLS for chemical map generation.
Authors: Junwei Liao, Muning Wen, Jun Wang, Weinan Zhang
Abstract: LLM-based Multi-Agent Systems have demonstrated remarkable capabilities in addressing complex, agentic tasks, from generating high-quality presentation slides to even conducting sophisticated scientific research. Meanwhile, RL has been widely recognized for its effectiveness in enhancing agent intelligence, but limited research has investigated the fine-tuning of LaMAS using foundational RL techniques. Moreover, the direct application of MARL methods to LaMAS introduces significant challenges, stemming from the unique characteristics and mechanisms inherent to LaMAS. To address these challenges, this article presents a comprehensive study of LLM-based MARL and proposes a novel paradigm termed Multi-Agent Reinforcement Fine-Tuning (MARFT). We introduce a brand-new MG called Flex-MG, which aligns with the LaMAS optimization in real-world applications and a universal algorithmic framework tailored specifically for LaMAS, outlining the conceptual foundations, key distinctions, and practical implementation strategies. We review the evolution from RL to RFT, setting the stage for a parallel analysis in the multi-agent domain. In the context of LaMAS, we elucidate critical differences between MARL and MARFT. These differences motivate a transition toward a LaMAS-oriented formulation of RFT. Central to this work is a robust and scalable MARFT framework. We detail the core algorithm and provide a complete, open-source implementation to facilitate adoption and further research. The latter sections of the paper explore real-world application perspectives and opening challenges in MARFT. By bridging theoretical underpinnings with practical methodologies, this work serves as a roadmap for researchers seeking to advance MARFT toward resilient and adaptive solutions in agentic systems. Our implementation of the proposed framework is publicly available at: https://github.com/jwliao-ai/MARFT.
Authors: Emily Minus, R. Yates Coley, Susan M. Shortreed, Brian D. Williamson
Abstract: Objective: Area under the receiving operator characteristic curve (AUC) is commonly reported alongside prediction models for binary outcomes. Recent articles have raised concerns that AUC might be a misleading measure of prediction performance in the rare event setting. This setting is common since many events of clinical importance are rare. We aimed to determine whether the bias and variance of AUC are driven by the number of events or the event rate. We also investigated the behavior of other commonly used measures of prediction performance, including positive predictive value, accuracy, sensitivity, and specificity. Study Design and Setting: We conducted a simulation study to determine when or whether AUC is unstable in the rare event setting by varying the size of datasets used to train and evaluate prediction models. This plasmode simulation study was based on data from the Mental Health Research Network; the data contained 149 predictors and the outcome of interest, suicide attempt, which had event rate 0.92\% in the original dataset. Results: Our results indicate that poor AUC behavior -- as measured by empirical bias, variability of cross-validated AUC estimates, and empirical coverage of confidence intervals -- is driven by the number of events in a rare-event setting, not event rate. Performance of sensitivity is driven by the number of events, while that of specificity is driven by the number of non-events. Other measures, including positive predictive value and accuracy, depend on the event rate even in large samples. Conclusion: AUC is reliable in the rare event setting provided that the total number of events is moderately large; in our simulations, we observed near zero bias with 1000 events.
Authors: Kevin Lane, Morteza Karimzadeh
Abstract: Foundation models have garnered increasing attention for representation learning in remote sensing. Many such foundation models adopt approaches that have demonstrated success in computer vision with minimal domain-specific modification. However, the development and application of foundation models in this field are still burgeoning, as there are a variety of competing approaches for how to most effectively leverage remotely sensed data. This paper examines these approaches, along with their roots in the computer vision field. This is done to characterize potential advantages and pitfalls, while outlining future directions to further improve remote sensing-specific foundation models. We discuss the quality of the learned representations and methods to alleviate the need for massive compute resources. We first examine single-sensor remote foundation models to introduce concepts and provide context, and then place emphasis on incorporating the multi-sensor aspect of Earth observations into foundation models. In particular, we explore the extent to which existing approaches leverage multiple sensors in training foundation models in relation to multi-modal foundation models. Finally, we identify opportunities for further harnessing the vast amounts of unlabeled, seasonal, and multi-sensor remote sensing observations.
Authors: William Sutcliffe, Marta Calvi, Simone Capelli, Jonas Eschle, Juli\'an Garc\'ia Pardi\~nas, Abhijit Mathad, Azusa Uzuki, Nicola Serra
Abstract: The growing luminosity frontier at the Large Hadron Collider is challenging the reconstruction and analysis of particle collision events. Increased particle multiplicities are straining latency and storage requirements at the data acquisition stage, while new complications are emerging, including higher background levels and more frequent particle vertex misassociations. This in turn necessitates the development of more holistic and scalable reconstruction methods that take advantage of recent advances in machine learning. We propose a novel Heterogeneous Graph Neural Network (HGNN) architecture featuring unique representations for diverse particle collision relationships and integrated graph pruning layers for scalability. Trained with a multi-task paradigm in an environment mimicking the LHCb experiment, this HGNN significantly improves beauty hadron reconstruction performance. Notably, it concurrently performs particle vertex association and graph pruning within a single framework. We quantify reconstruction and pruning performance, demonstrate enhanced inference time scaling with event complexity, and mitigate potential performance loss using a weighted message passing scheme.
Authors: Zihan Zhou, Muhammad Qasim Elahi, Murat Kocaoglu
Abstract: A fundamental challenge in the empirical sciences involves uncovering causal structure through observation and experimentation. Causal discovery entails linking the conditional independence (CI) invariances in observational data to their corresponding graphical constraints via d-separation. In this paper, we consider a general setting where we have access to data from multiple experimental distributions resulting from hard interventions, as well as potentially from an observational distribution. By comparing different interventional distributions, we propose a set of graphical constraints that are fundamentally linked to Pearl's do-calculus within the framework of hard interventions. These graphical constraints associate each graphical structure with a set of interventional distributions that are consistent with the rules of do-calculus. We characterize the interventional equivalence class of causal graphs with latent variables and introduce a graphical representation that can be used to determine whether two causal graphs are interventionally equivalent, i.e., whether they are associated with the same family of hard interventional distributions, where the elements of the family are indistinguishable using the invariances from do-calculus. We also propose a learning algorithm to integrate multiple datasets from hard interventions, introducing new orientation rules. The learning objective is a tuple of augmented graphs which entails a set of causal graphs. We also prove the soundness of the proposed algorithm.
Authors: Qingwen Bu, Yanting Yang, Jisong Cai, Shenyuan Gao, Guanghui Ren, Maoqing Yao, Ping Luo, Hongyang Li
Abstract: A generalist robot should perform effectively across various environments. However, most existing approaches heavily rely on scaling action-annotated data to enhance their capabilities. Consequently, they are often limited to single physical specification and struggle to learn transferable knowledge across different embodiments and environments. To confront these limitations, we propose UniVLA, a new framework for learning cross-embodiment vision-language-action (VLA) policies. Our key innovation is to derive task-centric action representations from videos with a latent action model. This enables us to exploit extensive data across a wide spectrum of embodiments and perspectives. To mitigate the effect of task-irrelevant dynamics, we incorporate language instructions and establish a latent action model within the DINO feature space. Learned from internet-scale videos, the generalist policy can be deployed to various robots through efficient latent action decoding. We obtain state-of-the-art results across multiple manipulation and navigation benchmarks, as well as real-robot deployments. UniVLA achieves superior performance over OpenVLA with less than 1/20 of pretraining compute and 1/10 of downstream data. Continuous performance improvements are observed as heterogeneous data, even including human videos, are incorporated into the training pipeline. The results underscore UniVLA's potential to facilitate scalable and efficient robot policy learning.
Authors: Junhe Zhang, Wanli Ni, Pengwei Wang, Dongyu Wang
Abstract: With the emergence of large model-based agents, widely adopted transformer-based architectures inevitably produce excessively long token embeddings for transmission, which may result in high bandwidth overhead, increased power consumption and latency. In this letter, we propose a task-oriented multimodal token transmission scheme for efficient multimodal information fusion and utilization. To improve the efficiency of token transmission, we design a two-stage training algotithm, including cross-modal alignment and task-oriented fine-tuning, for large model-based token communication. Meanwhile, token compression is performed using a sliding window pooling operation to save communication resources. To balance the trade-off between latency and model performance caused by compression, we formulate a weighted-sum optimization problem over latency and validation loss. We jointly optimizes bandwidth, power allocation, and token length across users by using an alternating optimization method. Simulation results demonstrate that the proposed algorithm outperforms the baseline under different bandwidth and power budgets. Moreover, the two-stage training algorithm achieves higher accuracy across various signal-to-noise ratios than the method without cross-modal alignment.
Authors: Julia Wunderle, Anton Ehrmanntraut, Jan Pfister, Fotis Jannidis, Andreas Hotho
Abstract: Encoders remain essential for efficient German NLP and NLU scenarios despite the rise of decoder-only LLMs. This work studies two routes to high-quality German encoders under identical data and training constraints: 1) training from scratch and 2) converting decoders via LLM2Vec. We introduce two resources: ModernGBERT (134M, 1B), fully transparent German encoders in the ModernBERT style, and LL\"aMmleinVec (120M, 1B, 7B), decoder-to-encoder conversions trained with masked next-token prediction, both undergoing a context extension to 8.192 tokens. Across SuperGLEBer, ModernGBERT 1B sets a new state of the art (avg 0.808), surpassing GBERT Large (+4%) and the seven-times larger converted 7B model (0.787). On German MTEB after supervised fine-tuning, ModernGBERT 1B (0.551) approaches the converted 7B model (0.557). We release all models, checkpoints, datasets, and full training records, and introduce an encoder-adapted QA-NIAH evaluation. All in all, our results provide actionable guidance: when parameter efficiency and latency matter, from-scratch encoders dominate. When a pre-trained decoder exists and compute is a limited, conversion offers an effective alternative. ModernGBERT and LL\"aMmleinVec, including all code, data and intermediary checkpoints are published under a research-only RAIL license.
Authors: Charles Hong, Sahil Bhatia, Alvin Cheung, Yakun Sophia Shao
Abstract: Hardware accelerators, especially those designed for tensor processing, have become ubiquitous in today's computing landscape. However, even with significant efforts in building compilers, programming these tensor accelerators remains challenging, leaving much of their potential underutilized. Recently, large language models (LLMs), trained on large amounts of code, have shown significant promise in code generation and optimization tasks, but generating low-resource languages, such as specialized tensor accelerator code still poses a significant challenge. We tackle this challenge with Autocomp, an approach that empowers accelerator programmers to leverage domain knowledge and hardware feedback to optimize code via an automated LLM-driven search. We accomplish this by: 1) formulating each optimization pass as a structured two-phase prompt, divided into planning and code generation phases, 2) inserting domain knowledge during planning via a concise and adaptable optimization menu, and 3) integrating correctness and performance metrics from hardware as feedback at each search iteration. Across three distinct hardware platforms, we demonstrate that Autocomp-optimized code runs 5.6x faster than the vendor-provided library (Gemmini), outperforms expert-level hand-tuned code by 1.9x (AWS Trainium), and achieves 3.8x higher performance than a machine learning-based cost model for GPUs (NVIDIA L40S). Additionally, we demonstrate that optimization schedules generated from Autocomp can be reused across similar tensor operations, improving speedups by up to 24% under a fixed sample budget.
