# Research Papers — 2025-10 · bundle page 151 of 217 > Full-text markdown of papers #7501–7550 of 10843 published in 2025-10, newest first. AI-readable mirrors of the full 10843-paper 2025-10 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2025-10/page/151 [← Previous bundle](https://aipod.dev/papers/archive/2025-10/page/150.md) [Next bundle →](https://aipod.dev/papers/archive/2025-10/page/152.md) ### Deceptive Exploration in Multi-armed Bandits Authors: I. Arda Vurankaya, Mustafa O. Karabag, Wesley A. Suttle, Jesse Milzman, David Fridovich-Keil, Ufuk Topcu Published: 2025-10-09 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.08794v1 We consider a multi-armed bandit setting in which each arm has a public and a private reward distribution. An observer expects an agent to follow Thompson Sampling according to the public rewards, however, the deceptive agent aims to quickly identify the best private arm without being noticed. The observer can observe the public rewards and the pulled arms, but not the private rewards. The agent, on the other hand, observes both the public and private rewards. We formalize detectability as a stepwise Kullback-Leibler (KL) divergence constraint between the actual pull probabilities used by the agent and the anticipated pull probabilities by the observer. We model successful pulling of public suboptimal arms as a % Bernoulli process where the success probability decreases with each successful pull, and show these pulls can happen at most at a $Θ(\sqrt{T}) $ rate under the KL constraint. We then formulate a maximin problem based on public and private means, whose solution characterizes the optimal error exponent for best private arm identification. We finally propose an algorithm inspired by top-two algorithms. This algorithm naturally adapts its exploration according to the hardness of pulling arms based on the public suboptimality gaps. We provide numerical examples illustrating the $Θ(\sqrt{T}) $ rate and the behavior of the proposed algorithm. --- ### Alignment, Mining and Fusion: Representation Alignment with Hard Negative Mining and Selective Knowledge Fusion for Medical Visual Question Answering Authors: Yuanhao Zou, Zhaozheng Yin Published: 2025-10-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.08791v1 Medical Visual Question Answering (Med-VQA) is a challenging task that requires a deep understanding of both medical images and textual questions. Although recent works leveraging Medical Vision-Language Pre-training (Med-VLP) have shown strong performance on the Med-VQA task, there is still no unified solution for modality alignment, and the issue of hard negatives remains under-explored. Additionally, commonly used knowledge fusion techniques for Med-VQA may introduce irrelevant information. In this work, we propose a framework to address these challenges through three key contributions: (1) a unified solution for heterogeneous modality alignments across multiple levels, modalities, views, and stages, leveraging methods like contrastive learning and optimal transport theory; (2) a hard negative mining method that employs soft labels for multi-modality alignments and enforces the hard negative pair discrimination; and (3) a Gated Cross-Attention Module for Med-VQA that integrates the answer vocabulary as prior knowledge and selects relevant information from it. Our framework outperforms the previous state-of-the-art on widely used Med-VQA datasets like RAD-VQA, SLAKE, PathVQA and VQA-2019. --- ### COMPASS: Enhancing Agent Long-Horizon Reasoning with Evolving Context Authors: Guangya Wan, Mingyang Ling, Xiaoqi Ren, Rujun Han, Sheng Li, Zizhao Zhang Published: 2025-10-09 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2510.08790v1 Long-horizon tasks that require sustained reasoning and multiple tool interactions remain challenging for LLM agents: small errors compound across steps, and even state-of-the-art models often hallucinate or lose coherence. We identify context management as the central bottleneck -- extended histories cause agents to overlook critical evidence or become distracted by irrelevant information, thus failing to replan or reflect from previous mistakes. To address this, we propose COMPASS (Context-Organized Multi-Agent Planning and Strategy System), a lightweight hierarchical framework that separates tactical execution, strategic oversight, and context organization into three specialized components: (1) a Main Agent that performs reasoning and tool use, (2) a Meta-Thinker that monitors progress and issues strategic interventions, and (3) a Context Manager that maintains concise, relevant progress briefs for different reasoning stages. Across three challenging benchmarks -- GAIA, BrowseComp, and Humanity's Last Exam -- COMPASS improves accuracy by up to 20% relative to both single- and multi-agent baselines. We further introduce a test-time scaling extension that elevates performance to match established DeepResearch agents, and a post-training pipeline that delegates context management to smaller models for enhanced efficiency. --- ### Q-Router: Agentic Video Quality Assessment with Expert Model Routing and Artifact Localization Authors: Shuo Xing, Soumik Dey, Mingyang Wu, Ashirbad Mishra, Naveen Ravipati, Binbin Li, Hansi Wu, Zhengzhong Tu Published: 2025-10-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.08789v2 Video quality assessment (VQA) is a fundamental computer vision task that aims to predict the perceptual quality of a given video in alignment with human judgments. Existing performant VQA models trained with direct score supervision suffer from (1) poor generalization across diverse content and tasks, ranging from user-generated content (UGC), short-form videos, to AI-generated content (AIGC), (2) limited interpretability, and (3) lack of extensibility to novel use cases or content types. We propose Q-Router, an agentic framework for universal VQA with a multi-tier model routing system. Q-Router integrates a diverse set of expert models and employs vision--language models (VLMs) as real-time routers that dynamically reason and then ensemble the most appropriate experts conditioned on the input video semantics. We build a multi-tiered routing system based on the computing budget, with the heaviest tier involving a specific spatiotemporal artifacts localization for interpretability. This agentic design enables Q-Router to combine the complementary strengths of specialized experts, achieving both flexibility and robustness in delivering consistent performance across heterogeneous video sources and tasks. Extensive experiments demonstrate that Q-Router matches or surpasses state-of-the-art VQA models on a variety of benchmarks, while substantially improving generalization and interpretability. Moreover, Q-Router excels on the quality-based question answering benchmark, Q-Bench-Video, highlighting its promise as a foundation for next-generation VQA systems. Finally, we show that Q-Router capably localizes spatiotemporal artifacts, showing potential as a reward function for post-training video generation models. --- ### MLLM as a UI Judge: Benchmarking Multimodal LLMs for Predicting Human Perception of User Interfaces Authors: Reuben A. Luera, Ryan Rossi, Franck Dernoncourt, Samyadeep Basu, Sungchul Kim, Subhojyoti Mukherjee, Puneet Mathur, Ruiyi Zhang, Jihyung Kil, Nedim Lipka, Seunghyun Yoon, Jiuxiang Gu et al. Published: 2025-10-09 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2510.08783v1 In an ideal design pipeline, user interface (UI) design is intertwined with user research to validate decisions, yet studies are often resource-constrained during early exploration. Recent advances in multimodal large language models (MLLMs) offer a promising opportunity to act as early evaluators, helping designers narrow options before formal testing. Unlike prior work that emphasizes user behavior in narrow domains such as e-commerce with metrics like clicks or conversions, we focus on subjective user evaluations across varied interfaces. We investigate whether MLLMs can mimic human preferences when evaluating individual UIs and comparing them. Using data from a crowdsourcing platform, we benchmark GPT-4o, Claude, and Llama across 30 interfaces and examine alignment with human judgments on multiple UI factors. Our results show that MLLMs approximate human preferences on some dimensions but diverge on others, underscoring both their potential and limitations in supplementing early UX research. --- ### Weights initialization of neural networks for function approximation Authors: Xinwen Hu, Yunqing Huang, Nianyu Yi, Peimeng Yin Published: 2025-10-09 Categories: cs.LG, math.NA arXiv: https://arxiv.org/abs/2510.08780v1 Neural network-based function approximation plays a pivotal role in the advancement of scientific computing and machine learning. Yet, training such models faces several challenges: (i) each target function often requires training a new model from scratch; (ii) performance is highly sensitive to architectural and hyperparameter choices; and (iii) models frequently generalize poorly beyond the training domain. To overcome these challenges, we propose a reusable initialization framework based on basis function pretraining. In this approach, basis neural networks are first trained to approximate families of polynomials on a reference domain. Their learned parameters are then used to initialize networks for more complex target functions. To enhance adaptability across arbitrary domains, we further introduce a domain mapping mechanism that transforms inputs into the reference domain, thereby preserving structural correspondence with the pretrained models. Extensive numerical experiments in one- and two-dimensional settings demonstrate substantial improvements in training efficiency, generalization, and model transferability, highlighting the promise of initialization-based strategies for scalable and modular neural function approximation. The full code is made publicly available on Gitee. --- ### Guiding Exploration in Reinforcement Learning Through LLM-Augmented