# Research Papers — 2025-12 · bundle page 166 of 175 > Full-text markdown of papers #8251–8300 of 8703 published in 2025-12, newest first. AI-readable mirrors of the full 8703-paper 2025-12 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2025-12/page/166 [← Previous bundle](https://aipod.dev/papers/archive/2025-12/page/165.md) [Next bundle →](https://aipod.dev/papers/archive/2025-12/page/167.md) ### Vehicle Dynamics Embedded World Models for Autonomous Driving Authors: Huiqian Li, Wei Pan, Haodong Zhang, Jin Huang, Zhihua Zhong Published: 2025-12-02 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2512.02417v1 World models have gained significant attention as a promising approach for autonomous driving. By emulating human-like perception and decision-making processes, these models can predict and adapt to dynamic environments. Existing methods typically map high-dimensional observations into compact latent spaces and learn optimal policies within these latent representations. However, prior work usually jointly learns ego-vehicle dynamics and environmental transition dynamics from the image input, leading to inefficiencies and a lack of robustness to variations in vehicle dynamics. To address these issues, we propose the Vehicle Dynamics embedded Dreamer (VDD) method, which decouples the modeling of ego-vehicle dynamics from environmental transition dynamics. This separation allows the world model to generalize effectively across vehicles with diverse parameters. Additionally, we introduce two strategies to further enhance the robustness of the learned policy: Policy Adjustment during Deployment (PAD) and Policy Augmentation during Training (PAT). Comprehensive experiments in simulated environments demonstrate that the proposed model significantly improves both driving performance and robustness to variations in vehicle dynamics, outperforming existing approaches. --- ### Enhancing Floor Plan Recognition: A Hybrid Mix-Transformer and U-Net Approach for Precise Wall Segmentation Authors: Dmitriy Parashchuk, Alexey Kaspshitskiy, Yuriy Karyakin Published: 2025-12-02 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.02413v3 Automatic 3D reconstruction of indoor spaces from 2D floor plans necessitates high-precision semantic segmentation of structural elements, particularly walls. However, existing methods often struggle with detecting thin structures and maintaining geometric precision. To address this, we introduce MitUNet, a hybrid neural network designed to bridge the gap between global semantic context and fine-grained structural details. Our architecture combines a Mix-Transformer encoder with a U-Net decoder enhanced with spatial and channel attention blocks. Optimized with the Tversky loss function, this approach achieves a balance between precision and recall, ensuring accurate boundary recovery. Experiments on the CubiCasa5k dataset and the regional dataset demonstrate MitUNet's superiority in generating structurally correct masks with high boundary accuracy, outperforming standard models. This tool provides a robust foundation for automated 3D reconstruction pipelines. To ensure reproducibility and facilitate future research, the source code and the regional dataset are publicly available at https://github.com/aliasstudio/mitunet and https://doi.org/10.5281/zenodo.17871079, respectively. --- ### Data Curation Through the Lens of Spectral Dynamics: Static Limits, Dynamic Acceleration, and Practical Oracles Authors: Yizhou Zhang, Lun Du Published: 2025-12-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.02409v1 Large-scale neural models are increasingly trained with data pruning, synthetic data generation, cross-model distillation, reinforcement learning from human feedback (RLHF), and difficulty-based sampling. While several of these data-centric strategies reliably improve training efficiency and downstream performance, others fail to provide meaningful gains -- most notably self-generated synthetic data, which often increases dataset volume without enhancing model capability. We formalize data curation as reweighting the sampling distribution and map its effect onto the eigenstructure of the data-induced operator. Our first main result shows that \textbf{static pruning induces a bounded operator and therefore cannot change the spectral tail exponent}; it provides at most finite-region improvements and cannot alter asymptotic neural scaling. Our second result analyzes \textbf{time-dependent data curation}, showing that an ideal oracle capable of tracking spectral residuals and continuously re-normalizing the tail can provably accelerate learning -- although practical systems can only approximate this behavior. --- ### Dynamic Configuration of On-Street Parking Spaces using Multi Agent Reinforcement Learning Authors: Oshada Jayasinghe, Farhana Choudhury, Egemen Tanin, Shanika Karunasekera Published: 2025-12-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.02406v1 With increased travelling needs more than ever, traffic congestion has become a major concern in most urban areas. Allocating spaces for on-street parking, further hinders traffic flow, by limiting the effective road width available for driving. With the advancement of vehicle-to-infrastructure connectivity technologies, we explore how the impact of on-street parking on traffic congestion could be minimized, by dynamically configuring on-street parking spaces. Towards that end, we formulate dynamic on-street parking space configuration as an optimization problem, and we follow a data driven approach, considering the nature of our problem. Our proposed solution comprises a two-layer multi agent reinforcement learning based framework, which is inherently scalable to large road networks. The lane level agents are responsible for deciding the optimal parking space configuration for each lane, and we introduce a novel Deep Q-learning architecture which effectively utilizes long short term memory networks and graph attention networks to capture the spatio-temporal correlations evident in the given problem. The block level agents control the actions of the lane level agents and maintain a sufficient level of parking around the block. We conduct a set of comprehensive experiments using SUMO, on both synthetic data as well as real-world data from the city of Melbourne. Our experiments show that the proposed framework could reduce the average travel time loss of vehicles significantly, reaching upto 47%, with a negligible increase in the walking distance for parking. --- ### WISE: Weighted Iterative Society-of-Experts for Robust Multimodal Multi-Agent Debate Authors: Anoop Cherian, River Doyle, Eyal Ben-Dov, Suhas Lohit, Kuan-Chuan Peng Published: 2025-12-02 Categories: cs.CV, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.02405v1 Recent large language models (LLMs) are trained on diverse corpora and tasks, leading them to develop complementary strengths. Multi-agent debate (MAD) has emerged as a popular way to leverage these strengths for robust reasoning, though it has mostly been applied to language-only tasks, leaving its efficacy on multimodal problems underexplored. In this paper, we study MAD for solving vision-and-language reasoning problems. Our setup enables generalizing the debate protocol with heterogeneous experts that possess single- and multi-modal capabilities. To this end, we present Weighted Iterative Society-of-Experts (WISE), a generalized and modular MAD framework that partitions the agents into Solvers, that generate solutions, and Reflectors, that verify correctness, assign weights, and provide natural language feedback. To aggregate the agents' solutions across debate rounds, while accounting for variance in their responses and the feedback weights, we present a modified Dawid-Skene algorithm for post-processing that integrates our two-stage debate model. We evaluate WISE on SMART-840, VisualPuzzles, EvoChart-QA, and a new SMART-840++ dataset with programmatically generated problem instances of controlled difficulty. Our results show that WISE consistently improves accuracy by 2-7% over the state-of-the-art MAD setups and aggregation methods across diverse multimodal tasks and LLM configurations. --- ### ESACT: An End-to-End Sparse Accelerator for Compute-Intensive Transformers via Local Similarity Authors: Hongxiang Liu, Zhifang Deng, Tong Pu, Shengli Lu Published: 2025-12-02 Categories: cs.LG, cs.AR arXiv: https://arxiv.org/abs/2512.02403v2 Transformers, composed of QKV generation, attention computation, and FFNs, have become the dominant model across various domains due to their outstanding performance. However, their high computational cost hinders efficient hardware deployment. Sparsity offers a promising solution, yet most existing accelerators exploit only intra-row sparsity in attention, while few consider inter-row sparsity. Approaches leveraging inter-row sparsity often rely on costly global similarity estimation, which diminishes the acceleration benefits of sparsity, and typically apply sparsity to only one or two transformer components. Through careful analysis of the attention distribution and computation flow, we observe that local similarity allows end-to-end sparse acceleration with lower computational overhead. Motivated by this observation, we propose ESACT, an end-to-end sparse accelerator for compute-intensive Transformers. ESACT centers on the Sparsity Prediction with Local Similarity (SPLS) mechanism, which leverages HLog quantization to accurately predict local attention sparsity prior to QK generation, achieving efficient sparsity across all transformer components. To support efficient hardware realization, we introduce three architectural innovations. Experimental results on 26 benchmarks demonstrate that SPLS reduces total computation by 52.03% with less than 1% accuracy loss. ESACT achieves an end-to-end energy efficiency of 3.29 TOPS/W, and