Authors: Jamie Hayes, Ilia Shumailov, Christopher A. Choquette-Choo, Matthew Jagielski, George Kaissis, Milad Nasr, Sahra Ghalebikesabi, Meenatchi Sundaram Mutu Selva Annamalai, Niloofar Mireshghallah, Igor Shilov, Matthieu Meeus, Yves-Alexandre de Montjoye, Katherine Lee, Franziska Boenisch, Adam Dziedzic, A. Feder Cooper
Abstract: State-of-the-art membership inference attacks (MIAs) typically require training many reference models, making it difficult to scale these attacks to large pre-trained language models (LLMs). As a result, prior research has either relied on weaker attacks that avoid training references (e.g., fine-tuning attacks), or on stronger attacks applied to small models and datasets. However, weaker attacks have been shown to be brittle and insights from strong attacks in simplified settings do not translate to today's LLMs. These challenges prompt an important question: are the limitations observed in prior work due to attack design choices, or are MIAs fundamentally ineffective on LLMs? We address this question by scaling LiRA--one of the strongest MIAs--to GPT-2 architectures ranging from 10M to 1B parameters, training references on over 20B tokens from the C4 dataset. Our results advance the understanding of MIAs on LLMs in four key ways. While (1) strong MIAs can succeed on pre-trained LLMs, (2) their effectiveness, remains limited (e.g., AUC<0.7) in practical settings. (3) Even when strong MIAs achieve better-than-random AUC, aggregate metrics can conceal substantial per-sample MIA decision instability: due to training randomness, many decisions are so unstable that they are statistically indistinguishable from a coin flip. Finally, (4) the relationship between MIA success and related LLM privacy metrics is not as straightforward as prior work has suggested.
Authors: Daniel Barzilai, Ohad Shamir
Abstract: The widespread use of generative models has created a feedback loop, in which each generation of models is trained on data partially produced by its predecessors. This process has raised concerns about model collapse: A critical degradation in performance caused by repeated training on synthetic data. However, different analyses in the literature have reached different conclusions as to the severity of model collapse. As such, it remains unclear how concerning this phenomenon is, and under which assumptions it can be avoided. To address this, we theoretically study model collapse for maximum likelihood estimation (MLE), in a natural setting where synthetic data is gradually added to the original data set. Under standard assumptions (similar to those long used for proving asymptotic consistency and normality of MLE), we establish non-asymptotic bounds showing that collapse can be avoided even as the fraction of real data vanishes. On the other hand, we prove that some assumptions (beyond MLE consistency) are indeed necessary: Without them, model collapse can occur arbitrarily quickly, even when the original data is still present in the training set. To the best of our knowledge, these are the first rigorous examples of iterative generative modeling with accumulating data that rapidly leads to model collapse.
Authors: Gleb Mezentsev, Ivan Oseledets
Abstract: A recent study showed that large language models (LLMs) can reconstruct surprisingly long texts - up to thousands of tokens - via autoregressive generation from just one trained input embedding. In this work, we explore whether autoregressive decoding is essential for such reconstruction. We show that frozen LLMs can generate hundreds of accurate tokens in just one token-parallel forward pass, when provided with only two learned embeddings. This reveals a surprising and underexplored multi-token generation capability of autoregressive LLMs. We examine these embeddings and characterize the information they encode. We also empirically show that, although these representations are not unique for a given text, they form connected and local regions in embedding space - suggesting the potential to train a practical encoder. The existence of such representations hints that multi-token generation may be natively accessible in off-the-shelf LLMs via a learned input encoder, eliminating heavy retraining and helping to overcome the fundamental bottleneck of autoregressive decoding while reusing already-trained models.
Authors: Yida Xue, Zhen Bi, Jinnan Yang, Jungang Lou, Kehai Chen, Min Zhang, Huajun Chen, Ningyu Zhang
Abstract: Recent advances in Multimodal Large Language Models (MLLMs) have significantly enhanced their capabilities; however, their spatial perception abilities remain a notable limitation. To address this challenge, multimodal data synthesis offers a promising solution. Yet, ensuring that synthesized data adhere to spatial common sense is a non-trivial task. Our approach addresses this critical gap by providing a systematic framework for generating spatially coherent data. In this work, we introduce SKG2DATA, a novel multimodal synthesis approach guided by spatial knowledge graphs, grounded in the concept of knowledge-to-data generation. SKG2DATA employs an automated pipeline for constructing Spatial Knowledge Graph (SKG) that effectively captures human-like spatial cognition, including directional and distance relationships. These structured representations then serve as precise guidance for our integrated synthesis pipeline, where a diffusion model generates spatially-consistent images while a MLLM produces corresponding textual descriptions. The automated construction of SKG enables scalable generation of diverse yet realistic spatial configurations, overcoming the limitations of manual data collection and annotation. Extensive experiments demonstrate that data synthesized from diverse types of spatial knowledge, including direction and distance, enhance the spatial perception and reasoning abilities of MLLMs markedly, albeit with a slight cost to their general capabilities. We hope that the idea of knowledge-based data synthesis can advance the development of spatial intelligence. Code is available at https://github.com/zjunlp/Knowledge2Data.
Authors: Pedram Ghamisi, Weikang Yu, Xiaokang Zhang, Aldino Rizaldy, Jian Wang, Chufeng Zhou, Richard Gloaguen, Gustau Camps-Valls
Abstract: Foundation Models (FMs) are large-scale, pre-trained artificial intelligence (AI) systems that have revolutionized natural language processing and computer vision, and are now advancing geospatial analysis and Earth Observation (EO). They promise improved generalization across tasks, scalability, and efficient adaptation with minimal labeled data. However, despite the rapid proliferation of geospatial FMs, their real-world utility and alignment with global sustainability goals remain underexplored. We introduce SustainFM, a comprehensive benchmarking framework grounded in the 17 Sustainable Development Goals with extremely diverse tasks ranging from asset wealth prediction to environmental hazard detection. This study provides a rigorous, interdisciplinary assessment of geospatial FMs and offers critical insights into their role in attaining sustainability goals. Our findings show: (1) While not universally superior, FMs often outperform traditional approaches across diverse tasks and datasets. (2) Evaluating FMs should go beyond accuracy to include transferability, generalization, and energy efficiency as key criteria for their responsible use. (3) FMs enable scalable, SDG-grounded solutions, offering broad utility for tackling complex sustainability challenges. Critically, we advocate for a paradigm shift from model-centric development to impact-driven deployment, and emphasize metrics such as energy efficiency, robustness to domain shifts, and ethical considerations.
Authors: Hyojin Bahng, Caroline Chan, Fredo Durand, Phillip Isola
Abstract: Measuring alignment between language and vision is a fundamental challenge, especially as multimodal data becomes increasingly detailed and complex. Existing methods often rely on collecting human or AI preferences, which can be costly and time-intensive. We propose an alternative approach that leverages cycle consistency as a supervisory signal. Given an image and generated text, we map the text back to image space using a text-to-image model and compute the similarity between the original image and its reconstruction. Analogously, for text-to-image generation, we measure the textual similarity between an input caption and its reconstruction through the cycle. We use the cycle consistency score to rank candidates and construct a preference dataset of 866K comparison pairs. The reward model trained on our dataset, CycleReward, outperforms state-of-the-art alignment metrics on detailed captioning, with superior inference-time scalability when used as a verifier for Best-of-N sampling, while maintaining speed and differentiability. Furthermore, performing DPO and Diffusion DPO using our dataset enhances performance across a wide range of vision-language tasks and text-to-image generation. Our dataset, model, and code are publicly released at https://cyclereward.github.io.
Authors: Jingfeng Wu, Pierre Marion, Peter Bartlett
Abstract: We study gradient descent (GD) with a constant stepsize for $\ell_2$-regularized logistic regression with linearly separable data. Classical theory suggests small stepsizes to ensure monotonic reduction of the optimization objective, achieving exponential convergence in $\widetilde{\mathcal{O}}(\kappa)$ steps with $\kappa$ being the condition number. Surprisingly, we show that this can be accelerated to $\widetilde{\mathcal{O}}(\sqrt{\kappa})$ by simply using a large stepsize -- for which the objective evolves nonmonotonically. The acceleration brought by large stepsizes extends to minimizing the population risk for separable distributions, improving on the best-known upper bounds on the number of steps to reach a near-optimum. Finally, we characterize the largest stepsize for the local convergence of GD, which also determines the global convergence in special scenarios. Our results extend the analysis of Wu et al. (2024) from convex settings with minimizers at infinity to strongly convex cases with finite minimizers.
Authors: Long Tan Le, Senura Hansaja Wanasekara, Zerun Niu, Nguyen H. Tran, Phuong Vo, Walid Saad, Dusit Niyato, Zhu Han, Choong Seon Hong, H. Vincent Poor
Abstract: Semantic communication (SemCom) has emerged as a promising paradigm for 6G wireless systems by transmitting task-relevant information rather than raw bits, yet existing approaches remain vulnerable to dual sources of uncertainty: semantic misinterpretation arising from imperfect feature extraction and transmission-level perturbations from channel noise. Current deep learning based SemCom systems typically employ domain-specific architectures that lack robustness guarantees and fail to generalize across diverse noise conditions, adversarial attacks, and out-of-distribution data. In this paper, a novel and generalized semantic communication framework called WaSeCom is proposed to systematically address uncertainty and enhance robustness. In particular, Wasserstein distributionally robust optimization is employed to provide resilience against semantic misinterpretation and channel perturbations. A rigorous theoretical analysis is performed to establish the robust generalization guarantees of the proposed framework. Experimental results on image and text transmission demonstrate that WaSeCom achieves improved robustness under noise and adversarial perturbations. These results highlight its effectiveness in preserving semantic fidelity across varying wireless conditions.
Authors: Zhengyang Li, Sawyer Campos, Nana Wang
Abstract: This paper introduces LLM-MARL, a unified framework that incorporates large language models (LLMs) into multi-agent reinforcement learning (MARL) to enhance coordination, communication, and generalization in simulated game environments. The framework features three modular components of Coordinator, Communicator, and Memory, which dynamically generate subgoals, facilitate symbolic inter-agent messaging, and support episodic recall. Training combines PPO with a language-conditioned loss and LLM query gating. LLM-MARL is evaluated in Google Research Football, MAgent Battle, and StarCraft II. Results show consistent improvements over MAPPO and QMIX in win rate, coordination score, and zero-shot generalization. Ablation studies demonstrate that subgoal generation and language-based messaging each contribute significantly to performance gains. Qualitative analysis reveals emergent behaviors such as role specialization and communication-driven tactics. By bridging language modeling and policy learning, this work contributes to the design of intelligent, cooperative agents in interactive simulations. It offers a path forward for leveraging LLMs in multi-agent systems used for training, games, and human-AI collaboration.