Observations Authors: Vaibhav Jain, Gerrit Grossmann Published: 2025-10-09 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.08779v1 Reinforcement Learning (RL) agents often struggle in sparse-reward environments where traditional exploration strategies fail to discover effective action sequences. Large Language Models (LLMs) possess procedural knowledge and reasoning capabilities from text pretraining that could guide RL exploration, but existing approaches create rigid dependencies where RL policies must follow LLM suggestions or incorporate them directly into reward functions. We propose a framework that provides LLM-generated action recommendations through augmented observation spaces, allowing RL agents to learn when to follow or ignore this guidance. Our method leverages LLMs' world knowledge and reasoning abilities while maintaining flexibility through soft constraints. We evaluate our approach on three BabyAI environments of increasing complexity and show that the benefits of LLM guidance scale with task difficulty. In the most challenging environment, we achieve 71% relative improvement in final success rates over baseline. The approach provides substantial sample efficiency gains, with agents reaching performance thresholds up to 9 times faster, and requires no modifications to existing RL algorithms. Our results demonstrate an effective method for leveraging LLM planning capabilities to accelerate RL training in challenging environments. --- ### Measuring Moral LLM Responses in Multilingual Capacities Authors: Kimaya Basu, Savi Kolari, Allison Yu Published: 2025-10-09 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2510.08776v1 With LLM usage becoming widespread across countries, languages, and humanity more broadly, the need to understand and guardrail their multilingual responses increases. Large-scale datasets for testing and benchmarking have been created to evaluate and facilitate LLM responses across multiple dimensions. In this study, we evaluate the responses of frontier and leading open-source models in five dimensions across low and high-resource languages to measure LLM accuracy and consistency across multilingual contexts. We evaluate the responses using a five-point grading rubric and a judge LLM. Our study shows that GPT-5 performed the best on average in each category, while other models displayed more inconsistency across language and category. Most notably, in the Consent & Autonomy and Harm Prevention & Safety categories, GPT scored the highest with averages of 3.56 and 4.73, while Gemini 2.5 Pro scored the lowest with averages of 1.39 and 1.98, respectively. These findings emphasize the need for further testing on how linguistic shifts impact LLM responses across various categories and improvement in these areas. --- ### Re-Identifying Kākā with AI-Automated Video Key Frame Extraction Authors: Paula Maddigan, Andrew Lensen, Rachael C. Shaw Published: 2025-10-09 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2510.08775v1 Accurate recognition and re-identification of individual animals is essential for successful wildlife population monitoring. Traditional methods, such as leg banding of birds, are time consuming and invasive. Recent progress in artificial intelligence, particularly computer vision, offers encouraging solutions for smart conservation and efficient automation. This study presents a unique pipeline for extracting high-quality key frames from videos of kākā (Nestor meridionalis), a threatened forest-dwelling parrot in New Zealand. Key frame extraction is well-studied in person re-identification, however, its application to wildlife is limited. Using video recordings at a custom-built feeder, we extract key frames and evaluate the re-identification performance of our pipeline. Our unsupervised methodology combines object detection using YOLO and Grounding DINO, optical flow blur detection, image encoding with DINOv2, and clustering methods to identify representative key frames. The results indicate that our proposed key frame selection methods yield image collections which achieve high accuracy in kākā re-identification, providing a foundation for future research using media collected in more diverse and challenging environments. Through the use of artificial intelligence and computer vision, our non-invasive and efficient approach provides a valuable alternative to traditional physical tagging methods for recognising kākā individuals and therefore improving the monitoring of populations. This research contributes to developing fresh approaches in wildlife monitoring, with applications in ecology and conservation biology. --- ### Struc-EMB: The Potential of Structure-Aware Encoding in Language Embeddings Authors: Shikun Liu, Haoyu Wang, Mufei Li, Pan Li Published: 2025-10-09 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2510.08774v1 Text embeddings from Large Language Models (LLMs) have become foundational for numerous applications. However, these models typically operate on raw text, overlooking the rich structural information, such as hyperlinks or citations, that provides crucial context in many real-world datasets. This paper introduces and systematically evaluates a new paradigm for generating structure-aware text embeddings by integrating these structural relations directly into the LLM's internal encoding process, rather than relying on traditional post-hoc aggregation. We investigate two primary in-process methods: sequential concatenation and parallel caching. Through extensive zero-shot experiments across retrieval, clustering, classification, and recommendation tasks, we demonstrate that our structure-aware approaches consistently outperform both text-only and post-hoc baselines. Our analysis reveals critical trade-offs: sequential concatenation excels with noisy, moderate-length contexts, while parallel caching scales more effectively to long, high-signal contexts but is more susceptible to distractors. To address the challenge of noisy structural data, we also introduce and validate two effective techniques: Context Distillation and Semantic Balancing. This work provides the first comprehensive analysis of in-process structure-aware encoding, offering a blueprint for building more powerful and contextually aware embedding models. --- ### LinearSR: Unlocking Linear Attention for Stable and Efficient Image Super-Resolution Authors: Xiaohui Li, Shaobin Zhuang, Shuo Cao, Yang Yang, Yuandong Pu, Qi Qin, Siqi Luo, Bin Fu, Yihao Liu Published: 2025-10-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.08771v4 Generative models for Image Super-Resolution (SR) are increasingly powerful, yet their reliance on self-attention's quadratic complexity (O(N^2)) creates a major computational bottleneck. Linear Attention offers an O(N) solution, but its promise for photorealistic SR has remained largely untapped, historically hindered by a cascade of interrelated and previously unsolved challenges. This paper introduces LinearSR, a holistic framework that, for the first time, systematically overcomes these critical hurdles. Specifically, we resolve a fundamental, training instability that causes catastrophic model divergence using our novel "knee point"-based Early-Stopping Guided Fine-tuning (ESGF) strategy. Furthermore, we mitigate the classic perception-distortion trade-off with a dedicated SNR-based Mixture of Experts (MoE) architecture. Finally, we establish an effective and lightweight guidance paradigm, TAG, derived from our "precision-over-volume" principle. Our resulting LinearSR model simultaneously delivers state-of-the-art perceptual quality with exceptional efficiency. Its core diffusion forward pass (1-NFE) achieves SOTA-level speed, while its overall multi-step inference time remains highly competitive. This work provides the first robust methodology for applying Linear Attention in the photorealistic SR domain, establishing a foundational paradigm for future research in efficient generative super-resolution. --- ### Detecting spills using thermal imaging, pretrained deep learning models, and a robotic platform Authors: Gregory Yeghiyan, Jurius Azar, Devson Butani, Chan-Jin Chung Published: 2025-10-09 Categories: cs.CV, cs.LG, cs.RO arXiv: https://arxiv.org/abs/2510.08770v1 This paper presents a real-time spill detection system that utilizes pretrained deep learning models with RGB and thermal imaging to classify spill vs. no-spill scenarios across varied environments. Using a balanced binary dataset (4,000 images), our experiments demonstrate the advantages of thermal imaging in inference speed, accuracy, and model size. We achieve up to 100% accuracy using lightweight models like VGG19 and NasNetMobile, with thermal models performing faster and more robustly across different lighting conditions. Our system runs on consumer-grade hardware (RTX 4080) and achieves inference times as low as 44 ms with model sizes under 350 MB, highlighting its deployability in safety-critical contexts. Results from experiments with a real robot and test datasets indicate that a VGG19 model trained on thermal imaging performs best. --- ### Prioritizing Latency with Profit: A DRL-Based Admission Control for 5G Network Slices Authors: Proggya Chakraborty, Aaquib Asrar, Jayasree Sengupta, Sipra Das Bit Published: 2025-10-09 Categories: cs.NI, cs.LG, cs.PF arXiv: https://arxiv.org/abs/2510.08769v1 5G networks enable diverse services such as eMBB, URLLC, and mMTC through network slicing, necessitating intelligent admission control and resource allocation to meet stringent QoS requirements while maximizing Network Service Provider (NSP) profits. However, existing Deep Reinforcement Learning (DRL) frameworks focus primarily on profit optimization without explicitly accounting for service delay, potentially leading to QoS violations for latency-sensitive slices. Moreover, commonly used epsilon-greedy exploration of DRL often results in unstable convergence and suboptimal policy learning. To address these gaps, we propose DePSAC -- a Delay and Profit-aware Slice Admission Control scheme. Our