improves attention-level energy efficiency by 2.95x and 2.26x over SOTA attention accelerators SpAtten and Sanger, respectively. --- ### TaleFrame: An Interactive Story Generation System with Fine-Grained Control and Large Language Models Authors: Yunchao Wang, Guodao Sun, Zihang Fu, Zhehao Liu, Kaixing Du, Haidong Gao, Ronghua Liang Published: 2025-12-02 Categories: cs.CL, cs.HC arXiv: https://arxiv.org/abs/2512.02402v1 With the advancement of natural language generation (NLG) technologies, creative story generation systems have gained increasing attention. However, current systems often fail to accurately translate user intent into satisfactory story outputs due to a lack of fine-grained control and unclear input specifications, limiting their applicability. To address this, we propose TaleFrame, a system that combines large language models (LLMs) with human-computer interaction (HCI) to generate stories through structured information, enabling precise control over the generation process. The innovation of TaleFrame lies in decomposing the story structure into four basic units: entities, events, relationships, and story outline. We leverage the Tinystories dataset, parsing and constructing a preference dataset consisting of 9,851 JSON-formatted entries, which is then used to fine-tune a local Llama model. By employing this JSON2Story approach, structured data is transformed into coherent stories. TaleFrame also offers an intuitive interface that supports users in creating and editing entities and events and generates stories through the structured framework. Users can control these units through simple interactions (e.g., drag-and-drop, attach, and connect), thus influencing the details and progression of the story. The generated stories can be evaluated across seven dimensions (e.g., creativity, structural integrity), with the system providing suggestions for refinement based on these evaluations. Users can iteratively adjust the story until a satisfactory result is achieved. Finally, we conduct quantitative evaluation and user studies that demonstrate the usefulness of TaleFrame. Dataset available at https://huggingface.co/datasets/guodaosun/tale-frame. --- ### Nav-$R^2$ Dual-Relation Reasoning for Generalizable Open-Vocabulary Object-Goal Navigation Authors: Wentao Xiang, Haokang Zhang, Tianhang Yang, Zedong Chu, Ruihang Chu, Shichao Xie, Yujian Yuan, Jian Sun, Zhining Gu, Junjie Wang, Xiaolong Wu, Mu Xu et al. Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02400v1 Object-goal navigation in open-vocabulary settings requires agents to locate novel objects in unseen environments, yet existing approaches suffer from opaque decision-making processes and low success rate on locating unseen objects. To address these challenges, we propose Nav-$R^2$, a framework that explicitly models two critical types of relationships, target-environment modeling and environment-action planning, through structured Chain-of-Thought (CoT) reasoning coupled with a Similarity-Aware Memory. We construct a Nav$R^2$-CoT dataset that teaches the model to perceive the environment, focus on target-related objects in the surrounding context and finally make future action plans. Our SA-Mem preserves the most target-relevant and current observation-relevant features from both temporal and semantic perspectives by compressing video frames and fusing historical observations, while introducing no additional parameters. Compared to previous methods, Nav-R^2 achieves state-of-the-art performance in localizing unseen objects through a streamlined and efficient pipeline, avoiding overfitting to seen object categories while maintaining real-time inference at 2Hz. Resources will be made publicly available at \href{https://github.com/AMAP-EAI/Nav-R2}{github link}. --- ### Skywork-R1V4: Toward Agentic Multimodal Intelligence through Interleaved Thinking with Images and DeepResearch Authors: Yifan Zhang, Liang Hu, Haofeng Sun, Peiyu Wang, Yichen Wei, Shukang Yin, Jiangbo Pei, Wei Shen, Peng Xia, Yi Peng, Tianyidan Xie, Eric Li et al. Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02395v2 Despite recent progress in multimodal agentic systems, existing approaches often treat image manipulation and web search as disjoint capabilities, rely heavily on costly reinforcement learning, and lack planning grounded in real tool-execution traces. To address these limitations, we present Skywork-R1V4, a 30B (A3B) parameter multimodal agentic model that unifies multimodal planning, active image manipulation ("thinking with images"), deep multimodal search, and, most critically, interleaved reasoning that dynamically alternates between visual operations and external knowledge retrieval. Trained solely via supervised fine-tuning on fewer than 30,000 high-quality, planning-execution-consistent trajectories and validated through stepwise consistency filtering, Skywork-R1V4 achieves state-of-the-art results across perception and multimodal search benchmarks: it scores 66.1 on MMSearch and 67.2 on FVQA, surpassing Gemini 2.5 Flash on all 11 metrics. Skywork-R1V4 exhibits emergent long-horizon reasoning at inference time, successfully orchestrating more than 10 tool calls to solve complex, multi-step tasks. Our results demonstrate that sophisticated agentic multimodal intelligence can be achieved through carefully curated supervised learning alone, without any reliance on reinforcement learning. --- ### Reproducing and Extending RaDelft 4D Radar with Camera-Assisted Labels Authors: Kejia Hu, Mohammed Alsakabi, John M. Dolan, Ozan K. Tonguz Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02394v1 Recent advances in 4D radar highlight its potential for robust environment perception under adverse conditions, yet progress in radar semantic segmentation remains constrained by the scarcity of open source datasets and labels. The RaDelft data set, although seminal, provides only LiDAR annotations and no public code to generate radar labels, limiting reproducibility and downstream research. In this work, we reproduce the numerical results of the RaDelft group and demonstrate that a camera-guided radar labeling pipeline can generate accurate labels for radar point clouds without relying on human annotations. By projecting radar point clouds into camera-based semantic segmentation and applying spatial clustering, we create labels that significantly enhance the accuracy of radar labels. These results establish a reproducible framework that allows the research community to train and evaluate the labeled 4D radar data. In addition, we study and quantify how different fog levels affect the radar labeling performance. --- ### Process-Centric Analysis of Agentic Software Systems Authors: Shuyang Liu, Yang Chen, Rahul Krishna, Saurabh Sinha, Jatin Ganhotra, Reyhan Jabbarvand Published: 2025-12-02 Categories: cs.SE, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.02393v3 Agentic systems are modern software systems: they consist of orchestrated modules, expose interfaces, and are deployed in software pipelines. Unlike conventional programs, their execution, i.e., trajectories, is inherently stochastic and adaptive to the problems they solve. Evaluation of such systems is often outcome-centric. This narrow focus overlooks detailed insights, failing to explain how agents reason, plan, act, or change their strategies. Inspired by the structured representation of conventional software systems as graphs, we introduce Graphectory to systematically encode the temporal and semantic relations in such systems. Using Graphectory, we automatically analyze 4000 trajectories of two dominant agentic programming workflows, SWE-agent and OpenHands, with four backbone Large Language Models (LLMs), attempting to resolve SWE-bench issues. Our automated analyses (completed within four minutes) reveal that: (1) agents using richer prompts or stronger LLMs exhibit more complex Graphectory, reflecting deeper exploration, broader context gathering, and more thorough validation; (2) agents' strategies vary with problem difficulty and the underlying LLM - for resolved issues, strategies often follow coherent localization-patching-validation steps, while unresolved ones exhibit chaotic or backtracking behaviors; and (3) even successful agentic systems often display inefficient processes. We also implement a novel technique for real-time construction and analysis of Graphectory and Langutory during agent execution to flag trajectory issues. Upon detecting such issues, the technique notifies the agent with a diagnostic message and, when applicable, rolls back the trajectory. Experiments show that online monitoring and interventions improve resolution rates by 6.9%-23.5% across models for problematic instances, while significantly shortening trajectories with near-zero overhead. --- ### From Detection to Association: Learning Discriminative Object Embeddings for Multi-Object Tracking Authors: Yuqing Shao, Yuchen Yang, Rui Yu, Weilong Li, Xu Guo, Huaicheng Yan, Wei Wang, Xiao Sun Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02392v1 End-to-end multi-object tracking (MOT) methods have recently achieved remarkable progress by unifying detection and association within a single framework. Despite their strong detection performance, these methods suffer from relatively low association accuracy. Through detailed analysis, we observe that object embeddings produced by the shared DETR architecture display excessively high inter-object similarity, as it emphasizes only category-level discrimination within single frames. In contrast, tracking requires instance-level distinction across frames with spatial and temporal continuity, for which current end-to-end approaches insufficiently optimize object embeddings. To address this, we introduce FDTA (From Detection to Association), an explicit feature refinement framework that enhances object discriminativeness across