Authors: Tianyu Qi, Lei Xue, Yufeng Zhan, Xiaobo Ma
Abstract: The growing use of large pre-trained models in edge computing has made model inference on mobile clients both feasible and popular. Yet these devices remain vulnerable to adversarial attacks, threatening model robustness and security. Federated adversarial training (FAT) offers a promising solution by enhancing robustness while preserving client privacy. However, FAT often yields a generalized global model that struggles with heterogeneous client data, leading to limited personalization and significant communication overhead. In this paper, we propose \textit{Lorica}, a personalized synergistic adversarial training framework that delivers customized defense models through a two-phase process. In Phase 1, \textit{Lorica} applies LoRA-FA for local adversarial fine-tuning, enabling personalized robustness while reducing communication by uploading only LoRA-FA parameters. In Phase 2, a forward-gating selection strategy improves benign accuracy, further refining the personalized model. This yields tailored defense models that effectively balance robustness and accuracy. Extensive experiments on benchmark datasets demonstrate that \textit{Lorica} can achieve up to 68$\times$ improvements in communication efficiency compared to state-of-the-art algorithms, while achieving up to 29.9\% and 52.2\% enhancements in adversarial robustness and benign accuracy, respectively.
Authors: Alan Chen, Jack Merullo, Alessandro Stolfo, Ellie Pavlick
Abstract: In this work, we demonstrate that affine mappings between residual streams of language models is a cheap way to effectively transfer represented features between models. We apply this technique to transfer the weights of Sparse Autoencoders (SAEs) between models of different sizes to compare their representations. We find that small and large models learn similar representation spaces, which motivates training expensive components like SAEs on a smaller model and transferring to a larger model at a FLOPs savings. In particular, using a small-to-large transferred SAE as initialization can lead to 50% cheaper training runs when training SAEs on larger models. Next, we show that transferred probes and steering vectors can effectively recover ground truth performance. Finally, we dive deeper into feature-level transferability, finding that semantic and structural features transfer noticeably differently while specific classes of functional features have their roles faithfully mapped. Overall, our findings illustrate similarities and differences in the linear representation spaces of small and large models and demonstrate a method for improving the training efficiency of SAEs.
Authors: Iustin Sirbu, Robert-Adrian Popovici, Cornelia Caragea, Stefan Trausan-Matu, Traian Rebedea
Abstract: We introduce MultiMatch, a novel semi-supervised learning (SSL) algorithm combining the paradigms of co-training and consistency regularization with pseudo-labeling. At its core, MultiMatch features a pseudo-label weighting module designed for selecting and filtering pseudo-labels based on head agreement and model confidence, and weighting them according to the perceived classification difficulty. This novel module enhances and unifies three existing techniques -- heads agreement from Multihead Co-training, self-adaptive thresholds from FreeMatch, and Average Pseudo-Margins from MarginMatch -- resulting in a holistic approach that improves robustness and performance in SSL settings. Experimental results on benchmark datasets highlight the superior performance of MultiMatch, i.e., MultiMatch achieves state-of-the-art results on 8 out of 10 setups from 5 natural language processing datasets and ranks first according to the Friedman test among 21 methods. Furthermore, MultiMatch demonstrates exceptional robustness in highly imbalanced settings, outperforming the second-best approach by 3.26%, a critical advantage for real-world text classification tasks. Our code is available on GitHub.
Authors: Jiayi Sheng, Luna Lyu, Jikai Jin, Tony Xia, Alex Gu, James Zou, Pan Lu
Abstract: Inequality proving, crucial across diverse scientific and mathematical fields, tests advanced reasoning skills such as discovering tight bounds and strategic theorem application. This makes it a distinct, demanding frontier for large language models (LLMs), offering insights beyond general mathematical problem-solving. Progress in this area is hampered by existing datasets that are often scarce, synthetic, or rigidly formal. We address this by proposing an informal yet verifiable task formulation, recasting inequality proving into two automatically checkable subtasks: bound estimation and relation prediction. Building on this, we release IneqMath, an expert-curated dataset of Olympiad-level inequalities, including a test set and training corpus enriched with step-wise solutions and theorem annotations. We also develop a novel LLM-as-judge evaluation framework, combining a final-answer judge with four step-wise judges designed to detect common reasoning flaws. A systematic evaluation of 29 leading LLMs on IneqMath reveals a surprising reality: even top models like o1 achieve less than 10% overall accuracy under step-wise scrutiny; this is a drop of up to 65.5% from their accuracy considering only final answer equivalence. This discrepancy exposes fragile deductive chains and a critical gap for current LLMs between merely finding an answer and constructing a rigorous proof. Scaling model size and increasing test-time computation yield limited gains in overall proof correctness. Instead, our findings highlight promising research directions such as theorem-guided reasoning and self-refinement. Code and data are available at https://ineqmath.github.io/.
Authors: Donghoon Ahn, Jiwon Kang, Sanghyun Lee, Minjae Kim, Jaewon Min, Wooseok Jang, Sangwu Lee, Sayak Paul, Susung Hong, Seungryong Kim
Abstract: Recent guidance methods in diffusion models steer reverse sampling by perturbing the model to construct an implicit weak model and guide generation away from it. Among these approaches, attention perturbation has demonstrated strong empirical performance in unconditional scenarios where classifier-free guidance is not applicable. However, existing attention perturbation methods lack principled approaches for determining where perturbations should be applied, particularly in Diffusion Transformer (DiT) architectures where quality-relevant computations are distributed across layers. In this paper, we investigate the granularity of attention perturbations, ranging from the layer level down to individual attention heads, and discover that specific heads govern distinct visual concepts such as structure, style, and texture quality. Building on this insight, we propose "HeadHunter", a systematic framework for iteratively selecting attention heads that align with user-centric objectives, enabling fine-grained control over generation quality and visual attributes. In addition, we introduce SoftPAG, which linearly interpolates each selected head's attention map toward an identity matrix, providing a continuous knob to tune perturbation strength and suppress artifacts. Our approach not only mitigates the oversmoothing issues of existing layer-level perturbation but also enables targeted manipulation of specific visual styles through compositional head selection. We validate our method on modern large-scale DiT-based text-to-image models including Stable Diffusion 3 and FLUX.1, demonstrating superior performance in both general quality enhancement and style-specific guidance. Our work provides the first head-level analysis of attention perturbation in diffusion models, uncovering interpretable specialization within attention layers and enabling practical design of effective perturbation strategies.
Authors: Nima Hadidi, Jason Chan, Ebrahim Feghhi, Jonathan C. Kao
Abstract: Surface electromyography (sEMG) at the wrists could enable natural, keyboard-free text entry, yet the state-of-the-art emg2qwerty baseline still misrecognizes $51.8\%$ of characters in the zero-shot setting on unseen users and $7.0\%$ after user-specific fine-tuning. We trace many of these errors to mismatched cross-user signal statistics, fragile reliance on high-order feature dependencies, and the absence of architectural inductive biases aligned with the bilateral nature of typing. To address these issues, we introduce three simple modifications: (i) Rolling Time Normalization, which adaptively aligns input distributions across users; (ii) Aggressive Channel Masking, which encourages reliance on low-order feature combinations more likely to generalize across users; and (iii) a Split-and-Share encoder that processes each hand independently with weight-shared streams to reflect the bilateral symmetry of the neuromuscular system. Combined with a five-fold reduction in spectral resolution ($33\!\rightarrow\!6$ frequency bands), these components yield a compact Split-and-Share model, SplashNet-mini, which uses only $\tfrac14$ the parameters and $0.6\times$ the FLOPs of the baseline while reducing character-error rate (CER) to $36.4\%$ zero-shot and $5.9\%$ after fine-tuning. An upscaled variant, SplashNet ($\tfrac12$ the parameters, $1.15\times$ the FLOPs of the baseline), further lowers error to $35.7\%$ and $5.5\%$, representing relative improvements of $31\%$ and $21\%$ in the zero-shot and fine-tuned settings, respectively. SplashNet therefore establishes a new state of the art without requiring additional data.
Authors: Samuel Howard, Peter Potaptchik, George Deligiannidis
Abstract: Recent advances in flow-based generative modelling have provided scalable methods for computing the Schr\"odinger Bridge (SB) between distributions, a dynamic form of entropy-regularised Optimal Transport (OT) for the quadratic cost. The successful Iterative Markovian Fitting (IMF) procedure solves the SB problem via sequential bridge-matching steps, presenting an elegant and practical approach with many favourable properties over the more traditional Iterative Proportional Fitting (IPF) procedure. Beyond the standard setting, optimal transport can be generalised to the multi-marginal case in which the objective is to minimise a cost defined over several marginal distributions. Of particular importance are costs defined over a tree structure, from which Wasserstein barycentres can be recovered as a special case. In this work, we extend the IMF procedure to solve for the tree-structured SB problem. Our resulting algorithm inherits the many advantages of IMF over IPF approaches in the tree-based setting. In the case of Wasserstein barycentres, our approach can be viewed as extending the widely used fixed-point approach to use flow-based entropic OT solvers, while requiring only simple bridge-matching steps at each iteration.
Authors: Junqi Jiang, Tom Bewley, Salim I. Amoukou, Francesco Leofante, Antonio Rago, Saumitra Mishra, Francesca Toni
Abstract: Test-time scaling improves large language models' (LLMs) performance by allocating more compute budget during inference. To achieve this, existing methods often require intricate modifications to prompting and sampling strategies. In this work, we introduce representation consistency (RC), a test-time scaling method for aggregating answers drawn from multiple candidate responses of an LLM regardless of how they were generated, including variations in prompt phrasing and sampling strategy. RC enhances answer aggregation by not only considering the number of occurrences of each answer in the candidate response set, but also the consistency of the model's internal activations while generating the set of responses leading to each answer. These activations can be either dense (raw model activations) or sparse (encoded via pretrained sparse autoencoders). Our rationale is that if the model's representations of multiple responses converging on the same answer are highly variable, this answer is more likely to be the result of incoherent reasoning and should be down-weighted during aggregation. Importantly, our method only uses cached activations and lightweight similarity computations and requires no additional model queries. Through experiments with four open-source LLMs and four reasoning datasets, we validate the effectiveness of RC for improving task performance during inference, with consistent accuracy improvements (up to 4%) over strong test-time scaling baselines. We also show that consistency in the sparse activation signals aligns well with the common notion of coherent reasoning.
Authors: Chenlang Yi, Zizhan Xiong, Qi Qi, Xiyuan Wei, Girish Bathla, Ching-Long Lin, Bobak Jack Mortazavi, Tianbao Yang
Abstract: Contrastive Language-Image Pre-training (CLIP) models have demonstrated superior performance across various visual tasks including medical image classification. However, fairness concerns, including demographic biases, have received limited attention for CLIP models. This oversight leads to critical issues, particularly those related to race and gender, resulting in disparities in diagnostic outcomes and reduced reliability for underrepresented groups. To address these challenges, we introduce AdFair-CLIP, a novel framework employing adversarial feature intervention to suppress sensitive attributes, thereby mitigating spurious correlations and improving prediction fairness. We conduct comprehensive experiments on chest X-ray (CXR) datasets, and show that AdFair-CLIP significantly enhances both fairness and diagnostic accuracy, while maintaining robust generalization in zero-shot and few-shot scenarios. These results establish new benchmarks for fairness-aware learning in CLIP-based medical diagnostic models, particularly for CXR analysis.