DRL-based approach incorporates a delay-aware reward function, where penalties due to service delay incentivize the prioritization of latency-critical slices such as URLLC. Additionally, we employ Boltzmann exploration to achieve smoother and faster convergence. We implement and evaluate DePSAC on a simulated 5G core network substrate with realistic Network Slice Request (NSLR) arrival patterns. Experimental results demonstrate that our method outperforms the DSARA baseline in terms of overall profit, reduced URLLC slice delays, improved acceptance rates, and improved resource consumption. These findings validate the effectiveness of the proposed DePSAC in achieving better QoS-profit trade-offs for practical 5G network slicing scenarios. --- ### Zero-Shot Policy Transfer in Reinforcement Learning using Buckingham's Pi Theorem Authors: Francisco Pascoa, Ian Lalonde, Alexandre Girard Published: 2025-10-09 Categories: cs.LG, cs.RO arXiv: https://arxiv.org/abs/2510.08768v1 Reinforcement learning (RL) policies often fail to generalize to new robots, tasks, or environments with different physical parameters, a challenge that limits their real-world applicability. This paper presents a simple, zero-shot transfer method based on Buckingham's Pi Theorem to address this limitation. The method adapts a pre-trained policy to new system contexts by scaling its inputs (observations) and outputs (actions) through a dimensionless space, requiring no retraining. The approach is evaluated against a naive transfer baseline across three environments of increasing complexity: a simulated pendulum, a physical pendulum for sim-to-real validation, and the high-dimensional HalfCheetah. Results demonstrate that the scaled transfer exhibits no loss of performance on dynamically similar contexts. Furthermore, on non-similar contexts, the scaled policy consistently outperforms the naive transfer, significantly expanding the volume of contexts where the original policy remains effective. These findings demonstrate that dimensional analysis provides a powerful and practical tool to enhance the robustness and generalization of RL policies. --- ### Understanding Exoplanet Habitability: A Bayesian ML Framework for Predicting Atmospheric Absorption Spectra Authors: Vasuda Trehan, Kevin H. Knuth, M. J. Way Published: 2025-10-09 Categories: astro-ph.EP, astro-ph.IM, cs.LG arXiv: https://arxiv.org/abs/2510.08766v1 The evolution of space technology in recent years, fueled by advancements in computing such as Artificial Intelligence (AI) and machine learning (ML), has profoundly transformed our capacity to explore the cosmos. Missions like the James Webb Space Telescope (JWST) have made information about distant objects more easily accessible, resulting in extensive amounts of valuable data. As part of this work-in-progress study, we are working to create an atmospheric absorption spectrum prediction model for exoplanets. The eventual model will be based on both collected observational spectra and synthetic spectral data generated by the ROCKE-3D general circulation model (GCM) developed by the climate modeling program at NASA's Goddard Institute for Space Studies (GISS). In this initial study, spline curves are used to describe the bin heights of simulated atmospheric absorption spectra as a function of one of the values of the planetary parameters. Bayesian Adaptive Exploration is then employed to identify areas of the planetary parameter space for which more data are needed to improve the model. The resulting system will be used as a forward model so that planetary parameters can be inferred given a planet's atmospheric absorption spectrum. This work is expected to contribute to a better understanding of exoplanetary properties and general exoplanet climates and habitability. --- ### Reinforcement Learning-Based Optimization of CT Acquisition and Reconstruction Parameters Through Virtual Imaging Trials Authors: David Fenwick, Navid NaderiAlizadeh, Vahid Tarokh, Nicholas Felice, Darin Clark, Jayasai Rajagopal, Anuj Kapadia, Benjamin Wildman-Tobriner, Ehsan Samei, Ehsan Abadi Published: 2025-10-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.08763v1 Protocol optimization is critical in Computed Tomography (CT) to achieve high diagnostic image quality while minimizing radiation dose. However, due to the complex interdependencies among CT acquisition and reconstruction parameters, traditional optimization methods rely on exhaustive testing of combinations of these parameters, which is often impractical. This study introduces a novel methodology that combines virtual imaging tools with reinforcement learning to optimize CT protocols more efficiently. Human models with liver lesions were imaged using a validated CT simulator and reconstructed with a novel CT reconstruction toolkit. The optimization parameter space included tube voltage, tube current, reconstruction kernel, slice thickness, and pixel size. The optimization process was performed using a Proximal Policy Optimization (PPO) agent, which was trained to maximize an image quality objective, specifically the detectability index (d') of liver lesions in the reconstructed images. Optimization performance was compared against an exhaustive search performed on a supercomputer. The proposed reinforcement learning approach achieved the global maximum d' across test cases while requiring 79.7% fewer steps than the exhaustive search, demonstrating both accuracy and computational efficiency. The proposed framework is flexible and can accommodate various image quality objectives. The findings highlight the potential of integrating virtual imaging tools with reinforcement learning for CT protocol management. --- ### Spatial Deconfounder: Interference-Aware Deconfounding for Spatial Causal Inference Authors: Ayush Khot, Miruna Oprescu, Maresa Schröder, Ai Kagawa, Xihaier Luo Published: 2025-10-09 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2510.08762v2 Causal inference in spatial domains faces two intertwined challenges: (1) unmeasured spatial factors, such as weather, air pollution, or mobility, that confound treatment and outcome, and (2) interference from nearby treatments that violate standard no-interference assumptions. While existing methods typically address one by assuming away the other, we show they are deeply connected: interference reveals structure in the latent confounder. Leveraging this insight, we propose the Spatial Deconfounder, a two-stage method that reconstructs a substitute confounder from local treatment vectors using a conditional variational autoencoder (C-VAE) with a spatial prior, then estimates causal effects with a flexible outcome model. We show that this enables nonparametric identification of direct and spillover effects under weak assumptions--without multiple treatment types or a known latent-field model. Empirically, we extend SpaCE, a benchmark suite for spatial confounding, to include treatment interference, and show that the Spatial Deconfounder consistently improves effect estimation across real-world environmental health and social science datasets. By turning local interference into a multi-cause proxy for latent spatial confounding, our framework advances robust causal inference for spatial data. --- ### SAFER-AiD: Saccade-Assisted Foveal-peripheral vision Enhanced Reconstruction for Adversarial Defense Authors: Jiayang Liu, Daniel Tso, Yiming Bu, Qinru Qiu Published: 2025-10-09 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2510.08761v1 Adversarial attacks significantly challenge the safe deployment of deep learning models, particularly in real-world applications. Traditional defenses often rely on computationally intensive optimization (e.g., adversarial training or data augmentation) to improve robustness, whereas the human visual system achieves inherent robustness to adversarial perturbations through evolved biological mechanisms. We hypothesize that attention guided non-homogeneous sparse sampling and predictive coding plays a key role in this robustness. To test this hypothesis, we propose a novel defense framework incorporating three key biological mechanisms: foveal-peripheral processing, saccadic eye movements, and cortical filling-in. Our approach employs reinforcement learning-guided saccades to selectively capture multiple foveal-peripheral glimpses, which are integrated into a reconstructed image before classification. This biologically inspired preprocessing effectively mitigates adversarial noise, preserves semantic integrity, and notably requires no retraining or fine-tuning of downstream classifiers, enabling seamless integration with existing systems. Experiments on the ImageNet dataset demonstrate that our method improves system robustness across diverse classifiers and attack types, while significantly reducing training overhead compared to both biologically and non-biologically inspired defense techniques. --- ### Dissecting Embodied Abilities in Multimodal Language Models through Skill-level Evaluation and Diagnosis Authors: Yu Qi, Haibo Zhao, Ziyu Guo, Siyuan Ma, Ziyan Chen, Yaokun Han, Renrui Zhang, Zitiantao Lin, Yizhe Zhu, Shiji Xin, Yijian Huang, Boce Hu et al. Published: 2025-10-09 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2510.08759v3 Understanding the capability bottlenecks of embodied multimodal large language models (MLLMs) is crucial for improving embodied agents. However, existing embodied benchmarks mainly focus on task-level evaluation and fail to provide actionable insights into the underlying causes of model failures. To address this limitation, we introduce BEAR, a benchmark that decomposes embodied tasks into 14 atomic skills for fine-grained skill-level evaluation. BEAR comprises 4,469 interleaved image-video-text samples spanning 14 skills across 6 categories, ranging from low-level perception to high-level planning. We evaluate 20 MLLMs on BEAR under a hierarchical skill-level diagnosis framework and uncover two key findings: (1) perceptual capabilities are major bottlenecks behind reasoning failures, and (2) current models suffer