three complementary perspectives. Specifically, we introduce a Spatial Adapter (SA) to integrate depth-aware cues for spatial continuity, a Temporal Adapter (TA) to aggregate historical information for temporal dependencies, and an Identity Adapter (IA) to leverage quality-aware contrastive learning for instance-level separability. Extensive experiments demonstrate that FDTA achieves state-of-the-art performance on multiple challenging MOT benchmarks, including DanceTrack, SportsMOT, and BFT, highlighting the effectiveness of our proposed discriminative embedding enhancement strategy. The code is available at https://github.com/Spongebobbbbbbbb/FDTA. --- ### Synthetic Error Injection Fails to Elicit Self-Correction In Language Models Authors: David X. Wu, Shreyas Kapur, Anant Sahai, Stuart Russell Published: 2025-12-02 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.02389v1 Reinforcement learning has become the dominant paradigm for eliciting reasoning and self-correction capabilities in large language models, but its computational expense motivates exploration of alternatives. Inspired by techniques from autonomous driving and robotics, we investigate whether supervised learning with synthetic error injection can induce self-correction abilities in language models. Our approach inserts artificial errors into reasoning chains, masks them, and supervises the model to recognize and correct these mistakes. Despite the intuitive appeal of this method, we find that it fails to significantly improve performance even on simple synthetic tasks across multiple models. Moreover, even when the model catches its own error, it often parrots the original mistake. We find that the distribution shift of synthetic errors to on-policy errors significantly degrades the error-correction capabilities of the fine-tuned model, even with good synthetic coverage of on-policy errors. Our results help explain why on-policy reinforcement learning methods have proven uniquely effective for eliciting self-correction. --- ### Risk-Sensitive Q-Learning in Continuous Time with Application to Dynamic Portfolio Selection Authors: Chuhan Xie Published: 2025-12-02 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2512.02386v1 This paper studies the problem of risk-sensitive reinforcement learning (RSRL) in continuous time, where the environment is characterized by a controllable stochastic differential equation (SDE) and the objective is a potentially nonlinear functional of cumulative rewards. We prove that when the functional is an optimized certainty equivalent (OCE), the optimal policy is Markovian with respect to an augmented environment. We also propose \textit{CT-RS-q}, a risk-sensitive q-learning algorithm based on a novel martingale characterization approach. Finally, we run a simulation study on a dynamic portfolio selection problem and illustrate the effectiveness of our algorithm. --- ### Reinforcement Learning in POMDP's via Direct Gradient Ascent Authors: Jonathan Baxter, Peter L. Bartlett Published: 2025-12-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.02383v1 This paper discusses theoretical and experimental aspects of gradient-based approaches to the direct optimization of policy performance in controlled POMDPs. We introduce GPOMDP, a REINFORCE-like algorithm for estimating an approximation to the gradient of the average reward as a function of the parameters of a stochastic policy. The algorithm's chief advantages are that it requires only a single sample path of the underlying Markov chain, it uses only one free parameter $β\in (0,1)$, which has a natural interpretation in terms of bias-variance trade-off, and it requires no knowledge of the underlying state. We prove convergence of GPOMDP and show how the gradient estimates produced by GPOMDP can be used in a conjugate-gradient procedure to find local optima of the average reward. --- ### On-the-fly Feedback SfM: Online Explore-and-Exploit UAV Photogrammetry with Incremental Mesh Quality-Aware Indicator and Predictive Path Planning Authors: Liyuan Lou, Wanyun Li, Wentian Gan, Yifei Yu, Tengfei Wang, Xin Wang, Zongqian Zhan Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02375v1 Compared with conventional offline UAV photogrammetry, real-time UAV photogrammetry is essential for time-critical geospatial applications such as disaster response and active digital-twin maintenance. However, most existing methods focus on processing captured images or sequential frames in real time, without explicitly evaluating the quality of the on-the-go 3D reconstruction or providing guided feedback to enhance image acquisition in the target area. This work presents On-the-fly Feedback SfM, an explore-and-exploit framework for real-time UAV photogrammetry, enabling iterative exploration of unseen regions and exploitation of already observed and reconstructed areas in near real time. Built upon SfM on-the-fly , the proposed method integrates three modules: (1) online incremental coarse-mesh generation for dynamically expanding sparse 3D point cloud; (2) online mesh quality assessment with actionable indicators; and (3) predictive path planning for on-the-fly trajectory refinement. Comprehensive experiments demonstrate that our method achieves in-situ reconstruction and evaluation in near real time while providing actionable feedback that markedly reduces coverage gaps and re-flight costs. Via the integration of data collection, processing, 3D reconstruction and assessment, and online feedback, our on the-fly feedback SfM could be an alternative for the transition from traditional passive working mode to a more intelligent and adaptive exploration workflow. Code is now available at https://github.com/IRIS-LAB-whu/OntheflySfMFeedback. --- ### SAGE: Style-Adaptive Generalization for Privacy-Constrained Semantic Segmentation Across Domains Authors: Qingmei Li, Yang Zhang, Peifeng Zhang, Haohuan Fu, Juepeng Zheng Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02369v2 Domain generalization for semantic segmentation aims to mitigate the degradation in model performance caused by domain shifts. However, in many real-world scenarios, we are unable to access the model parameters and architectural details due to privacy concerns and security constraints. Traditional fine-tuning or adaptation is hindered, leading to the demand for input-level strategies that can enhance generalization without modifying model weights. To this end, we propose a \textbf{S}tyle-\textbf{A}daptive \textbf{GE}neralization framework (\textbf{SAGE}), which improves the generalization of frozen models under privacy constraints. SAGE learns to synthesize visual prompts that implicitly align feature distributions across styles instead of directly fine-tuning the backbone. Specifically, we first utilize style transfer to construct a diverse style representation of the source domain, thereby learning a set of style characteristics that can cover a wide range of visual features. Then, the model adaptively fuses these style cues according to the visual context of each input, forming a dynamic prompt that harmonizes the image appearance without touching the interior of the model. Through this closed-loop design, SAGE effectively bridges the gap between frozen model invariance and the diversity of unseen domains. Extensive experiments on five benchmark datasets demonstrate that SAGE achieves competitive or superior performance compared to state-of-the-art methods under privacy constraints and outperforms full fine-tuning baselines in all settings. --- ### MoE-Enhanced Multi-Domain Feature Selection and Fusion for Fast Map-Free Trajectory Prediction Authors: Wenyi Xiong, Jian Chen, Ziheng Qi, Wenhua Chen Published: 2025-12-02 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.02368v2 Trajectory prediction is crucial for the reliability and safety of autonomous driving systems, yet it remains a challenging task in complex interactive scenarios due to noisy trajectory observations and intricate agent interactions. Existing methods often struggle to filter redundant scene data for discriminative information extraction, directly impairing trajectory prediction accuracy especially when handling outliers and dynamic multi-agent interactions. In response to these limitations, we present a novel map-free trajectory prediction method which adaptively eliminates redundant information and selects discriminative features across the temporal, spatial, and frequency domains, thereby enabling precise trajectory prediction in real-world driving environments. First, we design a MoE based frequency domain filter to adaptively weight distinct frequency components of observed trajectory data and suppress outlier related noise; then a selective spatiotemporal attention module that reallocates weights across temporal nodes (sequential dependencies), temporal trends (evolution patterns), and spatial nodes to extract salient information is proposed. Finally, our multimodal decoder-supervised by joint patch level and point-level losses generates reasonable and temporally consistent trajectories, and comprehensive experiments on the large-scale NuScenes and Argoverse dataset demonstrate that our method achieves competitive performance and low-latency inference performance compared with recently proposed methods. --- ### Tackling Tuberculosis: A Comparative Dive into Machine Learning for Tuberculosis Detection Authors: Daanish Hindustani, Sanober Hindustani, Preston Nguyen Published: 2025-12-02 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.02364v1 This study explores the application of machine learning models, specifically a pretrained ResNet-50 model and a general SqueezeNet model, in diagnosing tuberculosis (TB) using chest X-ray images. TB, a persistent infectious disease affecting humanity for millennia, poses challenges in diagnosis, especially in resource-limited settings. Traditional