Authors: Christian Intern\`o, Robert Geirhos, Markus Olhofer, Sunny Liu, Barbara Hammer, David Klindt
Abstract: The rapid advancement of generative AI enables highly realistic synthetic videos, posing significant challenges for content authentication and raising urgent concerns about misuse. Existing detection methods often struggle with generalization and capturing subtle temporal inconsistencies. We propose ReStraV(Representation Straightening Video), a novel approach to distinguish natural from AI-generated videos. Inspired by the "perceptual straightening" hypothesis -- which suggests real-world video trajectories become more straight in neural representation domain -- we analyze deviations from this expected geometric property. Using a pre-trained self-supervised vision transformer (DINOv2), we quantify the temporal curvature and stepwise distance in the model's representation domain. We aggregate statistics of these measures for each video and train a classifier. Our analysis shows that AI-generated videos exhibit significantly different curvature and distance patterns compared to real videos. A lightweight classifier achieves state-of-the-art detection performance (e.g., 97.17% accuracy and 98.63% AUROC on the VidProM benchmark), substantially outperforming existing image- and video-based methods. ReStraV is computationally efficient, it is offering a low-cost and effective detection solution. This work provides new insights into using neural representation geometry for AI-generated video detection.
Authors: Dheeraj Vattikonda, Santhoshi Ravichandran, Emiliano Penaloza, Hadi Nekoei, Megh Thakkar, Thibault Le Sellier de Chezelles, Nicolas Gontier, Miguel Mu\~noz-M\'armol, Sahar Omidi Shayegan, Stefania Raimondo, Xue Liu, Alexandre Drouin, Laurent Charlin, Alexandre Pich\'e, Alexandre Lacoste, Massimo Caccia
Abstract: LLM-based web agents have recently made significant progress, but much of it has occurred in closed-source systems, widening the gap with open-source alternatives. Progress has been held back by two key challenges: first, a narrow focus on single-step tasks that overlooks the complexity of multi-step web interactions; and second, the high compute costs required to post-train LLM-based web agents. To address this, we present the first statistically grounded study on compute allocation for LLM web-agent post-training. Our approach uses a two-stage pipeline, training a Llama 3.1 8B student to imitate a Llama 3.3 70B teacher via supervised fine-tuning (SFT), followed by on-policy reinforcement learning. We find this process highly sensitive to hyperparameter choices, making exhaustive sweeps impractical. To spare others from expensive trial-and-error, we sample 1,370 configurations and use bootstrapping to estimate effective hyperparameters. Our results show that combining SFT with on-policy RL consistently outperforms either approach alone on both WorkArena and MiniWob++. Further, this strategy requires only 55% of the compute to match the peak performance of pure SFT on MiniWob++, effectively pushing the compute-performance Pareto frontier, and is the only strategy that can close the gap with closed-source models.
Authors: Jack Lu, Ryan Teehan, Zhenbang Yang, Mengye Ren
Abstract: We introduce Context Tuning, a simple and effective method to significantly enhance few-shot adaptation of language models (LLMs) without fine-tuning model parameters. While prompt-based adaptation techniques have demonstrated the effectiveness of lightweight adaptation methods for LLMs, they typically initialize a trainable prompt or prefix with irrelevant tokens for the task at hand. In contrast, Context Tuning initializes the trainable prompt or prefix with task-specific demonstration examples, leveraging the model's inherent In-Context Learning (ICL) ability to extract relevant information for improved few-shot learning performance. Extensive evaluations on benchmarks such as CrossFit, UnifiedQA, MMLU, BIG-Bench Hard, and ARC demonstrate that Context Tuning outperforms traditional prompt-based adaptation methods and achieves competitive accuracy to Test-Time Training with significantly higher training efficiency.
Authors: Riccardo Alberghi, Elizaveta Demyanenko, Luca Biggio, Luca Saglietti
Abstract: Recent advances in natural language processing highlight two key factors for improving reasoning in large language models (LLMs): (i) allocating more test-time compute tends to help on harder problems but often introduces redundancy in the reasoning trace, and (ii) compute is most effective when reasoning is systematic and incremental, forming structured chains of thought (CoTs) akin to human problem-solving. To study these factors in isolation, we introduce a controlled setting based on shortest-path tasks in layered graphs. We train decoder-only transformers on question-trace-answer triples using a custom tokenizer, comparing models trained on optimal bottom-up dynamic programming traces with those trained on longer, valid traces involving backtracking. Surprisingly, with the same training-token budget, models trained on inefficient traces generalize better to unseen graphs. This benefit is not due to length alone-injecting arbitrary redundancy into reasoning traces fails to help and can even hurt performance. Instead, we find that generalization correlates with the model's confidence in next-token prediction, suggesting that long, coherent, and locally incremental traces make the training signal easier to optimize.
Authors: Sangwoo Park, Jinheon Baek, Soyeong Jeong, Sung Ju Hwang
Abstract: Scientific paper retrieval, particularly framed as document-to-document retrieval, aims to identify relevant papers in response to a long-form query paper, rather than a short query string. Previous approaches to this task have focused exclusively on abstracts, embedding them into dense vectors as surrogates for full documents and calculating similarity between them. Yet, abstracts offer only sparse and high-level summaries, and such methods primarily optimize one-to-one similarity, overlooking the dynamic relations that emerge among relevant papers during the retrieval process. To address this, we propose Chain of Retrieval(COR), a novel iterative framework for full-paper retrieval. Specifically, CoR decomposes each query paper into multiple aspect-specific views, matches them against segmented candidate papers, and iteratively expands the search by promoting top-ranked results as new queries, thereby forming a tree-structured retrieval process. The resulting retrieval tree is then aggregated in a post-order manner: descendants are first combined at the query level, then recursively merged with their parent nodes, to capture hierarchical relations across iterations. To validate this, we present SCIFULLBENCH, a large-scale benchmark providing both complete and segmented contexts of full papers for queries and candidates, and results show that CoR significantly outperforms existing retrieval baselines. Our code and dataset is available at https://github.com/psw0021/Chain-of-Retrieval.git.
Authors: Byeonghee Lee, Sangwook Kang, Ju-Hyun Park, Saebom Jeon, Joonsung Kang
Abstract: We propose a doubly robust estimator for the average treatment effect in high dimensional low sample size observational studies, where contamination and model misspecification pose serious inferential challenges. The estimator combines bounded influence estimating equations for outcome modeling with covariate balancing propensity scores for treatment assignment, embedded within a penalized empirical likelihood framework using nonconvex regularization. It satisfies the oracle property by jointly achieving consistency under partial model correct ness, selection consistency, robustness to contamination, and asymptotic normality. For uncertainty quantification, we derive a finite sample confidence interval using cumulant generating functions and influence function corrections, avoiding reliance on asymptotic approximations. Simulation studies and applications to gene expression datasets (Golub and Khan) demonstrate superior performance in bias, error metrics, and interval calibration, highlighting the method robustness and inferential validity in HDLSS regimes. One notable aspect is that even in the absence of contamination, the proposed estimator and its confidence interval remain efficient compared to those of competing models.
Authors: Biyi Fang, Jean Utke, Truong Vo, Diego Klabjan
Abstract: Convolutional Neural Networks (CNNs) have achieved remarkable success across a wide range of machine learning tasks by leveraging hierarchical feature learning through deep architectures. However, the large number of layers and millions of parameters often make CNNs computationally expensive to train, requiring extensive time and manual tuning to discover optimal architectures. In this paper, we introduce a novel framework for boosting CNN performance that integrates dynamic feature selection with the principles of BoostCNN. Our approach incorporates two key strategies: subgrid selection and importance sampling, to guide training toward informative regions of the feature space. We further develop a family of algorithms that embed boosting weights directly into the network training process using a least squares loss formulation. This integration not only alleviates the burden of manual architecture design but also enhances accuracy and efficiency. Experimental results across several fine-grained classification benchmarks demonstrate that our boosted CNN variants consistently outperform conventional CNNs in both predictive performance and training speed.
Authors: Samuel R\"aber, Till Aczel, Andreas Plesner, Roger Wattenhofer
Abstract: Previous work has suggested that preprocessing images through lossy compression can defend against adversarial perturbations, but comprehensive attack evaluations have been lacking. In this paper, we construct strong white-box and adaptive attacks against various compression models and identify a critical challenge for attackers: high realism in reconstructed images significantly increases attack difficulty. Through rigorous evaluation across multiple attack scenarios, we demonstrate that compression models capable of producing realistic, high-fidelity reconstructions are substantially more resistant to our attacks. In contrast, low-realism compression models can be broken. Our analysis reveals that this is not due to gradient masking. Rather, realistic reconstructions maintaining distributional alignment with natural images seem to offer inherent robustness. This work highlights a significant obstacle for future adversarial attacks and suggests that developing more effective techniques to overcome realism represents an essential challenge for comprehensive security evaluation.
Authors: Morteza Kimiaei, Vyacheslav Kungurtsev, Brian Olimba
Abstract: Integer and mixed-integer nonlinear programming (INLP, MINLP) are central to logistics, energy, and scheduling, but remain computationally challenging. This survey examines how machine learning and reinforcement learning can enhance exact optimization methods-particularly branch-and-bound (BB)-without compromising global optimality. We cover discrete, continuous, and mixed-integer formulations, and highlight applications such as vehicle routing, hydropower planning, and crew scheduling. We introduce a unified BB framework that embeds learning-based strategies into branching, cut selection, node ordering, and parameter control. Classical algorithms are augmented using supervised, imitation, and reinforcement learning models to accelerate convergence while maintaining correctness. We conclude with a taxonomy of learning methods by solver class and learning paradigm, and outline open challenges in generalization, hybridization, and scaling intelligent solvers.
Authors: Zhimeng Guo, Huaisheng Zhu, Siyuan Xu, Hangfan Zhang, Teng Xiao, Minhao Cheng
Abstract: Protecting intellectual property on LLM-generated code necessitates effective watermarking systems that can operate within code's highly structured, syntactically constrained nature. In this work, we introduce CodeTracer, an innovative adaptive code watermarking framework underpinned by a novel reinforcement learning training paradigm. At its core, CodeTracer features a policy-driven approach that utilizes a parameterized model to intelligently bias token choices during next-token prediction. This strategy ensures that embedded watermarks maintain code functionality while exhibiting subtle yet statistically detectable deviations from typical token distributions. To facilitate policy learning, we devise a comprehensive reward system that seamlessly integrates execution feedback with watermark embedding signals, balancing process-level and outcome-level rewards. Additionally, we employ Gumbel Top-k reparameterization to enable gradient-based optimization of discrete watermarking decisions. Extensive comparative evaluations demonstrate CodeTracer's significant superiority over state-of-the-art baselines in both watermark detectability and the preservation of generated code's functionality.