from unstable spatiotemporal modeling that remains largely unexposed in prior benchmarks. Motivated by these findings, we further propose BEAR-Agent, a multimodal conversational agent that augments MLLMs with visual and spatial reasoning tools. BEAR-Agent substantially improves performance across embodied skills, achieving a relative improvement of 17.5% on GPT-5 over the base model on BEAR, while also outperforming strong baselines in both simulation and real-world robotic experiments. Project page: https://bear-official66.github.io/ --- ### A Design-based Solution for Causal Inference with Text: Can a Language Model Be Too Large? Authors: Graham Tierney, Srikar Katta, Christopher Bail, Sunshine Hillygus, Alexander Volfovsky Published: 2025-10-09 Categories: stat.ME, cs.CL, cs.LG, stat.AP arXiv: https://arxiv.org/abs/2510.08758v1 Many social science questions ask how linguistic properties causally affect an audience's attitudes and behaviors. Because text properties are often interlinked (e.g., angry reviews use profane language), we must control for possible latent confounding to isolate causal effects. Recent literature proposes adapting large language models (LLMs) to learn latent representations of text that successfully predict both treatment and the outcome. However, because the treatment is a component of the text, these deep learning methods risk learning representations that actually encode the treatment itself, inducing overlap bias. Rather than depending on post-hoc adjustments, we introduce a new experimental design that handles latent confounding, avoids the overlap issue, and unbiasedly estimates treatment effects. We apply this design in an experiment evaluating the persuasiveness of expressing humility in political communication. Methodologically, we demonstrate that LLM-based methods perform worse than even simple bag-of-words models using our real text and outcomes from our experiment. Substantively, we isolate the causal effect of expressing humility on the perceived persuasiveness of political statements, offering new insights on communication effects for social media platforms, policy makers, and social scientists. --- ### LOTION: Smoothing the Optimization Landscape for Quantized Training Authors: Mujin Kwun, Depen Morwani, Chloe Huangyuan Su, Stephanie Gil, Nikhil Anand, Sham Kakade Published: 2025-10-09 Categories: cs.LG, cs.AR arXiv: https://arxiv.org/abs/2510.08757v1 Optimizing neural networks for quantized objectives is fundamentally challenging because the quantizer is piece-wise constant, yielding zero gradients everywhere except at quantization thresholds where the derivative is undefined. Most existing methods deal with this issue by relaxing gradient computations with techniques like Straight Through Estimators (STE) and do not provide any guarantees of convergence. In this work, taking inspiration from Nesterov smoothing, we approximate the quantized loss surface with a continuous loss surface. In particular, we introduce LOTION, \textbf{L}ow-precision \textbf{O}ptimization via s\textbf{T}ochastic-no\textbf{I}se sm\textbf{O}othi\textbf{N}g, a principled smoothing framework that replaces the raw quantized loss with its expectation under unbiased randomized-rounding noise. In this framework, standard optimizers are guaranteed to converge to a local minimum of the loss surface. Moreover, when using noise derived from stochastic rounding, we show that the global minima of the original quantized loss are preserved. We empirically demonstrate that this method outperforms standard QAT on synthetic testbeds and on 150M- and 300M- parameter language models. --- ### Robust Heuristic Algorithm Design with LLMs Authors: Pantea Karimi, Dany Rouhana, Pooria Namyar, Siva Kesava Reddy Kakarla, Venkat Arun, Behnaz Arzani Published: 2025-10-09 Categories: cs.AI, cs.CL, cs.NI arXiv: https://arxiv.org/abs/2510.08755v1 We posit that we can generate more robust and performant heuristics if we augment approaches using LLMs for heuristic design with tools that explain why heuristics underperform and suggestions about how to fix them. We find even simple ideas that (1) expose the LLM to instances where the heuristic underperforms; (2) explain why they occur; and (3) specialize design to regions in the input space, can produce more robust algorithms compared to existing techniques~ -- ~the heuristics we produce have a $\sim28\times$ better worst-case performance compared to FunSearch, improve average performance, and maintain the runtime. --- ### Exploring Cross-Client Memorization of Training Data in Large Language Models for Federated Learning Authors: Tinnakit Udsa, Can Udomcharoenchaikit, Patomporn Payoungkhamdee, Sarana Nutanong, Norrathep Rattanavipanon Published: 2025-10-09 Categories: cs.LG, cs.CL arXiv: https://arxiv.org/abs/2510.08750v2 Federated learning (FL) enables collaborative training without raw data sharing, but still risks training data memorization. Existing FL memorization detection techniques focus on one sample at a time, underestimating more subtle risks of cross-sample memorization. In contrast, recent work on centralized learning (CL) has introduced fine-grained methods to assess memorization across all samples in training data, but these assume centralized access to data and cannot be applied directly to FL. We bridge this gap by proposing a framework that quantifies both intra- and inter-client memorization in FL using fine-grained cross-sample memorization measurement across all clients. Based on this framework, we conduct two studies: (1) measuring subtle memorization across clients and (2) examining key factors that influence memorization, including decoding strategies, prefix length, and FL algorithms. Our findings reveal that FL models do memorize client data, particularly intra-client data, more than inter-client data, with memorization influenced by training and inferencing factors. --- ### Theoretical guarantees for change localization using conformal p-values Authors: Swapnaneel Bhattacharyya, Aaditya Ramdas Published: 2025-10-09 Categories: math.ST, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2510.08749v2 Changepoint localization aims to provide confidence sets for a changepoint (if one exists). Existing methods either relying on strong parametric assumptions or providing only asymptotic guarantees or focusing on a particular kind of change(e.g., change in the mean) rather than the entire distributional change. A method (possibly the first) to achieve distribution-free changepoint localization with finite-sample validity was recently introduced by \cite{dandapanthula2025conformal}. However, while they proved finite sample coverage, there was no analysis of set size. In this work, we provide rigorous theoretical guarantees for their algorithm. We also show the consistency of a point estimator for change, and derive its convergence rate without distributional assumptions. Along that line, we also construct a distribution-free consistent test to assess whether a particular time point is a changepoint or not. Thus, our work provides unified distribution-free guarantees for changepoint detection, localization, and testing. In addition, we present various finite sample and asymptotic properties of the conformal $p$-value in the distribution change setup, which provides a theoretical foundation for many applications of the conformal $p$-value. As an application of these properties, we construct distribution-free consistent tests for exchangeability against distribution-change alternatives and a new, computationally tractable method of optimizing the powers of conformal tests. We run detailed simulation studies to corroborate the performance of our methods and theoretical results. Together, our contributions offer a comprehensive and theoretically principled approach to distribution-free changepoint inference, broadening both the scope and credibility of conformal methods in modern changepoint analysis. --- ### Conformal Risk Training: End-to-End Optimization of Conformal Risk Control Authors: Christopher Yeh, Nicolas Christianson, Adam Wierman, Yisong Yue Published: 2025-10-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.08748v1 While deep learning models often achieve high predictive accuracy, their predictions typically do not come with any provable guarantees on risk or reliability, which are critical for deployment in high-stakes applications. The framework of conformal risk control (CRC) provides a distribution-free, finite-sample method for controlling the expected value of any bounded monotone loss function and can be conveniently applied post-hoc to any pre-trained deep learning model. However, many real-world applications are sensitive to tail risks, as opposed to just expected loss. In this work, we develop a method for controlling the general class of Optimized Certainty-Equivalent (OCE) risks, a broad class of risk measures which includes as special cases the expected loss (generalizing the original CRC method) and common tail risks like the conditional value-at-risk (CVaR). Furthermore, standard post-hoc CRC can degrade average-case performance due to its lack of feedback to the model. To address this, we introduce "conformal risk training," an end-to-end approach that differentiates through conformal OCE risk control during model training or fine-tuning. Our method achieves provable risk guarantees while demonstrating significantly improved average-case performance over post-hoc approaches on applications to controlling classifiers' false negative rate and controlling financial risk in battery storage operation. --- ### RFOD: Random Forest-based Outlier Detection for Tabular Data Authors: Yihao Ang, Peicheng Yao, Yifan Bao, Yushuo Feng, Qiang Huang, Anthony K. H. Tung, Zhiyong Huang Published: 2025-10-09 Categories: cs.LG, cs.DB arXiv: https://arxiv.org/abs/2510.08747v1 Outlier detection in tabular data is crucial for safeguarding data integrity in high-stakes domains such as cybersecurity, financial fraud detection, and healthcare, where anomalies can cause serious operational