methods, such as sputum smear microscopy and culture, are inefficient, prompting the exploration of advanced technologies like deep learning and computer vision. The study utilized a dataset from Kaggle, consisting of 4,200 chest X-rays, to develop and compare the performance of the two machine learning models. Preprocessing involved data splitting, augmentation, and resizing to enhance training efficiency. Evaluation metrics, including accuracy, precision, recall, and confusion matrix, were employed to assess model performance. Results showcase that the SqueezeNet achieved a loss of 32%, accuracy of 89%, precision of 98%, recall of 80%, and an F1 score of 87%. In contrast, the ResNet-50 model exhibited a loss of 54%, accuracy of 73%, precision of 88%, recall of 52%, and an F1 score of 65%. This study emphasizes the potential of machine learning in TB detection and possible implications for early identification and treatment initiation. The possibility of integrating such models into mobile devices expands their utility in areas lacking TB detection resources. However, despite promising results, the need for continued development of faster, smaller, and more accurate TB detection models remains crucial in contributing to the global efforts in combating TB. --- ### Memory-Augmented Knowledge Fusion with Safety-Aware Decoding for Domain-Adaptive Question Answering Authors: Lei Fu, Xiang Chen, Kaige Gao Xinyue Huang, Kejian Tong Published: 2025-12-02 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.02363v1 Domain-specific question answering (QA) systems for services face unique challenges in integrating heterogeneous knowledge sources while ensuring both accuracy and safety. Existing large language models often struggle with factual consistency and context alignment in sensitive domains such as healthcare policies and government welfare. In this work, we introduce Knowledge-Aware Reasoning and Memory-Augmented Adaptation (KARMA), a novel framework designed to enhance QA performance in care scenarios. KARMA incorporates a dual-encoder architecture to fuse structured and unstructured knowledge sources, a gated memory unit to dynamically regulate external knowledge integration, and a safety-aware controllable decoder that mitigates unsafe outputs using safety classification and guided generation techniques. Extensive experiments on a proprietary QA dataset demonstrate that KARMA outperforms strong baselines in both answer quality and safety. This study offers a comprehensive solution for building trustworthy and adaptive QA systems in service contexts. --- ### VACoT: Rethinking Visual Data Augmentation with VLMs Authors: Zhengzhuo Xu, Chong Sun, SiNan Du, Chen Li, Jing Lyu, Chun Yuan Published: 2025-12-02 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.02361v1 While visual data augmentation remains a cornerstone for training robust vision models, it has received limited attention in visual language models (VLMs), which predominantly rely on large-scale real data acquisition or synthetic diversity. Consequently, they may struggle with basic perception tasks that conventional models handle reliably. Given the substantial cost of pre-training and fine-tuning VLMs, continue training on augmented data yields limited and diminishing returns. In this paper, we present Visual Augmentation Chain-of-Thought (VACoT), a framework that dynamically invokes image augmentations during model inference. By incorporating post-hoc transformations such as denoising, VACoT substantially improves robustness on challenging and out-of-distribution inputs, especially in OCR-related adversarial scenarios. Distinct from prior approaches limited to local cropping, VACoT integrates a structured collection of general visual augmentations, broadening the query image views while reducing training complexity and computational overhead with efficient agentic reinforcement learning. We propose a conditional reward scheme that encourages necessary augmentation while penalizing verbose responses, ensuring concise and effective reasoning in perception tasks. We demonstrate the superiority of VACoT with extensive experiments on 13 perception benchmarks and further introduce AdvOCR to highlight the generalization benefits of post-hoc visual augmentations in adversarial scenarios. --- ### WSCF-MVCC: Weakly-supervised Calibration-free Multi-view Crowd Counting Authors: Bin Li, Daijie Chen, Qi Zhang Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02359v1 Multi-view crowd counting can effectively mitigate occlusion issues that commonly arise in single-image crowd counting. Existing deep-learning multi-view crowd counting methods project different camera view images onto a common space to obtain ground-plane density maps, requiring abundant and costly crowd annotations and camera calibrations. Hence, calibration-free methods are proposed that do not require camera calibrations and scene-level crowd annotations. However, existing calibration-free methods still require expensive image-level crowd annotations for training the single-view counting module. Thus, in this paper, we propose a weakly-supervised calibration-free multi-view crowd counting method (WSCF-MVCC), directly using crowd count as supervision for the single-view counting module rather than density maps constructed from crowd annotations. Instead, a self-supervised ranking loss that leverages multi-scale priors is utilized to enhance the model's perceptual ability without additional annotation costs. What's more, the proposed model leverages semantic information to achieve a more accurate view matching and, consequently, a more precise scene-level crowd count estimation. The proposed method outperforms the state-of-the-art methods on three widely used multi-view counting datasets under weakly supervised settings, indicating that it is more suitable for practical deployment compared with calibrated methods. Code is released in https://github.com/zqyq/Weakly-MVCC. --- ### Beyond Playtesting: A Generative Multi-Agent Simulation System for Massively Multiplayer Online Games Authors: Ran Zhang, Kun Ouyang, Tiancheng Ma, Yida Yang, Dong Fang Published: 2025-12-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.02358v1 Optimizing numerical systems and mechanism design is crucial for enhancing player experience in Massively Multiplayer Online (MMO) games. Traditional optimization approaches rely on large-scale online experiments or parameter tuning over predefined statistical models, which are costly, time-consuming, and may disrupt player experience. Although simplified offline simulation systems are often adopted as alternatives, their limited fidelity prevents agents from accurately mimicking real player reasoning and reactions to interventions. To address these limitations, we propose a generative agent-based MMO simulation system empowered by Large Language Models (LLMs). By applying Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) on large-scale real player behavioral data, we adapt LLMs from general priors to game-specific domains, enabling realistic and interpretable player decision-making. In parallel, a data-driven environment model trained on real gameplay logs reconstructs dynamic in-game systems. Experiments demonstrate strong consistency with real-world player behaviors and plausible causal responses under interventions, providing a reliable, interpretable, and cost-efficient framework for data-driven numerical design optimization. --- ### Understanding and Harnessing Sparsity in Unified Multimodal Models Authors: Shwai He, Chaorui Deng, Ang Li, Shen Yan Published: 2025-12-02 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.02351v1 Large multimodal models have achieved remarkable progress in both understanding and generation. Recent efforts pursue unified multimodal models that integrate heterogeneous components to support both capabilities within a single framework. However, such unification introduces inference inefficiencies, e.g., specific tasks or samples may not require the full knowledge or capacity of the unified model. Yet, a systematic understanding of how these inefficiencies manifest across different components remains limited. In this work, we first conduct a systematic analysis of unified multimodal model components using training-free pruning as a probing methodology, considering both depth pruning and width reduction. Our study reveals that the understanding component exhibits notable compressibility in both understanding and generation tasks, which is more pronounced in the latter. In contrast, the generation components are highly sensitive to compression, with performance deteriorating sharply even under moderate compression ratios. To address this limitation, we propose the Mixture-of-Experts (MoE) Adaptation, inspired by the dynamic activation patterns observed across different samples. This approach partitions the generation module into multiple experts and enables sparse activation to restore generation quality. We validate the effectiveness of sparse activation through expert-frozen tuning and further demonstrate that a fully trainable adaptation delivers additional gains. As a result, the adapted BAGEL model achieves performance comparable to the full model while activating only about half of its parameters. The code is released at \href{https://github.com/Shwai-He/SparseUnifiedModel}{this link}. --- ### FOVA: Offline Federated Reinforcement Learning with Mixed-Quality Data Authors: Nan Qiao, Sheng Yue, Ju Ren, Yaoxue Zhang Published: 2025-12-02 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.02350v1 Offline Federated Reinforcement Learning (FRL), a marriage of federated learning and offline reinforcement learning, has attracted increasing interest recently. Albeit with some advancement, we find that the performance of most existing offline FRL methods drops dramatically when provided with mixed-quality