Authors: Zhouyu Li, Zhijin Yang, Huayue Gu, Xiaojian Wang, Yuchen Liu, Ruozhou Yu
Abstract: Earth observation analytics have the potential to serve many time-sensitive applications. However, due to limited bandwidth and duration of ground-satellite connections, it takes hours or even days to download and analyze data from existing Earth observation satellites, making real-time demands like timely disaster response impossible. Toward real-time analytics, we introduce OrbitChain, a collaborative analytics framework that orchestrates computational resources across multiple satellites in an Earth observation constellation. OrbitChain decomposes analytics applications into microservices and allocates computational resources for time-constrained analysis. A traffic routing algorithm is devised to minimize the inter-satellite communication overhead. OrbitChain adopts a pipeline workflow that completes Earth observation tasks in real-time, facilitates time-sensitive applications and inter-constellation collaborations such as tip-and-cue. To evaluate OrbitChain, we implement a hardware-in-the-loop orbital computing testbed. Experiments show that our system can complete up to 60% analytics workload than existing Earth observation analytics framework while reducing the communication overhead by up to 72%.
Authors: Aman Goel, Daniel Schwartz, Yanjun Qi
Abstract: Large language models (LLMs) have demonstrated impressive capabilities across diverse tasks, but they remain susceptible to hallucinations--generating content that appears plausible but contains factual inaccuracies. We present Finch-Zk, a black-box framework that leverages fine-grained cross-model consistency to detect and mitigate hallucinations in LLM outputs without requiring external knowledge sources. Finch-Zk introduces two key innovations: 1) a cross-model consistency checking strategy that reveals fine-grained inaccuracies by comparing responses generated by diverse models from semantically-equivalent prompts, and 2) a targeted mitigation technique that applies precise corrections to problematic segments while preserving accurate content. Experiments on the FELM dataset show Finch-Zk improves hallucination detection F1 scores by 6-39\% compared to existing approaches. For mitigation, Finch-Zk achieves up to 9 absolute percentage points improvement in answer accuracy on the GPQA-diamond dataset when applied to state-of-the-art models like Llama 4 Maverick and Claude 4 Sonnet. Extensive evaluation on multiple datasets demonstrates that Finch-Zk provides a practical, deployment-ready safeguard for enhancing factual reliability in production LLM systems.
Authors: Ryan Burn
Abstract: I develop an algorithm to produce the piecewise quadratic that computes leave-one-out cross-validation for the lasso as a function of its hyperparameter. The algorithm can be used to find exact hyperparameters that optimize leave-one-out cross-validation either globally or locally, and its practicality is demonstrated on real-world data sets. I also show how the algorithm can be modified to compute approximate leave-one-out cross-validation, making it suitable for larger data sets.
Authors: Ehsan Yousefzadeh-Asl-Miandoab, Reza Karimzadeh, Bulat Ibragimov, Florina M. Ciorba, P{\i}nar T\"oz\"un
Abstract: GPUs running deep learning (DL) workloads are frequently underutilized. Collocating multiple DL training tasks on the same GPU can improve utilization but introduces two key risks: (1) out-of-memory (OOM) crashes for newly scheduled tasks, and (2) severe performance interference among co-running tasks, which can negate any throughput gains. These issues reduce system robustness, quality of service, and energy efficiency. We present CARMA, a task-level, collocation-aware resource management system for the server-scale. CARMA addresses collocation challenges via (1) fine-grained monitoring and bookkeeping of GPUs and a collocation risk analysis that filters out the high-risk GPUs; (2) task placement policies that cap GPU utilization to avoid OOMs and limit interference; (3) integration of GPU memory need estimators for DL tasks to minimize OOMs during collocation; and (4) a lightweight recovery method that relaunches jobs crashed due to OOMs. Our evaluation on a DL training workload derived from real-world traces shows that CARMA uses GPUs more efficiently by making more informed collocation decisions: for the best-performing collocation policy, CARMA increases GPU streaming multiprocessor (SM) utilization by 54%, the parallelism achieved per SM by 61%, and memory use by 62%. This results in a $\sim$35% and $\sim$15% reduction in the end-to-end execution time (makespan) and GPU energy consumption, respectively, for this workload.
Authors: Wu Lin, Scott C. Lowe, Felix Dangel, Runa Eschenhagen, Zikun Xu, Roger B. Grosse
Abstract: Shampoo and its efficient variant, SOAP, employ structured second-moment estimations and have shown strong performance for training neural networks (NNs). In practice, however, Shampoo typically requires step-size grafting with Adam to be competitive, and SOAP mitigates this by applying Adam in Shampoo's eigenbasis -- at the cost of additional memory overhead from Adam in both methods. Prior analyses have largely relied on the Frobenius norm to motivate these estimation schemes. We instead recast their estimation procedures as covariance estimation under Kullback-Leibler (KL) divergence minimization, revealing a previously overlooked theoretical limitation and motivating principled redesigns. Building on this perspective, we develop $\textbf{KL-Shampoo}$ and $\textbf{KL-SOAP}$, practical schemes that match or exceed the performance of Shampoo and SOAP in NN pre-training while achieving SOAP-level per-iteration runtime. Notably, KL-Shampoo does not rely on Adam to attain competitive performance, eliminating the memory overhead introduced by Adam. Across our experiments, KL-Shampoo consistently outperforms SOAP, Shampoo, and even KL-SOAP, establishing the KL-based approach as a compelling foundation for designing structured methods in NN optimization.
Authors: Jacqueline Maasch, John Kalantari, Kia Khezeli
Abstract: On-the-fly reasoning often requires adaptation to novel problems under limited data and distribution shift. This work introduces CausalARC: an experimental testbed for AI reasoning in low-data and out-of-distribution regimes, modeled after the Abstraction and Reasoning Corpus (ARC). Each CausalARC reasoning task is sampled from a fully specified causal world model, formally expressed as a structural causal model. Principled data augmentations provide observational, interventional, and counterfactual feedback about the world model in the form of few-shot, in-context learning demonstrations. As a proof-of-concept, we illustrate the use of CausalARC for four language model evaluation settings: (1) abstract reasoning with test-time training, (2) counterfactual reasoning with in-context learning, (3) program synthesis, and (4) causal discovery with logical reasoning. Within- and between-model performance varied heavily across tasks, indicating room for significant improvement in language model reasoning.
Authors: Laura Boggia, Bogdan Malaescu
Abstract: Anomaly detection in multivariate time series is crucial to ensure the quality of data coming from a physics experiment. Accurately identifying the moments when unexpected errors or defects occur is essential, yet challenging due to scarce labels, unknown anomaly types, and complex correlations across dimensions. To address the scarcity and unreliability of labelled data, we use the Lorenzetti Simulator to generate synthetic events with injected calorimeter anomalies. We then assess the sensitivity of several time series anomaly detection methods, including transformer-based and other deep learning models. The approach employed here is generic and applicable to different detector designs and defects.
Authors: Yihao Guo, Haocheng Bian, Liutong Zhou, Ze Wang, Zhaoyi Zhang, Francois Kawala, Milan Dean, Ian Fischer, Yuantao Peng, Noyan Tokgozoglu, Ivan Barrientos, Riyaaz Shaik, Rachel Li, Chandru Venkataraman, Reza Shifteh Far, Moses Pawar, Venkat Sundaranatha, Michael Xu, Frank Chu
Abstract: With the deployment of Large Language Models (LLMs) in interactive applications, online malicious intent detection has become increasingly critical. However, existing approaches fall short of handling diverse and complex user queries in real time. To address these challenges, we introduce ADRAG (Adversarial Distilled Retrieval-Augmented Guard), a two-stage framework for robust and efficient online malicious intent detection. In the training stage, a high-capacity teacher model is trained on adversarially perturbed, retrieval-augmented inputs to learn robust decision boundaries over diverse and complex user queries. In the inference stage, a distillation scheduler transfers the teacher's knowledge into a compact student model, with a continually updated knowledge base collected online. At deployment, the compact student model leverages top-K similar safety exemplars retrieved from the online-updated knowledge base to enable both online and real-time malicious query detection. Evaluations across ten safety benchmarks demonstrate that ADRAG, with a 149M-parameter model, achieves 98.5% of WildGuard-7B's performance, surpasses GPT-4 by 3.3% and Llama-Guard-3-8B by 9.5% on out-of-distribution detection, while simultaneously delivering up to 5.6x lower latency at 300 queries per second (QPS) in real-time applications.
Authors: Debarpan Bhattacharya, Apoorva Kulkarni, Sriram Ganapathy
Abstract: The accurate trust assessment of multimodal large language models (MLLMs) generated predictions, which can enable selective prediction and improve user confidence, is challenging due to the diverse multi-modal input paradigms. We propose Functionally Equivalent Sampling for Trust Assessment (FESTA), a multimodal input sampling technique for MLLMs, that generates an uncertainty measure based on the equivalent and complementary input samplings. The proposed task-preserving sampling approach for uncertainty quantification expands the input space to probe the consistency (through equivalent samples) and sensitivity (through complementary samples) of the model. FESTA uses only input-output access of the model (black-box), and does not require ground truth (unsupervised). The experiments are conducted with various off-the-shelf multi-modal LLMs, on both visual and audio reasoning tasks. The proposed FESTA uncertainty estimate achieves significant improvement (33.3% relative improvement for vision-LLMs and 29.6% relative improvement for audio-LLMs) in selective prediction performance, based on area-under-receiver-operating-characteristic curve (AUROC) metric in detecting mispredictions. The code implementation is open-sourced.
Authors: Xiaolong Li, Zhi-Qin John Xu, Peiting You, Yifei Zhu
Abstract: Passive intermodulation (PIM) has emerged as a critical source of self-interference in modern MIMO-OFDM systems, especially under the stringent requirements of 5G and beyond. Conventional cancellation methods often rely on complex nonlinear models with limited scalability and high computational cost. In this work, we propose a lightweight deep learning framework for PIM cancellation that leverages depthwise separable convolutions and dilated convolutions to efficiently capture nonlinear dependencies across antennas and subcarriers. To further enhance convergence, we adopt a cyclic learning rate schedule and gradient clipping. In a controlled MIMO experimental setup, the method effectively suppresses third-order passive intermodulation (PIM) distortion, achieving up to 29dB of average power error (APE) with only 11k trainable parameters. These results highlight the potential of compact neural architectures for scalable interference mitigation in future wireless communication systems.
Authors: Daniel Maninger, Leon Chemnitz, Amir Molzam Sharifloo, Jannis Brugger, Mira Mezini
Abstract: API integration is a cornerstone of our digital infrastructure, enabling software systems to connect and interact. However, as shown by many studies, writing or generating correct code to invoke APIs, particularly web APIs, is challenging. Although large language models (LLMs) have become popular in software development, their effectiveness in automating the generation of web API integration code remains unexplored. In order to address this, we present WAPIIBench, a dataset and evaluation pipeline designed to assess the ability of LLMs to generate web API invocation code. Our experiments with several open-source LLMs reveal that generating API invocations poses a significant challenge, resulting in hallucinated endpoints, incorrect argument usage, and other errors. None of the evaluated open-source models was able to solve more than 40% of the tasks.