and economic impacts. Despite advances in both data mining and deep learning, many existing methods struggle with mixed-type tabular data, often relying on encoding schemes that lose important semantic information. Moreover, they frequently lack interpretability, offering little insight into which specific values cause anomalies. To overcome these challenges, we introduce \textsf{\textbf{RFOD}}, a novel \textsf{\textbf{R}}andom \textsf{\textbf{F}}orest-based \textsf{\textbf{O}}utlier \textsf{\textbf{D}}etection framework tailored for tabular data. Rather than modeling a global joint distribution, \textsf{RFOD} reframes anomaly detection as a feature-wise conditional reconstruction problem, training dedicated random forests for each feature conditioned on the others. This design robustly handles heterogeneous data types while preserving the semantic integrity of categorical features. To further enable precise and interpretable detection, \textsf{RFOD} combines Adjusted Gower's Distance (AGD) for cell-level scoring, which adapts to skewed numerical data and accounts for categorical confidence, with Uncertainty-Weighted Averaging (UWA) to aggregate cell-level scores into robust row-level anomaly scores. Extensive experiments on 15 real-world datasets demonstrate that \textsf{RFOD} consistently outperforms state-of-the-art baselines in detection accuracy while offering superior robustness, scalability, and interpretability for mixed-type tabular data. --- ### Graph Diffusion Transformers are In-Context Molecular Designers Authors: Gang Liu, Jie Chen, Yihan Zhu, Michael Sun, Tengfei Luo, Nitesh V Chawla, Meng Jiang Published: 2025-10-09 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.08744v1 In-context learning allows large models to adapt to new tasks from a few demonstrations, but it has shown limited success in molecular design. Existing databases such as ChEMBL contain molecular properties spanning millions of biological assays, yet labeled data for each property remain scarce. To address this limitation, we introduce demonstration-conditioned diffusion models (DemoDiff), which define task contexts using a small set of molecule-score examples instead of text descriptions. These demonstrations guide a denoising Transformer to generate molecules aligned with target properties. For scalable pretraining, we develop a new molecular tokenizer with Node Pair Encoding that represents molecules at the motif level, requiring 5.5$\times$ fewer nodes. We curate a dataset containing millions of context tasks from multiple sources covering both drugs and materials, and pretrain a 0.7-billion-parameter model on it. Across 33 design tasks in six categories, DemoDiff matches or surpasses language models 100-1000$\times$ larger and achieves an average rank of 3.63 compared to 5.25-10.20 for domain-specific approaches. These results position DemoDiff as a molecular foundation model for in-context molecular design. Our code is available at https://github.com/liugangcode/DemoDiff. --- ### Coordinates from Context: Using LLMs to Ground Complex Location References Authors: Tessa Masis, Brendan O'Connor Published: 2025-10-09 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2510.08741v2 Geocoding is the task of linking a location reference to an actual geographic location and is essential for many downstream analyses of unstructured text. In this paper, we explore the challenging setting of geocoding compositional location references. Building on recent work demonstrating LLMs' abilities to reason over geospatial data, we evaluate LLMs' geospatial knowledge versus reasoning skills relevant to our task. Based on these insights, we propose an LLM-based strategy for geocoding compositional location references. We show that our approach improves performance for the task and that a relatively small fine-tuned LLM can achieve comparable performance with much larger off-the-shelf models. --- ### Faithful and Interpretable Explanations for Complex Ensemble Time Series Forecasts using Surrogate Models and Forecastability Analysis Authors: Yikai Zhao, Jiekai Ma Published: 2025-10-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.08739v1 Modern time series forecasting increasingly relies on complex ensemble models generated by AutoML systems like AutoGluon, delivering superior accuracy but with significant costs to transparency and interpretability. This paper introduces a comprehensive, dual-approach framework that addresses both the explainability and forecastability challenges in complex time series ensembles. First, we develop a surrogate-based explanation methodology that bridges the accuracy-interpretability gap by training a LightGBM model to faithfully mimic AutoGluon's time series forecasts, enabling stable SHAP-based feature attributions. We rigorously validated this approach through feature injection experiments, demonstrating remarkably high faithfulness between extracted SHAP values and known ground truth effects. Second, we integrated spectral predictability analysis to quantify each series' inherent forecastability. By comparing each time series' spectral predictability to its pure noise benchmarks, we established an objective mechanism to gauge confidence in forecasts and their explanations. Our empirical evaluation on the M5 dataset found that higher spectral predictability strongly correlates not only with improved forecast accuracy but also with higher fidelity between the surrogate and the original forecasting model. These forecastability metrics serve as effective filtering mechanisms and confidence scores, enabling users to calibrate their trust in both the forecasts and their explanations. We further demonstrated that per-item normalization is essential for generating meaningful SHAP explanations across heterogeneous time series with varying scales. The resulting framework delivers interpretable, instance-level explanations for state-of-the-art ensemble forecasts, while equipping users with forecastability metrics that serve as reliability indicators for both predictions and their explanations. --- ### SHAP-Based Supervised Clustering for Sample Classification and the Generalized Waterfall Plot Authors: Justin Lin, Julia Fukuyama Published: 2025-10-09 Categories: cs.LG, stat.ME, stat.ML arXiv: https://arxiv.org/abs/2510.08737v1 In this growing age of data and technology, large black-box models are becoming the norm due to their ability to handle vast amounts of data and learn incredibly complex input-output relationships. The deficiency of these methods, however, is their inability to explain the prediction process, making them untrustworthy and their use precarious in high-stakes situations. SHapley Additive exPlanations (SHAP) analysis is an explainable AI method growing in popularity for its ability to explain model predictions in terms of the original features. For each sample and feature in the data set, we associate a SHAP value that quantifies the contribution of that feature to the prediction of that sample. Clustering these SHAP values can provide insight into the data by grouping samples that not only received the same prediction, but received the same prediction for similar reasons. In doing so, we map the various pathways through which distinct samples arrive at the same prediction. To showcase this methodology, we present a simulated experiment in addition to a case study in Alzheimer's disease using data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. We also present a novel generalization of the waterfall plot for multi-classification. --- ### Transmuting prompts into weights Authors: Hanna Mazzawi, Benoit Dherin, Michael Munn, Adrian Goldwaser, Michael Wunder, Javier Gonzalvo Published: 2025-10-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.08734v3 A growing body of research has demonstrated that the behavior of large language models can be effectively controlled at inference time by directly modifying their internal states, either through vector additions to their activations or through updates to their weight matrices. These techniques, while powerful, are often guided by empirical heuristics, such as deriving ``steering vectors'' from the average activations of contrastive prompts. Building on the foundational work of Dherin et al. (2025), who discovered that a prompt's influence mathematically maps to token-dependent implicit weight updates and introduced the initial concept of a static thought patch for prompt compression, we elevate this framework into a robust algorithm for direct model editing. We derive a principled method for condensing this transient information into token-independent thought vectors and thought matrices. These constructs provide a theoretical explanation for existing vector-and-matrix-based model editing techniques and offer a direct, computationally-grounded method for transmuting textual input into reusable weight updates for complex architectures and new knowledge injection. --- ### When to Reason: Semantic Router for vLLM Authors: Chen Wang, Xunzhuo Liu, Yuhan Liu, Yue Zhu, Xiangxi Mo, Junchen Jiang, Huamin Chen Published: 2025-10-09 Categories: cs.ET, cs.AI, cs.CL, eess.SY arXiv: https://arxiv.org/abs/2510.08731v1 Large Language Models (LLMs) demonstrate substantial accuracy gains when augmented with reasoning modes such as chain-of-thought and inference-time scaling. However, reasoning also incurs significant costs in inference latency and token usage, with environmental and financial impacts, which are unnecessary for many simple prompts. We present a semantic router that classifies queries based on their reasoning requirements and selectively applies reasoning only when beneficial. Our approach achieves a 10.2 percentage point improvement in accuracy on the MMLU-Pro benchmark while reducing response latency by 47.1% and token consumption by 48.5% compared to direct inference with vLLM. These results demonstrate that semantic routing offers an effective mechanism for striking a balance between accuracy and efficiency in open-source LLM serving systems --- ### How Reliable is Language