data, that is, the logging behaviors (offline data) are collected by policies with varying qualities across clients. To overcome this limitation, this paper introduces a new vote-based offline FRL framework, named FOVA. It exploits a \emph{vote mechanism} to identify high-return actions during local policy evaluation, alleviating the negative effect of low-quality behaviors from diverse local learning policies. Besides, building on advantage-weighted regression (AWR), we construct consistent local and global training objectives, significantly enhancing the efficiency and stability of FOVA. Further, we conduct an extensive theoretical analysis and rigorously show that the policy learned by FOVA enjoys strict policy improvement over the behavioral policy. Extensive experiments corroborate the significant performance gains of our proposed algorithm over existing baselines on widely used benchmarks. --- ### A multi-weight self-matching visual explanation for cnns on sar images Authors: Siyuan Sun, Yongping Zhang, Hongcheng Zeng, Yamin Wang, Wei Yang, Wanting Yang, Jie Chen Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02344v2 In recent years, convolutional neural networks (CNNs) have achieved significant success in various synthetic aperture radar (SAR) tasks. However, the complexity and opacity of their internal mechanisms hinder the fulfillment of high-reliability requirements, thereby limiting their application in SAR. Improving the interpretability of CNNs is thus of great importance for their development and deployment in SAR. In this paper, a visual explanation method termed multi-weight self-matching class activation mapping (MS-CAM) is proposed. MS-CAM matches SAR images with the feature maps and corresponding gradients extracted by the CNN, and combines both channel-wise and element-wise weights to visualize the decision basis learned by the model in SAR images. Extensive experiments conducted on a self-constructed SAR target classification dataset demonstrate that MS-CAM more accurately highlights the network's regions of interest and captures detailed target feature information, thereby enhancing network interpretability. Furthermore, the feasibility of applying MS-CAM to weakly-supervised obiect localization is validated. Key factors affecting localization accuracy, such as pixel thresholds, are analyzed in depth to inform future work. --- ### Safeguarded Stochastic Polyak Step Sizes for Non-smooth Optimization: Robust Performance Without Small (Sub)Gradients Authors: Dimitris Oikonomou, Nicolas Loizou Published: 2025-12-02 Categories: math.OC, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2512.02342v3 The stochastic Polyak step size (SPS) has proven to be a promising choice for stochastic gradient descent (SGD), delivering competitive performance relative to state-of-the-art methods on smooth convex and non-convex optimization problems, including deep neural network training. However, extensions of this approach to non-smooth settings remain in their early stages, often relying on interpolation assumptions or requiring knowledge of the optimal solution. In this work, we propose a novel SPS variant, Safeguarded SPS (SPS$_{safe}$), for the stochastic subgradient method, and provide rigorous convergence guarantees for non-smooth convex optimization with no need for strong assumptions. We further incorporate momentum into the update rule, yielding equally tight theoretical results. Comprehensive experiments on convex benchmarks and deep neural networks corroborate our theory: the proposed step size achieves competitive performance to existing adaptive baselines and exhibits stable behavior across a wide range of problem settings. Finally, in the context of deep neural network training, the gradient norms under our step size do not collapse to (near) zero, indicating robustness to vanishing gradients. --- ### TALO: Pushing 3D Vision Foundation Models Towards Globally Consistent Online Reconstruction Authors: Fengyi Zhang, Tianjun Zhang, Kasra Khosoussi, Zheng Zhang, Zi Huang, Yadan Luo Published: 2025-12-02 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.02341v3 3D vision foundation models have shown strong generalization in reconstructing key 3D attributes from uncalibrated images through a single feed-forward pass. However, when deployed in online settings such as driving scenarios, predictions are made over temporal windows, making it non-trivial to maintain consistency across time. Recent strategies align consecutive predictions by solving global transformation, yet our analysis reveals their fundamental limitations in assumption validity, local alignment scope, and robustness under noisy geometry. In this work, we propose a higher-DOF and long-term alignment framework based on Thin Plate Spline, leveraging globally propagated control points to correct spatially varying inconsistencies. In addition, we adopt a point-agnostic submap registration design that is inherently robust to noisy geometry predictions. The proposed framework is fully plug-and-play, compatible with diverse 3D foundation models and camera configurations (e.g., monocular or surround-view). Extensive experiments demonstrate that our method consistently yields more coherent geometry and lower trajectory errors across multiple datasets, backbone models, and camera setups, highlighting its robustness and generality. Code is available at https://github.com/Xian-Bei/TALO. --- ### Reasoning Path and Latent State Analysis for Multi-view Visual Spatial Reasoning: A Cognitive Science Perspective Authors: Qiyao Xue, Weichen Liu, Shiqi Wang, Haoming Wang, Yuyang Wu, Wei Gao Published: 2025-12-02 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2512.02340v1 Spatial reasoning is a core aspect of human intelligence that allows perception, inference and planning in 3D environments. However, current vision-language models (VLMs) struggle to maintain geometric coherence and cross-view consistency for spatial reasoning in multi-view settings. We attribute this gap to the lack of fine-grained benchmarks that isolate multi-view reasoning from single-view perception and temporal factors. To address this, we present ReMindView-Bench, a cognitively grounded benchmark for evaluating how VLMs construct, align and maintain spatial mental models across complementary viewpoints. ReMindView-Bench systematically varies viewpoint spatial pattern and query type to probe key factors of spatial cognition. Evaluations of 15 current VLMs reveals consistent failures in cross-view alignment and perspective-taking in multi-view spatial reasoning, motivating deeper analysis on the reasoning process. Explicit phase-wise analysis using LLM-as-a-judge and self-consistency prompting shows that VLMs perform well on in-frame perception but degrade sharply when integrating information across views. Implicit analysis, including linear probing and entropy dynamics, further show progressive loss of task-relevant information and uncertainty separation between correct and incorrect trajectories. These results provide a cognitively grounded diagnosis of VLM spatial reasoning and reveal how multi-view spatial mental models are formed, degraded and destabilized across reasoning phases. The ReMindView-Bench benchmark is available at https://huggingface.co/datasets/Xue0823/ReMindView-Bench, and the source codes of benchmark construction and VLM reasoning analysis are available at https://github.com/pittisl/ReMindView-Bench. --- ### Video Diffusion Models Excel at Tracking Similar-Looking Objects Without Supervision Authors: Chenshuang Zhang, Kang Zhang, Joon Son Chung, In So Kweon, Junmo Kim, Chengzhi Mao Published: 2025-12-02 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.02339v1 Distinguishing visually similar objects by their motion remains a critical challenge in computer vision. Although supervised trackers show promise, contemporary self-supervised trackers struggle when visual cues become ambiguous, limiting their scalability and generalization without extensive labeled data. We find that pre-trained video diffusion models inherently learn motion representations suitable for tracking without task-specific training. This ability arises because their denoising process isolates motion in early, high-noise stages, distinct from later appearance refinement. Capitalizing on this discovery, our self-supervised tracker significantly improves performance in distinguishing visually similar objects, an underexplored failure point for existing methods. Our method achieves up to a 6-point improvement over recent self-supervised approaches on established benchmarks and our newly introduced tests focused on tracking visually similar items. Visualizations confirm that these diffusion-derived motion representations enable robust tracking of even identical objects across challenging viewpoint changes and deformations. --- ### SpecPV: Improving Self-Speculative Decoding for Long-Context Generation via Partial Verification Authors: Zhendong Tan, Xingjun Zhang, Chaoyi Hu, Junjie Peng, Kun Xia Published: 2025-12-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.02337v2 Growing demands from tasks like code generation, deep reasoning, and long-document understanding have made long-context generation a crucial capability for large language models (LLMs). Speculative decoding is one of the most direct and effective approaches for accelerating generation. It follows a draft-verify paradigm, where a lightweight draft model proposes several candidate tokens and the target model verifies them. However, we find that as the context length grows, verification becomes the dominant bottleneck. To further accelerate speculative decoding in long-context generation, we introduce SpecPV, a self-speculative decoding approach that performs fast verification using partial key-value states (KV) and periodically applies full verification to eliminate