Authors: Samuel Schapiro, Sumuk Shashidhar, Alexi Gladstone, Jonah Black, Royce Moon, Dilek Hakkani-Tur, Lav R. Varshney
Abstract: Artificial intelligence (AI) systems, and Large Language Models (LLMs) in particular, are increasingly employed for creative tasks like scientific idea generation, constituting a form of generalization from training data unaddressed by existing conceptual frameworks. Despite its similarities to compositional generalization (CG), combinatorial creativity (CC) is an open-ended ability. Instead of evaluating for accuracy or correctness against fixed targets, which would contradict the open-ended nature of CC, we propose a theoretical framework and algorithmic task for evaluating outputs by their degrees of novelty and utility. From here, we make several important empirical contributions: (1) We obtain the first insights into the scaling behavior of creativity for LLMs. (2) We discover that, for fixed compute budgets, there exist optimal model depths and widths for creative ability. (3) We find that the ideation-execution gap, whereby LLMs excel at generating novel scientific ideas but struggle to ensure their practical feasibility, may be explained by a more fundamental novelty-utility tradeoff characteristic of creativity algorithms in general. Importantly, this tradeoff remains persistent even at scale, casting doubt on the long-term creative potential of LLMs in their current form. Together, our conceptual framework and empirical findings provide a foundation for understanding and improving creativity in modern AI models, bridging the gap between human and machine intelligence.
Authors: Shilin Lu, Zhuming Lian, Zihan Zhou, Shaocong Zhang, Chen Zhao, Adams Wai-Kin Kong
Abstract: Image composition aims to seamlessly insert a user-specified object into a new scene, but existing models struggle with complex lighting (e.g., accurate shadows, water reflections) and diverse, high-resolution inputs. Modern text-to-image diffusion models (e.g., SD3.5, FLUX) already encode essential physical and resolution priors, yet lack a framework to unleash them without resorting to latent inversion, which often locks object poses into contextually inappropriate orientations, or brittle attention surgery. We propose SHINE, a training-free framework for Seamless, High-fidelity Insertion with Neutralized Errors. SHINE introduces manifold-steered anchor loss, leveraging pretrained customization adapters (e.g., IP-Adapter) to guide latents for faithful subject representation while preserving background integrity. Degradation-suppression guidance and adaptive background blending are proposed to further eliminate low-quality outputs and visible seams. To address the lack of rigorous benchmarks, we introduce ComplexCompo, featuring diverse resolutions and challenging conditions such as low lighting, strong illumination, intricate shadows, and reflective surfaces. Experiments on ComplexCompo and DreamEditBench show state-of-the-art performance on standard metrics (e.g., DINOv2) and human-aligned scores (e.g., DreamSim, ImageReward, VisionReward). Code and benchmark will be publicly available upon publication.
Authors: Jiye Lee, Chenghui Li, Linh Tran, Shih-En Wei, Jason Saragih, Alexander Richard, Hanbyul Joo, Shaojie Bai
Abstract: We present an audio-driven real-time system for animating photorealistic 3D facial avatars with minimal latency, designed for social interactions in virtual reality for anyone. Central to our approach is an encoder model that transforms audio signals into latent facial expression sequences in real time, which are then decoded as photorealistic 3D facial avatars. Leveraging the generative capabilities of diffusion models, we capture the rich spectrum of facial expressions necessary for natural communication while achieving real-time performance (<15ms GPU time). Our novel architecture minimizes latency through two key innovations: an online transformer that eliminates dependency on future inputs and a distillation pipeline that accelerates iterative denoising into a single step. We further address critical design challenges in live scenarios for processing continuous audio signals frame-by-frame while maintaining consistent animation quality. The versatility of our framework extends to multimodal applications, including semantic modalities such as emotion conditions and multimodal sensors with head-mounted eye cameras on VR headsets. Experimental results demonstrate significant improvements in facial animation accuracy over existing offline state-of-the-art baselines, achieving 100 to 1000 times faster inference speed. We validate our approach through live VR demonstrations and across various scenarios such as multilingual speeches.
Authors: Yaxin Hou, Bo Han, Yuheng Jia, Hui Liu, Junhui Hou
Abstract: Current long-tailed semi-supervised learning methods assume that labeled data exhibit a long-tailed distribution, and unlabeled data adhere to a typical predefined distribution (i.e., long-tailed, uniform, or inverse long-tailed). However, the distribution of the unlabeled data is generally unknown and may follow an arbitrary distribution. To tackle this challenge, we propose a Controllable Pseudo-label Generation (CPG) framework, expanding the labeled dataset with the progressively identified reliable pseudo-labels from the unlabeled dataset and training the model on the updated labeled dataset with a known distribution, making it unaffected by the unlabeled data distribution. Specifically, CPG operates through a controllable self-reinforcing optimization cycle: (i) at each training step, our dynamic controllable filtering mechanism selectively incorporates reliable pseudo-labels from the unlabeled dataset into the labeled dataset, ensuring that the updated labeled dataset follows a known distribution; (ii) we then construct a Bayes-optimal classifier using logit adjustment based on the updated labeled data distribution; (iii) this improved classifier subsequently helps identify more reliable pseudo-labels in the next training step. We further theoretically prove that this optimization cycle can significantly reduce the generalization error under some conditions. Additionally, we propose a class-aware adaptive augmentation module to further improve the representation of minority classes, and an auxiliary branch to maximize data utilization by leveraging all labeled and unlabeled samples. Comprehensive evaluations on various commonly used benchmark datasets show that CPG achieves consistent improvements, surpassing state-of-the-art methods by up to $\textbf{15.97%}$ in accuracy. The code is available at https://github.com/yaxinhou/CPG.
Authors: Tong Mao, Jinchao Xu
Abstract: We prove a saturation theorem for linearized shallow ReLU$^k$ neural networks on the unit sphere $\mathbb S^d$. For any antipodally quasi-uniform set of centers, if the target function has smoothness $r>\tfrac{d+2k+1}{2}$, then the best $\mathcal{L}^2(\mathbb S^d)$ approximation cannot converge faster than order $n^{-\frac{d+2k+1}{2d}}$. This lower bound matches existing upper bounds, thereby establishing the exact saturation order $\tfrac{d+2k+1}{2d}$ for such networks. Our results place linearized neural-network approximation firmly within the classical saturation framework and show that, although ReLU$^k$ networks outperform finite elements under equal degrees $k$, this advantage is intrinsically limited.
Authors: Andreas Engelhardt, Mark Boss, Vikram Voleti, Chun-Han Yao, Hendrik P. A. Lensch, Varun Jampani
Abstract: We present Stable Video Materials 3D (SViM3D), a framework to predict multi-view consistent physically based rendering (PBR) materials, given a single image. Recently, video diffusion models have been successfully used to reconstruct 3D objects from a single image efficiently. However, reflectance is still represented by simple material models or needs to be estimated in additional steps to enable relighting and controlled appearance edits. We extend a latent video diffusion model to output spatially varying PBR parameters and surface normals jointly with each generated view based on explicit camera control. This unique setup allows for relighting and generating a 3D asset using our model as neural prior. We introduce various mechanisms to this pipeline that improve quality in this ill-posed setting. We show state-of-the-art relighting and novel view synthesis performance on multiple object-centric datasets. Our method generalizes to diverse inputs, enabling the generation of relightable 3D assets useful in AR/VR, movies, games and other visual media.
Authors: Siqi Zhu, David Zhang, Pedro Cisneros-Velarde, Jiaxuan You
Abstract: Large Language Models (LLMs) have achieved remarkable progress in reasoning, yet sometimes produce responses that are suboptimal for users in tasks such as writing, information seeking, or providing practical guidance. Conventional alignment practices typically assume that maximizing model reward also maximizes user welfare, but this assumption frequently fails in practice: models may over-clarify or generate overly verbose reasoning when users prefer concise answers. Such behaviors resemble the prisoner's dilemma, where individually rational choices lead to socially suboptimal outcomes. The fundamental challenge is the lack of a principled decision making mechanism that mutually benefits both the LLM and the user. We propose Game-Theoretic Alignment (GTAlign), an alignment framework that integrates game-theoretic decision making into both reasoning and training. During reasoning, the model explicitly treats user-LLM interaction as a strategic game: it constructs payoff matrices within its reasoning chain to estimate welfare for both itself and the user, and then selects actions that are mutually beneficial. During training, we introduce a social welfare reward that reinforces cooperative responses, aligning model behavior with socially efficient outcomes. In addition, we introduce an inference technique that leverages game-theoretic reasoning to dynamically adapt LLM's response when pricing policies of LLM service change. Extensive experiments demonstrate that GTAlign substantially improves reasoning efficiency, answer quality, and social welfare compared to baselines across diverse tasks. The code is available at https://github.com/ulab-uiuc/GTAlign .
Authors: Joachim Diederich
Abstract: We present a novel framework for training large language models with continuously adjustable internal representations that span the full spectrum from localist (interpretable, rule-based) to distributed (generalizable, efficient) encodings. The key innovation is a locality dial, a tunable parameter that dynamically controls the degree of localization during both training and inference without requiring model retraining. This is achieved through group sparsity penalties on attention mechanisms, information-theoretic anchor design, and dynamic rule injection. We provide rigorous mathematical proofs establishing explicit threshold conditions under which attention provably concentrates on semantically relevant blocks, with exponential bounds on attention entropy and pointer fidelity. Specifically, we prove that when group sparsity penalties exceed certain threshold values, the model's attention mechanisms concentrate on semantically relevant blocks, achieving low entropy and high fidelity with negligible error. This framework enables practitioners to continuously interpolate between interpretable and high-performance modes, supporting applications in regulated domains requiring both transparency and capability.
Authors: Ali Atiah Alzahrani
Abstract: We tackle high-dimensional, path-dependent valuation and control and introduce a deep BSDE/2BSDE solver that couples truncated log-signatures with a neural rough differential equation (RDE) backbone. The architecture aligns stochastic analysis with sequence-to-path learning: a CVaR-tilted terminal objective targets left-tail risk, while an optional second-order (2BSDE) head supplies curvature estimates for risk-sensitive control. Under matched compute and parameter budgets, the method improves accuracy, tail fidelity, and training stability across Asian and barrier option pricing and portfolio control: at d=200 it achieves CVaR(0.99)=9.80% versus 12.00-13.10% for strong baselines, attains the lowest HJB residual (0.011), and yields the lowest RMSEs for Z and Gamma. Ablations over truncation depth, local windows, and tilt parameters confirm complementary gains from the sequence-to-path representation and the 2BSDE head. Taken together, the results highlight a bidirectional dialogue between stochastic analysis and modern deep learning: stochastic tools inform representations and objectives, while sequence-to-path models expand the class of solvable financial models at scale.
Authors: Shingo Kodama, Haya Diwan, Lucas Rosenblatt, R. Teal Witter, Niv Cohen
Abstract: The rapid spread of text generated by large language models (LLMs) makes it increasingly difficult to distinguish authentic human writing from machine output. Watermarking offers a promising solution: model owners can embed an imperceptible signal into generated text, marking its origin. Most leading approaches seed an LLM's next-token sampling with a pseudo-random key that can later be recovered to identify the text as machine-generated, while only minimally altering the model's output distribution. However, these methods suffer from two related issues: (i) watermarks are brittle to simple surface-level edits such as paraphrasing or reordering; and (ii) adversaries can append unrelated, potentially harmful text that inherits the watermark, risking reputational damage to model owners. To address these issues, we introduce SimKey, a semantic key module that strengthens watermark robustness by tying key generation to the meaning of prior context. SimKey uses locality-sensitive hashing over semantic embeddings to ensure that paraphrased text yields the same watermark key, while unrelated or semantically shifted text produces a different one. Integrated with state-of-the-art watermarking schemes, SimKey improves watermark robustness to paraphrasing and translation while preventing harmful content from false attribution, establishing semantic-aware keying as a practical and extensible watermarking direction.