Model Micro-Benchmarking? Authors: Gregory Yauney, Shahzaib Saqib Warraich, Swabha Swayamdipta Published: 2025-10-09 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2510.08730v2 Micro-benchmarking offers a solution to the often prohibitive time and cost of language model development: evaluate on a very small subset of existing benchmarks. Can these micro-benchmarks, however, rank models as consistently as the full benchmarks they replace? And can they rank models more consistently than selecting a random subset of data points? In many scenarios, we find that the answer is no. We introduce a meta-evaluation measure for micro-benchmarking which investigates how well a micro-benchmark can rank two models as a function of their performance difference on the full benchmark. This approach can determine which model pairs can be ranked correctly by a micro-benchmark, allowing for a finer-grained analysis of the trade-off between micro-benchmark size and reliability. Prior work has suggested selecting as few as 10 examples; we find that no micro-benchmarking method can consistently rank model pairs 3.5 points of accuracy apart on MMLU-Pro or 4 points apart on BIG-bench Hard. In order to consistently rank model pairs with relatively similar performances, we show that often as many as 250 examples must be selected, at which point random sampling is competitive with existing micro-benchmarking methods. When comparing only 8B instruction-tuned models on MMLU-Pro micro-benchmarks with 25 examples, we find that more than half of pairwise comparisons are not likely to be preserved. Our work provides actionable guidance for both micro-benchmark users and developers in navigating the trade-off between evaluation efficiency and reliability. --- ### Structured Output Regularization: a framework for few-shot transfer learning Authors: Nicolas Ewen, Jairo Diaz-Rodriguez, Kelly Ramsay Published: 2025-10-09 Categories: cs.CV, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2510.08728v1 Traditional transfer learning typically reuses large pre-trained networks by freezing some of their weights and adding task-specific layers. While this approach is computationally efficient, it limits the model's ability to adapt to domain-specific features and can still lead to overfitting with very limited data. To address these limitations, we propose Structured Output Regularization (SOR), a simple yet effective framework that freezes the internal network structures (e.g., convolutional filters) while using a combination of group lasso and $L_1$ penalties. This framework tailors the model to specific data with minimal additional parameters and is easily applicable to various network components, such as convolutional filters or various blocks in neural networks enabling broad applicability for transfer learning tasks. We evaluate SOR on three few shot medical imaging classification tasks and we achieve competitive results using DenseNet121, and EfficientNetB4 bases compared to established benchmarks. --- ### Neptune: Advanced ML Operator Fusion for Locality and Parallelism on GPUs Authors: Yifan Zhao, Egan Johnson, Prasanth Chatarasi, Vikram Adve, Sasa Misailovic Published: 2025-10-09 Categories: cs.PL, cs.LG arXiv: https://arxiv.org/abs/2510.08726v2 Operator fusion has become a key optimization for deep learning, which combines multiple deep learning operators to improve data reuse and reduce global memory transfers. However, existing tensor compilers struggle to fuse complex reduction computations involving loop-carried dependencies, such as attention mechanisms. This paper introduces Neptune, a tensor compiler for advanced operator fusion for sequences of reduction operators. Neptune presents a new approach for advanced operator fusion, which intentionally breaks some existing dependencies and compensates by constructing algebraic correction expressions that allow the kernel to produce the correct result. Applying Neptune's advanced operator fusion to a plain attention operator generates operators equivalent to FlashAttention and FlashDecoding. On ten attention-based benchmarks, Neptune, starting from a plain attention code and a high-level scheduling template, outperforms existing compilers like Triton, TVM, and FlexAttention, including Triton-based implementations of FlashAttention. Across four different GPU architectures from NVIDIA and AMD, Neptune-generated kernels have an average speedup of $1.35\times$ over the next best alternative, with up to $2.65\times$ speedup on Nvidia GPUs and up to $3.32\times$ on AMD GPUs, demonstrating its effectiveness for deep learning workloads. --- ### Counterfactually Fair Conformal Prediction Authors: Ozgur Guldogan, Neeraj Sarna, Yuanyuan Li, Michael Berger Published: 2025-10-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.08724v2 While counterfactual fairness of point predictors is well studied, its extension to prediction sets--central to fair decision-making under uncertainty--remains underexplored. On the other hand, conformal prediction (CP) provides efficient, distribution-free, finite-sample valid prediction sets, yet does not ensure counterfactual fairness. We close this gap by developing Counterfactually Fair Conformal Prediction (CF-CP) that produces counterfactually fair prediction sets. Through symmetrization of conformity scores across protected-attribute interventions, we prove that CF-CP results in counterfactually fair prediction sets while maintaining the marginal coverage property. Furthermore, we empirically demonstrate that on both synthetic and real datasets, across regression and classification tasks, CF-CP achieves the desired counterfactual fairness and meets the target coverage rate with minimal increase in prediction set size. CF-CP offers a simple, training-free route to counterfactually fair uncertainty quantification. --- ### The Impact of Semantic Pairs on Self-Supervised Representation Learning Authors: Mohammad Alkhalefi, Georgios Leontidis, Mingjun Zhong Published: 2025-10-09 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.08722v4 Instance discrimination learns visual representations by treating different augmented views of the same image as positive pairs. While this encourages invariance to handcrafted transformations, same-image positives can preserve nuisance correlations such as background, texture, illumination, and object-specific details. Semantic positive pairs, i.e., different same-class instances, may reduce these correlations by presenting objects across diverse contexts. However, previous studies often combine semantic pairs with augmented positives or false neighbors (i.e., incorrectly mapped semantic pairs), making it difficult to isolate the effect of semantic pairing. We present a controlled empirical study of semantic positive pairs for self-supervised representation learning. From ImageNet-1K, we construct two matched subsets: an augmented-pair baseline and a manually curated semantic-pair dataset with the same class composition and training-pair count. We use these datasets to compare representative contrastive and non-contrastive SSL methods under matched training conditions. Across transfer learning and object detection evaluations, semantic-pair pretraining consistently improves generalisation over augmented-pair pretraining. Additional ablations show that semantic pairs induce invariances beyond the standard transformation pipeline. Among the evaluated methods, contrastive learning benefits most strongly from semantic pairs, with SimCLR showing the largest relative improvement. These results clarify the role of semantic positive pairs in SSL and provide guidance for selecting and designing frameworks that can exploit semantic pair information effectively. --- ### How Many Code and Test Cases Are Enough? Evaluating Test Cases Generation from a Binary-Matrix Perspective Authors: Xianzhen Luo, Jinyang Huang, Wenzhen Zheng, Qingfu Zhu, Mingzheng Xu, Yiheng Xu, Yuantao Fan, Wanxiang Che Published: 2025-10-09 Categories: cs.CL arXiv: https://arxiv.org/abs/2510.08720v3 Evaluating test cases automatically generated by Large Language Models (LLMs) is a critical yet challenging task. Existing benchmarks often evaluate the exclusion ratio on large, unstructured collections of wrong codes, suffering from high computational costs and score inflation. Furthermore, they inadvertently reward generators that detect common, trivial bugs, while failing to penalize their inability to identify rare yet critical faults. In this work, we connect two fundamental questions: (1) What is the minimal set of wrong codes sufficient to represent the entire error space? and (2) What is the minimal set of test cases needed to distinguish them? We introduce a novel framework that formalizes benchmark construction as finding an optimal diagnostic basis in a binary code-test matrix, where rows represent wrong codes and columns represent test case results. The rank of this matrix specifies the minimal number of independent error patterns (wrong codes) and provides a tight upper bound on the number of test cases required for complete fault coverage. Our objective is to identify a basis of size equal to the matrix rank that maximizes internal diversity. To tackle this NP-hard problem, we propose WrongSelect, an efficient approximation algorithm to select maximally diverse wrong codes. Applying this framework to millions of competitive programming submissions, we construct TC-Bench, a compact, diverse, and inflation-resistant benchmark. Extensive experiments show that even the most advanced test case generation methods achieve only ~60% exclusion rates on TC-Bench, exposing a significant gap in their diagnostic power and highlighting substantial room for future improvement. Our dataset is available at: https://huggingface.co/datasets/Luoberta/TC-Bench and our code is at: https://github.com/Luowaterbi/TC-Bench. --- ### Towards Unified World Models for Visual Navigation via Memory-Augmented Planning and Foresight Authors: Yifei Dong, Fengyi Wu, Guangyu Chen, Lingdong Kong, Qiyu Hu, Yuxuan Zhou, Xu Zhu, Jingdong