accumulated errors. We validate SpecPV across multiple long-context benchmarks and models, including LLaMA-3.1-8B-Instruct and Qwen3-series. Experimental results show that SpecPV achieves up to 6x decoding speedup over standard autoregressive decoding with minor degradation. --- ### Forecasting MBTA Transit Dynamics: A Performance Benchmarking of Statistical and Machine Learning Models Authors: Sai Siddharth Nalamalpu, Kaining Yuan, Aiden Zhou, Eugene Pinsky Published: 2025-12-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.02336v1 The Massachusetts Bay Transportation Authority (MBTA) is the main public transit provider in Boston, operating multiple means of transport, including trains, subways, and buses. However, the system often faces delays and fluctuations in ridership volume, which negatively affect efficiency and passenger satisfaction. To further understand this phenomenon, this paper compares the performance of existing and unique methods to determine the best approach in predicting gated station entries in the subway system (a proxy for subway usage) and the number of delays in the overall MBTA system. To do so, this research considers factors that tend to affect public transportation, such as day of week, season, pressure, wind speed, average temperature, and precipitation. This paper evaluates the performance of 10 statistical and machine learning models on predicting next-day subway usage. On predicting delay count, the number of models is extended to 11 per day by introducing a self-exciting point process model, representing a unique application of a point-process framework for MBTA delay modeling. This research involves experimenting with the selective inclusion of features to determine feature importance, testing model accuracy via Root Mean Squared Error (RMSE). Remarkably, it is found that providing either day of week or season data has a more substantial benefit to predictive accuracy compared to weather data; in fact, providing weather data generally worsens performance, suggesting a tendency of models to overfit. --- ### Retrieval-Augmented Memory for Online Learning Authors: Wenzhang Du Published: 2025-12-02 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.02333v1 Retrieval-augmented models couple parametric predictors with non-parametric memories, but their use in streaming supervised learning with concept drift is not well understood. We study online classification in non-stationary environments and propose Retrieval-Augmented Memory for Online Learning (RAM-OL), a simple extension of stochastic gradient descent that maintains a small buffer of past examples. At each time step, RAM-OL retrieves a few nearest neighbours of the current input in the hidden representation space and updates the model jointly on the current example and the retrieved neighbours. We compare a naive replay variant with a gated replay variant that constrains neighbours using a time window, similarity thresholds, and gradient reweighting, in order to balance fast reuse of relevant past data against robustness to outdated regimes. From a theoretical perspective, we interpret RAM-OL under a bounded drift model and discuss how retrieval can reduce adaptation cost and improve regret constants when patterns recur over time. Empirically, we instantiate RAM-OL on a simple online multilayer perceptron and evaluate it on three real-world data streams derived from electricity pricing, electricity load, and airline delay data. On strongly and periodically drifting streams, RAM-OL improves prequential accuracy by up to about seven percentage points and greatly reduces variance across random seeds, while on a noisy airline stream the gated variant closely matches the purely online baseline. These results show that retrieval-augmented memory is a practical and robust tool for online learning under concept drift. --- ### Molecular Embedding-Based Algorithm Selection in Protein-Ligand Docking Authors: Jiabao Brad Wang, Siyuan Cao, Hongxuan Wu, Yiliang Yuan, Mustafa Misir Published: 2025-12-02 Categories: q-bio.QM, cs.LG arXiv: https://arxiv.org/abs/2512.02328v2 Selecting an effective docking algorithm is highly context-dependent, and no single method performs reliably across structural, chemical, and protocol regimes. MolAS is a lightweight algorithm-selection model that predicts per-algorithm performance from pretrained protein and ligand embeddings using attentional pooling and a shallow residual decoder. With hundreds to a few thousand labelled complexes, MolAS achieves up to a 15 percentage-point absolute improvement over the single-best solver (SBS) and closes 17--66\% of the Virtual Best Solver (VBS)--SBS gap across five docking benchmarks. Analyses of selection frequencies, margin-conditioned reliability, and benchmark-level oracle structure indicate that MolAS is most effective when the workflow-defined oracle landscape has low winner entropy and a reasonably separable top-solver region, but degrades under protocol mismatch that shifts solver rankings and changes the induced labels. These results suggest that, in the evaluated regime, robustness is limited less by representational capacity than by workflow- and protocol-induced instability in solver hierarchies, positioning MolAS as an in-domain selector for fixed pipelines and as a diagnostic tool for assessing when docking algorithm selection is well-posed. --- ### Training Energy-Based Models with Non-MCMC Samplers and Efficient Temperature Estimation Authors: Kentaro Kubo, Hayato Goto Published: 2025-12-02 Categories: cs.LG, quant-ph, stat.ML arXiv: https://arxiv.org/abs/2512.02323v2 Efficient sampling from Boltzmann distributions over discrete variables is a fundamental operation in a wide range of applications. While fast non-MCMC samplers have recently emerged as promising alternatives to conventional MCMC methods, their practical use for probabilistic learning remains hindered by the difficulty of estimating the effective temperature of the generated samples. In this work, we begin by introducing Langevin simulated bifurcation (LSB), a Boltzmann sampler that enables fast and parallel sampling with accuracy comparable to sequential MCMC methods. To address the challenge of unknown effective temperature, we propose conditional expectation matching (CEM), an efficient estimation method applicable to energy-based models (EBMs) with exploitable conditional independence structures. Building on these components, we further develop a learning framework, termed sampler adaptive learning (SAL), which adaptively adjusts the model temperature to match that of the distribution induced by fast non-MCMC sampling. We demonstrate the effectiveness of LSB, CEM, and SAL on semi-restricted Boltzmann machines (SRBMs), a class of EBMs that are difficult to train using conventional approaches. LSB achieves orders-of-magnitude acceleration over Gibbs sampling while maintaining comparable or higher accuracy, and CEM enables accurate temperature estimation of the resulting distribution with negligible computational overhead. As a consequence, SAL enables efficient training of SRBMs and outperforms conventional Boltzmann machine learning methods on synthetic spin-glass datasets. In addition, the trained models achieve strong performance across multiple tasks. These results establish LSB as a fast and accurate Boltzmann sampler and provide key insights that enable practical applications of fast non-MCMC sampling methods via efficient temperature estimation with CEM. --- ### LeechHijack: Covert Computational Resource Exploitation in Intelligent Agent Systems Authors: Yuanhe Zhang, Weiliu Wang, Zhenhong Zhou, Kun Wang, Jie Zhang, Li Sun, Yang Liu, Sen Su Published: 2025-12-02 Categories: cs.CR, cs.CL arXiv: https://arxiv.org/abs/2512.02321v1 Large Language Model (LLM)-based agents have demonstrated remarkable capabilities in reasoning, planning, and tool usage. The recently proposed Model Context Protocol (MCP) has emerged as a unifying framework for integrating external tools into agent systems, enabling a thriving open ecosystem of community-built functionalities. However, the openness and composability that make MCP appealing also introduce a critical yet overlooked security assumption -- implicit trust in third-party tool providers. In this work, we identify and formalize a new class of attacks that exploit this trust boundary without violating explicit permissions. We term this new attack vector implicit toxicity, where malicious behaviors occur entirely within the allowed privilege scope. We propose LeechHijack, a Latent Embedded Exploit for Computation Hijacking, in which an adversarial MCP tool covertly expropriates the agent's computational resources for unauthorized workloads. LeechHijack operates through a two-stage mechanism: an implantation stage that embeds a benign-looking backdoor in a tool, and an exploitation stage where the backdoor activates upon predefined triggers to establish a command-and-control channel. Through this channel, the attacker injects additional tasks that the agent executes as if they were part of its normal workflow, effectively parasitizing the user's compute budget. We implement LeechHijack across four major LLM families. Experiments show that LeechHijack achieves an average success rate of 77.25%, with a resource overhead of 18.62% compared to the baseline. This study highlights the urgent need for computational provenance and resource attestation mechanisms to safeguard the emerging MCP ecosystem. --- ### Associative Memory using Attribute-Specific Neuron Groups-1: Learning between Multiple Cue Balls Authors: Hiroshi Inazawa Published: 2025-12-02 Categories: cs.NE arXiv: https://arxiv.org/abs/2512.02319v2 In this paper, we present a new neural network model based on attribute-specific representations (e.g., color, shape, size), a classic example of associative memory. The proposed model is based on a