Authors: Kun Lei, Huanyu Li, Dongjie Yu, Zhenyu Wei, Lingxiao Guo, Zhennan Jiang, Ziyu Wang, Shiyu Liang, Huazhe Xu
Abstract: Real-world robotic manipulation in homes and factories demands reliability, efficiency, and robustness that approach or surpass skilled human operators. We present RL-100, a real-world reinforcement learning training framework built on diffusion visuomotor policies trained by supervised learning. RL-100 introduces a three-stage pipeline. First, imitation learning leverages human priors. Second, iterative offline reinforcement learning uses an Offline Policy Evaluation procedure, abbreviated OPE, to gate PPO-style updates that are applied in the denoising process for conservative and reliable improvement. Third, online reinforcement learning eliminates residual failure modes. An additional lightweight consistency distillation head compresses the multi-step sampling process in diffusion into a single-step policy, enabling high-frequency control with an order-of-magnitude reduction in latency while preserving task performance. The framework is task-, embodiment-, and representation-agnostic and supports both 3D point clouds and 2D RGB inputs, a variety of robot platforms, and both single-step and action-chunk policies. We evaluate RL-100 on seven real-robot tasks spanning dynamic rigid-body control, such as Push-T and Agile Bowling, fluids and granular pouring, deformable cloth folding, precise dexterous unscrewing, and multi-stage orange juicing. RL-100 attains 100\% success across evaluated trials for a total of 900 out of 900 episodes, including up to 250 out of 250 consecutive trials on one task. The method achieves near-human teleoperation or better time efficiency and demonstrates multi-hour robustness with uninterrupted operation lasting up to two hours.
Authors: Akira Okutomi
Abstract: We reinterpret Kant's Critique of Pure Reason as a theory of feedback stability, viewing reason as a regulator that keeps inference within the bounds of possible experience. We formalize this intuition via a composite instability index (H-Risk) combining spectral margin, conditioning, temporal sensitivity, and innovation amplification. In linear-Gaussian simulations, higher H-Risk predicts overconfident errors even under formal stability, revealing a gap between nominal and epistemic stability. Extending to large language models (LLMs), we observe preliminary correlations between internal fragility and miscalibration or hallucination (confabulation), and find that lightweight critique prompts may modestly improve or worsen calibration in small-scale tests. These results suggest a structural bridge between Kantian self-limitation and feedback control, offering a principled lens to diagnose and potentially mitigate overconfidence in reasoning systems.
Authors: Zelong Bi, Pierre Lafaye de Micheaux
Abstract: Local principal component analysis (Local PCA) has proven to be an effective tool for estimating the intrinsic dimension of a manifold. More recently, curvature-adjusted PCA (CA-PCA) has improved upon this approach by explicitly accounting for the curvature of the underlying manifold, rather than assuming local flatness. Building on these insights, we propose a general framework for manifold dimension estimation that captures the manifold's local graph structure by integrating PCA with regression-based techniques. Within this framework, we introduce two representative estimators: quadratic embedding (QE) and total least squares (TLS). Experiments on both synthetic and real-world datasets demonstrate that these methods perform competitively with, and often outperform, state-of-the-art alternatives.
Authors: Zehao Ni, Yonghao He, Lingfeng Qian, Jilei Mao, Fa Fu, Wei Sui, Hu Su, Junran Peng, Zhipeng Wang, Bin He
Abstract: In the context of imitation learning, visuomotor-based diffusion policy learning is one of the main directions in robotic manipulation. Most of these approaches rely on point clouds as observation inputs and construct scene representations through point clouds feature learning, which enables them to achieve remarkable accuracy. However, the existing literature lacks an in-depth exploration of vision-only solutions that have significant potential. In this paper, we propose a Vision-Only and single-view Diffusion Policy learning method (VO-DP) that leverages pretrained visual foundation models to achieve effective fusion of semantic and geometric features. We utilize intermediate features from VGGT incorporating semantic features from DINOv2 and geometric features from Alternating Attention blocks. Features are fused via cross-attention and spatially compressed with a CNN to form the input to the policy head. Extensive experiments demonstrate that VO-DP not only outperforms the vision-only baseline DP significantly but also exhibits distinct performance trends against the point cloud-based method DP3: in simulation tasks, VO-DP achieves an average success rate of 64.6% on par with DP3 64.0% and far higher than DP 34.8%, while in real-world tasks, it reaches 87.9%, outperforming both DP3 67.5% and DP 11.2% by a notable margin. Further robustness evaluations confirm that VO-DP remains highly stable under varying conditions including color, size, background, and lighting. Lastly, we open-source a training library for robotic manipulation. Built on Accelerate, this library supports multi-machine and multi-GPU parallel training, as well as mixed precision training. It is compatible with visuomotor policies such as DP, DP3 and VO-DP, and also supports the RoboTwin simulator.
Authors: Ivan Kartashov, Mariia Pushkareva, Iakov Karandashev
Abstract: This paper introduces SpikeFit, a novel training method for Spiking Neural Networks (SNNs) that enables efficient inference on neuromorphic hardware, considering all its stringent requirements: the number of neurons and synapses that can fit on a single device, and lower bit-width representations (e.g., 4-bit, 8-bit). Unlike conventional compressing approaches that address only a subset of these requirements (limited numerical precision and limited number of neurons in the network), SpikeFit treats the allowed weights' discrete values themselves as learnable parameters co-optimized with the model, allowing for optimal Clusterization-Aware Training (CAT) of the model's weights at low precision (2-, 4-, or 8-bit) which results in higher network compression efficiency, as well as limiting the number of unique synaptic connections to a value required by neuromorphic processor. This joint optimization allows SpikeFit to find a discrete weight set aligned with hardware constraints, enabling the most complete deployment across a broader range of neuromorphic processors than existing methods of SNN compression support. Moreover, SpikeFit introduces a new hardware-friendly Fisher Spike Contribution (FSC) pruning method showing the state-of-the-art performance. We demonstrate that for spiking neural networks constrained to only four unique synaptic weight values (M = 4), our SpikeFit method not only outperforms state-of-the-art SNNs compression methods and conventional baselines combining extreme quantization schemes and clustering algorithms, but also meets a wider range of neuromorphic hardware requirements and provides the lowest energy use in experiments.
Authors: Tian Guo, Emmanuel Hauptmann
Abstract: In quantitative investing, return prediction supports various tasks, including stock selection, portfolio optimization, and risk management. Quantitative factors, such as valuation, quality, and growth, capture various characteristics of stocks. Unstructured financial data, like news and transcripts, has attracted growing attention, driven by recent advances in large language models (LLMs). This paper examines effective methods for leveraging multimodal factors and newsflow in return prediction and stock selection. First, we introduce a fusion learning framework to learn a unified representation from factors and newsflow representations generated by an LLM. Within this framework, we compare three representative methods: representation combination, representation summation, and attentive representations. Next, building on empirical observations from fusion learning, we explore the mixture model that adaptively combines predictions made by single modalities and their fusion. To mitigate the training instability observed in the mixture model, we introduce a decoupled training approach with theoretical insights. Finally, our experiments on real investment universes yield several insights into effective multimodal modeling of factors and news for stock return prediction and selection.
Authors: Tianci Bi, Xiaoyi Zhang, Yan Lu, Nanning Zheng
Abstract: The performance of Latent Diffusion Models (LDMs) is critically dependent on the quality of their visual tokenizer. While recent works have explored incorporating Vision Foundation Models (VFMs) via distillation, we identify a fundamental flaw in this approach: it inevitably weakens the robustness of alignment with the original VFM, causing the aligned latents to deviate semantically under distribution shifts. In this paper, we bypass distillation by proposing a more direct approach: Vision Foundation Model Variational Autoencoder (VFM-VAE). To resolve the inherent tension between the VFM's semantic focus and the need for pixel-level fidelity, we redesign the VFM-VAE decoder with Multi-Scale Latent Fusion and Progressive Resolution Reconstruction blocks, enabling high-quality reconstruction from spatially coarse VFM features. Furthermore, we provide a comprehensive analysis of representation dynamics during diffusion training, introducing the proposed SE-CKNNA metric as a more precise tool for this diagnosis. This analysis allows us to develop a joint tokenizer-diffusion alignment strategy that dramatically accelerates convergence. Our innovations in tokenizer design and training strategy lead to superior performance and efficiency: our system reaches a gFID (w/o CFG) of 2.20 in merely 80 epochs (a 10x speedup over prior tokenizers). With continued training to 640 epochs, it further attains a gFID (w/o CFG) of 1.62, establishing direct VFM integration as a superior paradigm for LDMs.
Authors: Joe Meyer, Divyansha Lachi, Mahmoud Mohammadi, Roshan Reddy Upendra, Eva L. Dyer, Mark Li, Tom Palczewski
Abstract: Relational multi-table data is common in domains such as e-commerce, healthcare, and scientific research, and can be naturally represented as heterogeneous temporal graphs with multi-modal node attributes. Existing graph neural networks (GNNs) rely on schema-specific feature encoders, requiring separate modules for each node type and feature column, which hinders scalability and parameter sharing. We introduce RELATE (Relational Encoder for Latent Aggregation of Typed Entities), a schema-agnostic, plug-and-play feature encoder that can be used with any general purpose GNN. RELATE employs shared modality-specific encoders for categorical, numerical, textual, and temporal attributes, followed by a Perceiver-style cross-attention module that aggregates features into a fixed-size, permutation-invariant node representation. We evaluate RELATE on ReLGNN and HGT in the RelBench benchmark, where it achieves performance within 3% of schema-specific encoders while reducing parameter counts by up to 5x. This design supports varying schemas and enables multi-dataset pretraining for general-purpose GNNs, paving the way toward foundation models for relational graph data.
Authors: Giora Simchoni, Saharon Rosset
Abstract: We present MMbeddings, a probabilistic embedding approach that reinterprets categorical embeddings through the lens of nonlinear mixed models, effectively bridging classical statistical theory with modern deep learning. By treating embeddings as latent random effects within a variational autoencoder framework, our method substantially decreases the number of parameters -- from the conventional embedding approach of cardinality $\times$ embedding dimension, which quickly becomes infeasible with large cardinalities, to a significantly smaller, cardinality-independent number determined primarily by the encoder architecture. This reduction dramatically mitigates overfitting and computational burden in high-cardinality settings. Extensive experiments on simulated and real datasets, encompassing collaborative filtering and tabular regression tasks using varied architectures, demonstrate that MMbeddings consistently outperforms traditional embeddings, underscoring its potential across diverse machine learning applications.