Sun, Jun-Yan He, Qi Dai, Alexander G. Hauptmann, Zhi-Qi Cheng Published: 2025-10-09 Categories: cs.AI, cs.CV, cs.RO arXiv: https://arxiv.org/abs/2510.08713v3 Enabling embodied agents to imagine future states is essential for robust and generalizable visual navigation. Yet, state-of-the-art systems typically rely on modular designs that decouple navigation planning from visual world modeling, which often induces state-action misalignment and weak adaptability in novel or dynamic scenarios. We propose UniWM, a unified, memory-augmented world model that integrates egocentric visual foresight and planning within a single multimodal autoregressive backbone. UniWM explicitly grounds action selection in visually imagined outcomes, tightly aligning prediction with control. Meanwhile, a hierarchical memory mechanism fuses short-term perceptual cues with longer-term trajectory context, supporting stable and coherent reasoning over extended horizons. Extensive experiments on four challenging benchmarks (Go Stanford, ReCon, SCAND, HuRoN) and the 1X Humanoid Dataset show that UniWM improves navigation success rates by up to 30%, substantially reduces trajectory errors against strong baselines, generalizes zero-shot to the unseen TartanDrive dataset, and scales naturally to high-dimensional humanoid navigation. These results position UniWM as a principled step toward unified, imagination-driven embodied navigation. The code and models are available at https://github.com/UWMILab/UniWM. --- ### In-Context Learning for Non-Stationary MIMO Equalization Authors: Jiachen Jiang, Zhen Qin, Zhihui Zhu Published: 2025-10-09 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2510.08711v1 Channel equalization is fundamental for mitigating distortions such as frequency-selective fading and inter-symbol interference. Unlike standard supervised learning approaches that require costly retraining or fine-tuning for each new task, in-context learning (ICL) adapts to new channels at inference time with only a few examples. However, existing ICL-based equalizers are primarily developed for and evaluated on static channels within the context window. Indeed, to our knowledge, prior principled analyses and theoretical studies of ICL focus exclusively on the stationary setting, where the function remains fixed within the context. In this paper, we investigate the ability of ICL to address non-stationary problems through the lens of time-varying channel equalization. We employ a principled framework for designing efficient attention mechanisms with improved adaptivity in non-stationary tasks, leveraging algorithms from adaptive signal processing to guide better designs. For example, new attention variants can be derived from the Least Mean Square (LMS) adaptive algorithm, a Least Root Mean Square (LRMS) formulation for enhanced robustness, or multi-step gradient updates for improved long-term tracking. Experimental results demonstrate that ICL holds strong promise for non-stationary MIMO equalization, and that attention mechanisms inspired by classical adaptive algorithms can substantially enhance adaptability and performance in dynamic environments. Our findings may provide critical insights for developing next-generation wireless foundation models with stronger adaptability and robustness. --- ### Thinking Longer, Not Always Smarter: Evaluating LLM Capabilities in Hierarchical Legal Reasoning Authors: Li Zhang, Matthias Grabmair, Morgan Gray, Kevin Ashley Published: 2025-10-09 Categories: cs.CL arXiv: https://arxiv.org/abs/2510.08710v2 Case-based reasoning is a cornerstone of U.S. legal practice, requiring professionals to argue about a current case by drawing analogies to and distinguishing from past precedents. While Large Language Models (LLMs) have shown remarkable capabilities, their proficiency in this complex, nuanced form of reasoning needs further investigation. We propose a formal framework that decomposes the process of identifying significant distinctions between cases into three-stage reasoning tasks. Our framework models cases using factual predicates called factors, organizes them into a legal knowledge hierarchy, and defines verifiable rules for identifying distinctions, analyzing their argumentative support, and evaluating their significance. Through comprehensive evaluation of modern reasoning LLMs, we reveal a paradox: while models achieve high accuracy on surface-level reasoning (Task 1), performance degrades on hierarchical reasoning (Task 2: 64.82%-92.09%) and collapses on integrated analysis (Task 3: 11.46%-33.99%). Most strikingly, we find that models consistently expend more computational resources on incorrect responses than correct ones, suggesting that "thinking longer" does not always mean "thinking smarter." Our work provides a methodology for fine-grained analysis of LLM reasoning capabilities in complex domains and reveals fundamental limitations that must be addressed for robust and trustworthy legal AI. --- ### ConPoSe: LLM-Guided Contact Point Selection for Scalable Cooperative Object Pushing Authors: Noah Steinkrüger, Nisarga Nilavadi, Wolfram Burgard, Tanja Katharina Kaiser Published: 2025-10-09 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2510.08705v1 Object transportation in cluttered environments is a fundamental task in various domains, including domestic service and warehouse logistics. In cooperative object transport, multiple robots must coordinate to move objects that are too large for a single robot. One transport strategy is pushing, which only requires simple robots. However, careful selection of robot-object contact points is necessary to push the object along a preplanned path. Although this selection can be solved analytically, the solution space grows combinatorially with the number of robots and object size, limiting scalability. Inspired by how humans rely on common-sense reasoning for cooperative transport, we propose combining the reasoning capabilities of Large Language Models with local search to select suitable contact points. Our LLM-guided local search method for contact point selection, ConPoSe, successfully selects contact points for a variety of shapes, including cuboids, cylinders, and T-shapes. We demonstrate that ConPoSe scales better with the number of robots and object size than the analytical approach, and also outperforms pure LLM-based selection. --- ### Decoding Positive Selection in Mycobacterium tuberculosis with Phylogeny-Guided Graph Attention Models Authors: Linfeng Wang, Susana Campino, Taane G. Clark, Jody E. Phelan Published: 2025-10-09 Categories: q-bio.PE, cs.LG arXiv: https://arxiv.org/abs/2510.08703v1 Positive selection drives the emergence of adaptive mutations in Mycobacterium tuberculosis, shaping drug resistance, transmissibility, and virulence. Phylogenetic trees capture evolutionary relationships among isolates and provide a natural framework for detecting such adaptive signals. We present a phylogeny-guided graph attention network (GAT) approach, introducing a method for converting SNP-annotated phylogenetic trees into graph structures suitable for neural network analysis. Using 500 M. tuberculosis isolates from four major lineages and 249 single-nucleotide variants (84 resistance-associated and 165 neutral) across 61 drug-resistance genes, we constructed graphs where nodes represented isolates and edges reflected phylogenetic distances. Edges between isolates separated by more than seven internal nodes were pruned to emphasise local evolutionary structure. Node features encoded SNP presence or absence, and the GAT architecture included two attention layers, a residual connection, global attention pooling, and a multilayer perceptron classifier. The model achieved an accuracy of 0.88 on a held-out test set and, when applied to 146 WHO-classified "uncertain" variants, identified 41 candidates with convergent emergence across multiple lineages, consistent with adaptive evolution. This work demonstrates the feasibility of transforming phylogenies into GNN-compatible structures and highlights attention-based models as effective tools for detecting positive selection, aiding genomic surveillance and variant prioritisation. --- ### Scaling Laws for Code: A More Data-Hungry Regime Authors: Xianzhen Luo, Wenzhen Zheng, Qingfu Zhu, Rongyi Zhang, Houyi Li, Siming Huang, YuanTao Fan, Wanxiang Che Published: 2025-10-09 Categories: cs.CL arXiv: https://arxiv.org/abs/2510.08702v2 Code Large Language Models (LLMs) are revolutionizing software engineering. However, scaling laws that guide the efficient training are predominantly analyzed on Natural Language (NL). Given the fundamental differences like strict syntax between code and NL, it is unclear whether these laws are directly applicable to code. To address this gap, we conduct the first large-scale empirical study of scaling laws for code, comprising 117 experimental runs with model sizes from 0.2B to 3.8B and training tokens from 2B to 128B. We fit the Chinchilla law and the Farsser law. First, the results show that the more expressive Farseer law offers greater accuracy. Second, the analysis reveals that Code LLMs scale effectively with model size. Crucially, code represents a more data-hungry regime, requiring a substantially higher data-to-parameter ratio than NL. Finally, two additional sets of experiments on code-NL mixtures show that NL benefits resource-constrained scenarios, but becomes a detriment at higher compute budgets. --- ### BigCodeArena: Unveiling More Reliable Human Preferences in Code Generation via Execution Authors: Terry Yue Zhuo, Xiaolong Jin, Hange Liu, Juyong Jiang, Tianyang Liu, Chen Gong, Bhupesh Bishnoi, Vaisakhi Mishra, Marek Suppa, Noah Ziems, Saiteja Utpala, Ming Xu et al. Published: 2025-10-09 Categories: cs.SE, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2510.08697v2 Crowdsourced model evaluation platforms, such as Chatbot Arena, enable real-time evaluation from human perspectives to assess the quality of model responses. In the coding domain, manually examining the quality of LLM-generated content