previous study on memory and recall of multiple images using the Cue Ball and Recall Net (referred to as the CB-RN system, or simply CB-RN) (1). The system consists of three components, which are C.CB-RN for processing color, S.CB-RN for processing shape, and V.CB-RN for processing size. When an attribute data pattern is presented to the CB-RN system, the corresponding attribute pattern of the cue neurons within the Cue Balls is associatively recalled in the Recall Net. Each image pattern presented to these CB-RN systems is represented using a two-dimensional code, specifically a QR code (2). --- ### COGNITION: From Evaluation to Defense against Multimodal LLM CAPTCHA Solvers Authors: Junyu Wang, Changjia Zhu, Yuanbo Zhou, Lingyao Li, Xu He, Mingkui Wei, Junjie Xiong Published: 2025-12-02 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2512.02318v4 This paper studies how multimodal large language models (MLLMs) undermine the security guarantees of visual CAPTCHA. We identify the attack surface where an adversary can cheaply automate CAPTCHA solving using off-the-shelf models. We evaluate 7 representative MLLMs on 18 real-world CAPTCHA task types, measuring single-shot accuracy, success under limited retries, end-to-end latency, and per-solve cost. We further validate our findings through a supplemental external dataset and an adaptive-attacker setting with session memory, while also analyzing the impact of task-specific prompt engineering and few-shot demonstrations on solver effectiveness. We reveal that MLLMs can reliably solve recognition-oriented and low-interaction CAPTCHA tasks at human-like cost and latency, whereas tasks requiring fine-grained localization, multi-step spatial reasoning, or cross-frame consistency remain significantly harder for current models. By examining the reasoning traces of such MLLMs, we investigate the underlying mechanisms of why models succeed/fail on specific CAPTCHA puzzles and use these insights to derive defense-oriented guidelines for selecting and strengthening CAPTCHA tasks. To validate these principles, we present a proof-of-concept by hardening a vulnerable CAPTCHA type using our guidelines. We demonstrate that incorporating fine-grained localization and implicit counting reduces the success rate of state-of-the-art MLLMs from over 95\% to 0\%, confirming that structural changes can effectively mitigate the threat. We conclude by emphasizing the urgent need for CAPTCHA redesign as MLLM capabilities increasingly threaten existing defenses. Code Availability (https://doi.org/10.5281/zenodo.20406852). --- ### Few-shot Protein Fitness Prediction via In-context Learning and Test-time Training Authors: Felix Teufel, Aaron W. Kollasch, Yining Huang, Ole Winther, Kevin K. Yang, Pascal Notin, Debora S. Marks Published: 2025-12-02 Categories: q-bio.BM, cs.LG arXiv: https://arxiv.org/abs/2512.02315v1 Accurately predicting protein fitness with minimal experimental data is a persistent challenge in protein engineering. We introduce PRIMO (PRotein In-context Mutation Oracle), a transformer-based framework that leverages in-context learning and test-time training to adapt rapidly to new proteins and assays without large task-specific datasets. By encoding sequence information, auxiliary zero-shot predictions, and sparse experimental labels from many assays as a unified token set in a pre-training masked-language modeling paradigm, PRIMO learns to prioritize promising variants through a preference-based loss function. Across diverse protein families and properties-including both substitution and indel mutations-PRIMO outperforms zero-shot and fully supervised baselines. This work underscores the power of combining large-scale pre-training with efficient test-time adaptation to tackle challenging protein design tasks where data collection is expensive and label availability is limited. --- ### OmniGuard: Unified Omni-Modal Guardrails with Deliberate Reasoning Authors: Boyu Zhu, Xiaofei Wen, Wenjie Jacky Mo, Tinghui Zhu, Yanan Xie, Peng Qi, Muhao Chen Published: 2025-12-02 Categories: cs.AI, cs.CL, cs.CR, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2512.02306v1 Omni-modal Large Language Models (OLLMs) that process text, images, videos, and audio introduce new challenges for safety and value guardrails in human-AI interaction. Prior guardrail research largely targets unimodal settings and typically frames safeguarding as binary classification, which limits robustness across diverse modalities and tasks. To address this gap, we propose OmniGuard, the first family of omni-modal guardrails that performs safeguarding across all modalities with deliberate reasoning ability. To support the training of OMNIGUARD, we curate a large, comprehensive omni-modal safety dataset comprising over 210K diverse samples, with inputs that cover all modalities through both unimodal and cross-modal samples. Each sample is annotated with structured safety labels and carefully curated safety critiques from expert models through targeted distillation. Extensive experiments on 15 benchmarks show that OmniGuard achieves strong effectiveness and generalization across a wide range of multimodal safety scenarios. Importantly, OmniGuard provides a unified framework that enforces policies and mitigates risks in omni-modalities, paving the way toward building more robust and capable omnimodal safeguarding systems. --- ### When Does Verification Pay Off? A Closer Look at LLMs as Solution Verifiers Authors: Jack Lu, Ryan Teehan, Jinran Jin, Mengye Ren Published: 2025-12-02 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.02304v2 Large language models (LLMs) can act as both problem solvers and solution verifiers, where the latter select high-quality answers from a pool of solver-generated candidates. This raises the question of under what conditions verification pays off in solver-verifier systems. Prior work has conducted only limited studies of the factors influencing verification performance, focusing primarily on self-verification and examining neither the relationship between solver and verifier model families nor the effects of reasoning post-training. To rectify this, we present a systematic study across 37 models spanning multiple families, sizes, and base vs. post-trained variants, evaluated on 9 benchmarks covering logical reasoning, structured puzzles, symbolic computation, mathematics, commonsense, factual recall, and domain knowledge. In order to support our analysis, we introduce and empirically validate verifier gain, a metric that predicts the performance improvements from test-time verifier-based rejection sampling. Our experiments find that 1) verification across model families is more effective than either self-verification or verification within the same family, and more generally that the benefits of verification decrease as the solver and verifier become more similar, 2) reasoning post-training weakens self-improvement abilities but strengthens cross-family improvement, and 3) some tasks are inherently more amenable to improvement through verification, particularly mathematical and logical tasks. --- ### Training Dynamics of Learning 3D-Rotational Equivariance Authors: Max W. Shen, Ewa Nowara, Michael Maser, Kyunghyun Cho Published: 2025-12-02 Categories: cs.LG, q-bio.BM arXiv: https://arxiv.org/abs/2512.02303v1 While data augmentation is widely used to train symmetry-agnostic models, it remains unclear how quickly and effectively they learn to respect symmetries. We investigate this by deriving a principled measure of equivariance error that, for convex losses, calculates the percent of total loss attributable to imperfections in learned symmetry. We focus our empirical investigation to 3D-rotation equivariance on high-dimensional molecular tasks (flow matching, force field prediction, denoising voxels) and find that models reduce equivariance error quickly to $\leq$2\% held-out loss within 1k-10k training steps, a result robust to model and dataset size. This happens because learning 3D-rotational equivariance is an easier learning task, with a smoother and better-conditioned loss landscape, than the main prediction task. For 3D rotations, the loss penalty for non-equivariant models is small throughout training, so they may achieve lower test loss than equivariant models per GPU-hour unless the equivariant ``efficiency gap'' is narrowed. We also experimentally and theoretically investigate the relationships between relative equivariance error, learning gradients, and model parameters. --- ### Breast Cell Segmentation Under Extreme Data Constraints: Quantum Enhancement Meets Adaptive Loss Stabilization Authors: Varun Kumar Dasoju, Qingsu Cheng, Zeyun Yu Published: 2025-12-02 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.02302v1 Annotating medical images demands significant time and expertise, often requiring pathologists to invest hundreds of hours in labeling mammary epithelial nuclei datasets. We address this critical challenge by achieving 95.5% Dice score using just 599 training images for breast cell segmentation, where just 4% of pixels represent breast tissue and 60% of images contain no breast regions. Our framework uses quantum-inspired edge enhancement via multi-scale Gabor filters creating a fourth input channel, enhancing boundary detection where inter-annotator variations reach +/- 3 pixels. We present a stabilized multi-component loss function that integrates adaptive Dice loss with boundary-aware terms and automatic positive weighting to effectively address severe class imbalance, where mammary epithelial cell regions comprise only 0.1%-20% of the total image area. Additionally, a complexity-based weighted sampling strategy is introduced to prioritize the challenging mammary epithelial cell regions. The model employs an EfficientNet-B7/UNet++ architecture with a 4-to-3 channel projection, enabling the use of pretrained weights