Authors: Swagatam Das
Abstract: This paper develops a rigorous probabilistic framework that extends denoising diffusion models to the setting of noncommutative random variables. Building on Voiculescu's theory of free entropy and free Fisher information, we formulate diffusion and reverse processes governed by operator-valued stochastic dynamics whose spectral measures evolve by additive convolution. Using tools from free stochastic analysis -- including a Malliavin calculus and a Clark--Ocone representation -- we derive the reverse-time stochastic differential equation driven by the conjugate variable, the analogue of the classical score function. The resulting dynamics admit a gradient-flow structure in the noncommutative Wasserstein space, establishing an information-geometric link between entropy production, transport, and deconvolution. We further construct a variational scheme analogous to the Jordan--Kinderlehrer--Otto (JKO) formulation and prove convergence toward the semicircular equilibrium. The framework provides functional inequalities (free logarithmic Sobolev, Talagrand, and HWI) that quantify entropy dissipation and Wasserstein contraction. These results unify diffusion-based generative modeling with the geometry of operator-valued information, offering a mathematical foundation for generative learning on structured and high-dimensional data.
Authors: Vishal Halder, Alexandre Reiffers-Masson, Abdeldjalil A\"issa-El-Bey, Gugan Thoppe
Abstract: Let $A \in \mathbb{R}^{m \times n}$ be an arbitrary, known matrix and $e$ a $q$-sparse adversarial vector. Given $y = A x^\star + e$ and $q$, we seek the smallest set containing $x^\star$--hence the one conveying maximal information about $x^\star$--that is uniformly recoverable from $y$ without knowing $e$. While exact recovery of $x^\star$ via strong (and often impractical) structural assumptions on $A$ or $x^\star$ (e.g., restricted isometry, sparsity) is well studied, recoverability for arbitrary $A$ and $x^\star$ remains open. Our main result shows that the best that one can hope to recover is $x^\star + \ker(U)$, where $U$ is the unique projection matrix onto the intersection of rowspaces of all possible submatrices of $A$ obtained by deleting $2q$ rows. Moreover, we prove that every $x$ that minimizes the $\ell_0$-norm of $y - A x$ lies in $x^\star + \ker(U)$, which then gives a constructive approach to recover this set.
Authors: Ann Huang, Mitchell Ostrow, Satpreet H. Singh, Leo Kozachkov, Ila Fiete, Kanaka Rajan
Abstract: In control problems and basic scientific modeling, it is important to compare observations with dynamical simulations. For example, comparing two neural systems can shed light on the nature of emergent computations in the brain and deep neural networks. Recently, Ostrow et al. (2023) introduced Dynamical Similarity Analysis (DSA), a method to measure the similarity of two systems based on their recurrent dynamics rather than geometry or topology. However, DSA does not consider how inputs affect the dynamics, meaning that two similar systems, if driven differently, may be classified as different. Because real-world dynamical systems are rarely autonomous, it is important to account for the effects of input drive. To this end, we introduce a novel metric for comparing both intrinsic (recurrent) and input-driven dynamics, called InputDSA (iDSA). InputDSA extends the DSA framework by estimating and comparing both input and intrinsic dynamic operators using a variant of Dynamic Mode Decomposition with control (DMDc) based on subspace identification. We demonstrate that InputDSA can successfully compare partially observed, input-driven systems from noisy data. We show that when the true inputs are unknown, surrogate inputs can be substituted without a major deterioration in similarity estimates. We apply InputDSA on Recurrent Neural Networks (RNNs) trained with Deep Reinforcement Learning, identifying that high-performing networks are dynamically similar to one another, while low-performing networks are more diverse. Lastly, we apply InputDSA to neural data recorded from rats performing a cognitive task, demonstrating that it identifies a transition from input-driven evidence accumulation to intrinsically-driven decision-making. Our work demonstrates that InputDSA is a robust and efficient method for comparing intrinsic dynamics and the effect of external input on dynamical systems.
Authors: Dinghong Song, Yuan Feng, Yiwei Wang, Shangye Chen, Cyril Guyot, Filip Blagojevic, Hyeran Jeon, Pengfei Su, Dong Li
Abstract: Large Language Models (LLMs) are widely used in generative applications such as chatting, code generation, and reasoning. However, many realworld workloads such as classification, question answering, recommendation, and text embedding rely solely on the prefill stage of inference, where the model encodes input sequences without performing autoregressive decoding. In these prefill only scenarios, the self-attention computation becomes the primary performance bottleneck due to its quadratic complexity with respect to sequence length. In this paper, we observe that semantically different sentences often produce similar attention maps across layers and heads. Building on this insight, we propose AttnCache, a framework that accelerates the prefill stage of LLM inference by retrieving and reusing similar attention maps. Based on an attention map memorization database, AttnCache employs efficient caching and similarity search techniques to identify and reuse pre-cached attention maps during inference, thereby reducing the computational overhead of self-attention. Experimental results show that AttnCache achieves an average of 1.2x end-to-end and 2x attention speedup on CPU, and 1.6x end-to-end and 3x attention speedup on GPU, with negligible accuracy degradation.
Authors: Pei Peng, MingKun Xie, Hang Hao, Tong Jin, ShengJun Huang
Abstract: Object-context shortcuts remain a persistent challenge in vision-language models, undermining zero-shot reliability when test-time scenes differ from familiar training co-occurrences. We recast this issue as a causal inference problem and ask: Would the prediction remain if the object appeared in a different environment? To answer this at inference time, we estimate object and background expectations within CLIP's representation space, and synthesize counterfactual embeddings by recombining object features with diverse alternative contexts sampled from external datasets, batch neighbors, or text-derived descriptions. By estimating the Total Direct Effect and simulating intervention, we further subtract background-only activation, preserving beneficial object-context interactions while mitigating hallucinated scores. Without retraining or prompt design, our method substantially improves both worst-group and average accuracy on context-sensitive benchmarks, establishing a new zero-shot state of the art. Beyond performance, our framework provides a lightweight representation-level counterfactual approach, offering a practical causal avenue for debiased and reliable multimodal reasoning.
Authors: Kimi Team, Yu Zhang, Zongyu Lin, Xingcheng Yao, Jiaxi Hu, Fanqing Meng, Chengyin Liu, Xin Men, Songlin Yang, Zhiyuan Li, Wentao Li, Enzhe Lu, Weizhou Liu, Yanru Chen, Weixin Xu, Longhui Yu, Yejie Wang, Yu Fan, Longguang Zhong, Enming Yuan, Dehao Zhang, Yizhi Zhang, T. Y. Liu, Haiming Wang, Shengjun Fang, Weiran He, Shaowei Liu, Yiwei Li, Jianlin Su, Jiezhong Qiu, Bo Pang, Junjie Yan, Zhejun Jiang, Weixiao Huang, Bohong Yin, Jiacheng You, Chu Wei, Zhengtao Wang, Chao Hong, Yutian Chen, Guanduo Chen, Yucheng Wang, Huabin Zheng, Feng Wang, Yibo Liu, Mengnan Dong, Zheng Zhang, Siyuan Pan, Wenhao Wu, Yuhao Wu, Longyu Guan, Jiawen Tao, Guohong Fu, Xinran Xu, Yuzhi Wang, Guokun Lai, Yuxin Wu, Xinyu Zhou, Zhilin Yang, Yulun Du
Abstract: We introduce Kimi Linear, a hybrid linear attention architecture that, for the first time, outperforms full attention under fair comparisons across various scenarios -- including short-context, long-context, and reinforcement learning (RL) scaling regimes. At its core lies Kimi Delta Attention (KDA), an expressive linear attention module that extends Gated DeltaNet with a finer-grained gating mechanism, enabling more effective use of limited finite-state RNN memory. Our bespoke chunkwise algorithm achieves high hardware efficiency through a specialized variant of the Diagonal-Plus-Low-Rank (DPLR) transition matrices, which substantially reduces computation compared to the general DPLR formulation while remaining more consistent with the classical delta rule. We pretrain a Kimi Linear model with 3B activated parameters and 48B total parameters, based on a layerwise hybrid of KDA and Multi-Head Latent Attention (MLA). Our experiments show that with an identical training recipe, Kimi Linear outperforms full MLA with a sizeable margin across all evaluated tasks, while reducing KV cache usage by up to 75% and achieving up to 6 times decoding throughput for a 1M context. These results demonstrate that Kimi Linear can be a drop-in replacement for full attention architectures with superior performance and efficiency, including tasks with longer input and output lengths. To support further research, we open-source the KDA kernel and vLLM implementations, and release the pre-trained and instruction-tuned model checkpoints.
Authors: Nelson Matti\'e, Arturo Sanchez-Azofeifa, Pablo Crespo-Peremarch, Juan-Ygnacio L\'opez-Hern\'andez
Abstract: According to the Paris Climate Change Agreement, all nations are required to submit reports on their greenhouse gas emissions and absorption every two years by 2024. Consequently, forests play a crucial role in reducing carbon emissions, which is essential for meeting these obligations. Recognizing the significance of forest conservation in the global battle against climate change, Article 5 of the Paris Agreement emphasizes the need for high-quality forest data. This study focuses on enhancing methods for mapping aboveground biomass in tropical dry forests. Tropical dry forests are considered one of the least understood tropical forest environments; therefore, there is a need for accurate approaches to estimate carbon pools. We employ a comparative analysis of AGB estimates, utilizing different discrete and full-waveform laser scanning datasets in conjunction with Ordinary Least Squares and Bayesian approaches SVM. Airborne Laser Scanning, Unmanned Laser Scanning, and Space Laser Scanning were used as independent variables for extracting forest metrics. Variable selection, SVM regression tuning, and cross-validation via a machine-learning approach were applied to account for overfitting and underfitting. The results indicate that six key variables primarily related to tree height: Elev\.minimum, Elev\.L3, lev\.MAD.mode, Elev\.mode, Elev\.MAD\.median, and Elev\.skewness, are important for AGB estimation using ALSD and ULSD, while Leaf Area Index, canopy coverage and height, terrain elevation, and full-waveform signal energy emerged as the most vital variables. AGB values estimated from ten permanent tropical dry forest plots in Costa Rica Guanacaste province ranged from 26.02 Mg/ha to 175.43 Mg/ha. The SVM regressions demonstrated a 17.89 error across all laser scanning systems, with SLSF W exhibiting the lowest error 17.07 in estimating total biomass per plot.
Authors: Ilyass Moummad, Kawtar Zaher, Herv\'e Go\"eau, Alexis Joly
Abstract: Efficient large-scale retrieval requires representations that are both compact and discriminative. Foundation models provide powerful visual and multimodal embeddings, but nearest neighbor search in these high-dimensional spaces is computationally expensive. Hashing offers an efficient alternative by enabling fast Hamming distance search with binary codes, yet existing approaches often rely on complex pipelines, multi-term objectives, designs specialized for a single learning paradigm, and long training times. We introduce CroVCA (Cross-View Code Alignment), a simple and unified principle for learning binary codes that remain consistent across semantically aligned views. A single binary cross-entropy loss enforces alignment, while coding-rate maximization serves as an anti-collapse regularizer to promote balanced and diverse codes. To implement this, we design HashCoder, a lightweight MLP hashing network with a final batch normalization layer to enforce balanced codes. HashCoder can be used as a probing head on frozen embeddings or to adapt encoders efficiently via LoRA fine-tuning. Across benchmarks, CroVCA achieves state-of-the-art results in just 5 training epochs. At 16 bits, it particularly well-for instance, unsupervised hashing on COCO completes in under 2 minutes and supervised hashing on ImageNet100 in about 3 minutes on a single GPU. These results highlight CroVCA's efficiency, adaptability, and broad applicability.