is extremely challenging, as it requires understanding long chunks of raw code and deliberately simulating code execution. To this end, we introduce BigCodeArena, an open human evaluation platform for code generation backed by a comprehensive and on-the-fly execution environment. Built on top of Chatbot Arena, BigCodeArena enables the execution of LLM-generated code and allows humans to interact with the execution process and outcomes. We collected over 14,000 raw code-centric conversation sessions across 10 widely used LLMs, spanning 10 languages and 8 types of execution environments. Among these conversations, we identified more than 4,700 multi-turn samples with pairwise human preferences. Further analysis uncovers underexplored preferences of LLMs in fine-grained domains characterized by tasks, languages, and frameworks. To systematically examine code understanding and generation capabilities of frontier LLMs, we curated two benchmarks based on the collected data, namely BigCodeReward and AutoCodeArena. For BigCodeReward, we post-processed the 4,700 conversations and evaluated the consistency between reward models and human preferences. The evaluation shows that most LLMs have superior performance in judging coding preferences when the execution results are available. Inspired by these findings, we propose AutoCodeArena, an automatic Elo rating benchmark designed to assess the coding quality of LLMs without human involvement. We find that proprietary LLMs like GPT-5, Claude-Sonnet-4, and Claude-Opus-4 still lead in code generation performance among recent emerging models. --- ### Don't Waste Mistakes: Leveraging Negative RL-Groups via Confidence Reweighting Authors: Yunzhen Feng, Parag Jain, Anthony Hartshorn, Yaqi Duan, Julia Kempe Published: 2025-10-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2510.08696v1 Reinforcement learning with verifiable rewards (RLVR) has become a standard recipe for improving large language models (LLMs) on reasoning tasks, with Group Relative Policy Optimization (GRPO) widely used in practice. Yet GRPO wastes substantial compute on negative groups: groups in which no sampled response is correct yield zero advantage and thus no gradient. We ask whether negative groups can be leveraged without extra supervision. Starting from a maximum-likelihood (MLE) objective in reward modeling, we show that the MLE gradient is equivalent to a policy gradient for a modified value function. This value function adds a confidence-weighted penalty on incorrect responses, imposing larger penalties on more confident mistakes. We refer to this as \textbf{L}ikelihood \textbf{E}stimation with \textbf{N}egative \textbf{S}amples (\textbf{LENS}). LENS modifies GRPO to assign non-zero, confidence-dependent rewards to incorrect generations, making negative groups informative and converting previously wasted samples into useful gradient updates. On the MATH benchmark with Llama-3.1-8B and Qwen-2.5-3B, the proposed variant consistently outperforms GRPO baseline, with significant gains on harder items. These results demonstrate a principled and practical way to "rescue" negative groups, improving efficiency and performance in RLVR. --- ### Thinking with Camera: A Unified Multimodal Model for Camera-Centric Understanding and Generation Authors: Kang Liao, Size Wu, Zhonghua Wu, Linyi Jin, Chao Wang, Yikai Wang, Fei Wang, Wei Li, Chen Change Loy Published: 2025-10-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.08673v2 Camera-centric understanding and generation are two cornerstones of spatial intelligence, yet they are typically studied in isolation. We present Puffin, a unified camera-centric multimodal model that extends spatial awareness along the camera dimension. Puffin integrates language regression and diffusion-based generation to interpret and create scenes from arbitrary viewpoints. To bridge the modality gap between cameras and vision-language, we introduce a novel paradigm that treats camera as language, enabling thinking with camera. This guides the model to align spatially grounded visual cues with photographic terminology while reasoning across geometric context. Puffin is trained on Puffin-4M, a large-scale dataset of 4 million vision-language-camera triplets. We incorporate both global camera parameters and pixel-wise camera maps, yielding flexible and reliable spatial generation. Experiments demonstrate Puffin superior performance over specialized models for camera-centric generation and understanding. With instruction tuning, Puffin generalizes to diverse cross-view tasks such as spatial imagination, world exploration, and photography guidance. We will release the code, models, dataset pipeline, and benchmark to advance multimodal spatial intelligence research. --- ### Optimizing delivery for quick commerce factoring qualitative assessment of generated routes Authors: Milon Bhattacharya, Milan Kumar Published: 2025-10-09 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2510.08671v1 Indias e-commerce market is projected to grow rapidly, with last-mile delivery accounting for nearly half of operational expenses. Although vehicle routing problem (VRP) based solvers are widely used for delivery planning, their effectiveness in real-world scenarios is limited due to unstructured addresses, incomplete maps, and computational constraints in distance estimation. This study proposes a framework that employs large language models (LLMs) to critique VRP-generated routes against policy-based criteria, allowing logistics operators to evaluate and prioritise more efficient delivery plans. As a illustration of our approach we generate, annotate and evaluated 400 cases using large language models. Our study found that open-source LLMs identified routing issues with 79% accuracy, while proprietary reasoning models achieved reach upto 86%. The results demonstrate that LLM-based evaluation of VRP-generated routes can be an effective and scalable layer of evaluation which goes beyond beyond conventional distance and time based metrics. This has implications for improving cost efficiency, delivery reliability, and sustainability in last-mile logistics, especially for developing countries like India. --- ### FreqCa: Accelerating Diffusion Models via Frequency-Aware Caching Authors: Jiacheng Liu, Peiliang Cai, Qinming Zhou, Yuqi Lin, Deyang Kong, Benhao Huang, Yupei Pan, Haowen Xu, Chang Zou, Junshu Tang, Shikang Zheng, Linfeng Zhang Published: 2025-10-09 Categories: cs.LG, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2510.08669v1 The application of diffusion transformers is suffering from their significant inference costs. Recently, feature caching has been proposed to solve this problem by reusing features from previous timesteps, thereby skipping computation in future timesteps. However, previous feature caching assumes that features in adjacent timesteps are similar or continuous, which does not always hold in all settings. To investigate this, this paper begins with an analysis from the frequency domain, which reveal that different frequency bands in the features of diffusion models exhibit different dynamics across timesteps. Concretely, low-frequency components, which decide the structure of images, exhibit higher similarity but poor continuity. In contrast, the high-frequency bands, which decode the details of images, show significant continuity but poor similarity. These interesting observations motivate us to propose Frequency-aware Caching (FreqCa) which directly reuses features of low-frequency components based on their similarity, while using a second-order Hermite interpolator to predict the volatile high-frequency ones based on its continuity. Besides, we further propose to cache Cumulative Residual Feature (CRF) instead of the features in all the layers, which reduces the memory footprint of feature caching by 99%. Extensive experiments on FLUX.1-dev, FLUX.1-Kontext-dev, Qwen-Image, and Qwen-Image-Edit demonstrate its effectiveness in both generation and editing. Codes are available in the supplementary materials and will be released on GitHub. --- ### Hulu-Med: A Transparent Generalist Model towards Holistic Medical Vision-Language Understanding Authors: Songtao Jiang, Yuan Wang, Sibo Song, Tianxiang Hu, Chenyi Zhou, Bin Pu, Yan Zhang, Zhibo Yang, Yang Feng, Joey Tianyi Zhou, Jin Hao, Zijian Chen et al. Published: 2025-10-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2510.08668v2 Real-world clinical decision-making requires integrating heterogeneous data, including medical text, 2D images, 3D volumes, and videos, while existing AI systems fail to unify all these signals, limiting their utility. In this paper, we introduce Hulu-Med, a transparent, generalist medical Vision-Language Model (VLM) designed to unify language-only, 2D/3D vision-language, and video understanding within a single architecture. Hulu-Med is trained on a curated corpus of 16.7 million samples, comprising exclusively public or synthetic data, spanning 12 major anatomical systems and 14 medical imaging modalities. Hulu-Med employs a medical-aware token-reduction strategy that prunes redundant visual tokens, achieving up to a 55% reduction for 3D and video inputs, improving cross-modal efficiency, and enabling training at 7B-32B parameter scales in approximately 4,000-40,000 GPU hours. Across 30 public in-domain and out-of-domain medical benchmarks-covering text reasoning, visual question answering, report generation, multilingual dialogue, video understanding, and rare disease diagnosis-Hulu-Med surpasses existing open-source models on 27 of 30 benchmarks and outperforms proprietary systems such as GPT-4o on 16 benchmarks. Despite being a VLM, Hulu-Med outperforms GPT-4o and matches GPT-o1 on the text-only HealthBench. For the first time in the community, we provide a fully transparent, reproducible and cost-effective pipeline for holistic medical vision-language understanding by releasing our end-to-end data curation, training procedures, and model parameters. Code and models are available at https://github.com/ZJUI-AI4H/Hulu-Med. ---