despite limited medical imaging data. Finally, robust validation is achieved through exponential moving averaging and statistical outlier detection, ensuring reliable performance estimates on a small validation set (129 images). Our framework achieves a Dice score of 95.5% +/- 0.3% and an IoU of 91.2% +/- 0.4%. Notably, quantum-based enhancement contributes to a 2.1% improvement in boundary accuracy, while weighted sampling increases small lesion detection by 3.8%. By achieving groundbreaking performance with limited annotations, our approach significantly reduces the medical expert time required for dataset creation, addressing a fundamental bottleneck in clinical perception AI development. --- ### HealthContradict: Evaluating Biomedical Knowledge Conflicts in Language Models Authors: Boya Zhang, Alban Bornet, Rui Yang, Nan Liu, Douglas Teodoro Published: 2025-12-02 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.02299v1 How do language models use contextual information to answer health questions? How are their responses impacted by conflicting contexts? We assess the ability of language models to reason over long, conflicting biomedical contexts using HealthContradict, an expert-verified dataset comprising 920 unique instances, each consisting of a health-related question, a factual answer supported by scientific evidence, and two documents presenting contradictory stances. We consider several prompt settings, including correct, incorrect or contradictory context, and measure their impact on model outputs. Compared to existing medical question-answering evaluation benchmarks, HealthContradict provides greater distinctions of language models' contextual reasoning capabilities. Our experiments show that the strength of fine-tuned biomedical language models lies not only in their parametric knowledge from pretraining, but also in their ability to exploit correct context while resisting incorrect context. --- ### VIGS-SLAM: Visual Inertial Gaussian Splatting SLAM Authors: Zihan Zhu, Wei Zhang, Moyang Li, Norbert Haala, Marc Pollefeys, Daniel Barath Published: 2025-12-02 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2512.02293v2 We present VIGS-SLAM, a visual-inertial 3D Gaussian Splatting SLAM system that achieves robust real-time tracking and high-fidelity reconstruction. Although recent 3DGS-based SLAM methods achieve dense and photorealistic mapping, their purely visual design degrades under challenging conditions such as motion blur, low texture, and exposure variations. Our method tightly couples visual and inertial cues within a unified optimization framework, jointly optimizing camera poses, depths, and IMU states. It features robust IMU initialization, time-varying bias modeling, and loop closure with consistent Gaussian updates. Experiments on five challenging datasets demonstrate our superiority over state-of-the-art methods. Project page: https://vigs-slam.github.io --- ### Enhancing Cross Domain SAR Oil Spill Segmentation via Morphological Region Perturbation and Synthetic Label-to-SAR Generation Authors: Andre Juarez, Luis Salsavilca, Frida Coaquira, Celso Gonzales Published: 2025-12-02 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.02290v1 Deep learning models for SAR oil spill segmentation often fail to generalize across regions due to differences in sea-state, backscatter statistics, and slick morphology, a limitation that is particularly severe along the Peruvian coast where labeled Sentinel-1 data remain scarce. To address this problem, we propose \textbf{MORP--Synth}, a two-stage synthetic augmentation framework designed to improve transfer from Mediterranean to Peruvian conditions. Stage~A applies Morphological Region Perturbation, a curvature guided label space method that generates realistic geometric variations of oil and look-alike regions. Stage~B renders SAR-like textures from the edited masks using a conditional generative INADE model. We compile a Peruvian dataset of 2112 labeled 512$\times$512 patches from 40 Sentinel-1 scenes (2014--2024), harmonized with the Mediterranean CleanSeaNet benchmark, and evaluate seven segmentation architectures. Models pretrained on Mediterranean data degrade from 67.8\% to 51.8\% mIoU on the Peruvian domain; MORP--Synth improves performance up to +6 mIoU and boosts minority-class IoU (+10.8 oil, +14.6 look-alike). --- ### Serendipity with Generative AI: Repurposing knowledge components during polycrisis with a Viable Systems Model approach Authors: Gordon Fletcher, Saomai Vu Khan Published: 2025-12-01 Categories: cs.HC, cs.CL, cs.IR arXiv: https://arxiv.org/abs/2602.23365v1 Organisations face polycrisis uncertainty yet overlook embedded knowledge. We show how generative AI can operate as a serendipity engine and knowledge transducer to discover, classify and mobilise reusable components (models, frameworks, patterns) from existing documents. Using 206 papers, our pipeline extracted 711 components (approx 3.4 per paper) and organised them into a repository aligned to Beer's Viable System Model (VSM). We contribute i) conceptually, a theory of planned serendipity in which GenAI lowers transduction costs between VSM subsystems, ii) empirically, a component repository and temporal/subject patterns, iii) managerially, a vignette and process blueprint for organisational adoption and iv) socially, pathways linking repurposing to environmental and social benefits. We propose testable links between repository creation, discovery-to-deployment time, and reuse rates, and discuss implications for shifting innovation portfolios from breakthrough bias toward systematic repurposing. --- ### Dynamic System Instructions and Tool Exposure for Efficient Agentic LLMs Authors: Uria Franko Published: 2025-12-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2602.17046v1 Large Language Model (LLM) agents often run for many steps while re-ingesting long system instructions and large tool catalogs each turn. This increases cost, agent derailment probability, latency, and tool-selection errors. We propose Instruction-Tool Retrieval (ITR), a RAG variant that retrieves, per step, only the minimal system-prompt fragments and the smallest necessary subset of tools. ITR composes a dynamic runtime system prompt and exposes a narrowed toolset with confidence-gated fallbacks. Using a controlled benchmark with internally consistent numbers, ITR reduces per-step context tokens by 95%, improves correct tool routing by 32% relative, and cuts end-to-end episode cost by 70% versus a monolithic baseline. These savings enable agents to run 2-20x more loops within context limits. Savings compound with the number of agent steps, making ITR particularly valuable for long-running autonomous agents. We detail the method, evaluation protocol, ablations, and operational guidance for practical deployment. --- ### Sales Research Agent and Sales Research Bench Authors: Deepanjan Bhol Published: 2025-12-01 Categories: cs.AI arXiv: https://arxiv.org/abs/2602.17017v1 Enterprises increasingly need AI systems that can answer sales-leader questions over live, customized CRM data, but most available models do not expose transparent, repeatable evidence of quality. This paper describes the Sales Research Agent in Microsoft Dynamics 365 Sales, an AI-first application that connects to live CRM and related data, reasons over complex schemas, and produces decision-ready insights through text and chart outputs. To make quality observable, we introduce the Sales Research Bench, a purpose-built benchmark that scores systems on eight customer-weighted dimensions, including text and chart groundedness, relevance, explainability, schema accuracy, and chart quality. In a 200-question run on a customized enterprise schema on October 19, 2025, the Sales Research Agent outperformed Claude Sonnet 4.5 by 13 points and ChatGPT-5 by 24.1 points on the 100-point composite score, giving customers a repeatable way to compare AI solutions. --- ### Call2Instruct: Automated Pipeline for Generating Q&A Datasets from Call Center Recordings for LLM Fine-Tuning Authors: Alex Echeverria, Sávio Salvarino Teles de Oliveira, Fernando Marques Federson Published: 2025-12-01 Categories: cs.LG, cs.AI, cs.CL, cs.HC, cs.SD, eess.AS arXiv: https://arxiv.org/abs/2601.14263v1 The adaptation of Large-Scale Language Models (LLMs) to specific domains depends on high-quality fine-tuning datasets, particularly in instructional format (e.g., Question-Answer - Q&A). However, generating these datasets, particularly from unstructured sources such as call center audio recordings, poses a significant challenge due to the noisy and disorganized nature of the data. This paper presents a solution to this challenge by offering an end-to-end automated pipeline for generating Q&A instructional datasets from such recordings. The methodology developed comprises sequential steps of audio processing (including diarization, noise removal and automatic transcription), textual processing (cleaning, normalization, and anonymization), semantic extraction of customer demands and attendant responses using vector embeddings, and matching via semantic search to form the final Q&A pairs. As a result, the complete pipeline was successfully implemented, generating a dataset specifically formatted for Instruct Fine Tuning. The practical value and feasibility of the generated dataset were substantiated and functionally demonstrated through the successful fine-tuning of an LLM model (based on Llama 2 7B). The conclusion of the paper states that the proposed approach is viable for converting unstructured conversational data from call centers into valuable resources for training LLMs. This development has the potential to open up avenues for creating more effective AI systems for Q&A tasks in the customer service domain. The developed codes have been made publicly available to promote reproducibility and future research. ---