# Research Papers — 2025-12 · bundle page 46 of 175 > Full-text markdown of papers #2251–2300 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/46 [← Previous bundle](https://aipod.dev/papers/archive/2025-12/page/45.md) [Next bundle →](https://aipod.dev/papers/archive/2025-12/page/47.md) ### Localising Shortcut Learning in Pixel Space via Ordinal Scoring Correlations for Attribution Representations (OSCAR) Authors: Akshit Achara, Peter Triantafillou, Esther Puyol-Antón, Alexander Hammers, Andrew P. King Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18888v1 Deep neural networks often exploit shortcuts. These are spurious cues which are associated with output labels in the training data but are unrelated to task semantics. When the shortcut features are associated with sensitive attributes, shortcut learning can lead to biased model performance. Existing methods for localising and understanding shortcut learning are mostly based upon qualitative, image-level inspection and assume cues are human-visible, limiting their use in domains such as medical imaging. We introduce OSCAR (Ordinal Scoring Correlations for Attribution Representations), a model-agnostic framework for quantifying shortcut learning and localising shortcut features. OSCAR converts image-level task attribution maps into dataset-level rank profiles of image regions and compares them across three models: a balanced baseline model (BA), a test model (TS), and a sensitive attribute predictor (SA). By computing pairwise, partial, and deviation-based correlations on these rank profiles, we produce a set of quantitative metrics that characterise the degree of shortcut reliance for TS, together with a ranking of image-level regions that contribute most to it. Experiments on CelebA, CheXpert, and ADNI show that our correlations are (i) stable across seeds and partitions, (ii) sensitive to the level of association between shortcut features and output labels in the training data, and (iii) able to distinguish localised from diffuse shortcut features. As an illustration of the utility of our method, we show how worst-group performance disparities can be reduced using a simple test-time attenuation approach based on the identified shortcut regions. OSCAR provides a lightweight, pixel-space audit that yields statistical decision rules and spatial maps, enabling users to test, localise, and mitigate shortcut reliance. The code is available at https://github.com/acharaakshit/oscar --- ### Can LLMs Estimate Student Struggles? Human-AI Difficulty Alignment with Proficiency Simulation for Item Difficulty Prediction Authors: Ming Li, Han Chen, Yunze Xiao, Jian Chen, Hong Jiao, Tianyi Zhou Published: 2025-12-21 Categories: cs.CL, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2512.18880v2 Accurate estimation of item (question or task) difficulty is critical for educational assessment but suffers from the cold start problem. While Large Language Models demonstrate superhuman problem-solving capabilities, it remains an open question whether they can perceive the cognitive struggles of human learners. In this work, we present a large-scale empirical analysis of Human-AI Difficulty Alignment for over 20 models across diverse domains such as medical knowledge and mathematical reasoning. Our findings reveal a systematic misalignment where scaling up model size is not reliably helpful; instead of aligning with humans, models converge toward a shared machine consensus. We observe that high performance often impedes accurate difficulty estimation, as models struggle to simulate the capability limitations of students even when being explicitly prompted to adopt specific proficiency levels. Furthermore, we identify a critical lack of introspection, as models fail to predict their own limitations. These results suggest that general problem-solving capability does not imply an understanding of human cognitive struggles, highlighting the challenge of using current models for automated difficulty prediction. --- ### CrashChat: A Multimodal Large Language Model for Multitask Traffic Crash Video Analysis Authors: Kaidi Liang, Ke Li, Xianbiao Hu, Ruwen Qin Published: 2025-12-21 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.18878v1 Automating crash video analysis is essential to leverage the growing availability of driving video data for traffic safety research and accountability attribution in autonomous driving. Crash video analysis is a challenging multitask problem due to the complex spatiotemporal dynamics of crash events in video data and the diverse analytical requirements involved. It requires capabilities spanning crash recognition, temporal grounding, and high-level video understanding. Existing models, however, cannot perform all these tasks within a unified framework, and effective training strategies for such models remain underexplored. To fill these gaps, this paper proposes CrashChat, a multimodal large language model (MLLM) for multitask traffic crash analysis, built upon VideoLLaMA3. CrashChat acquires domain-specific knowledge through instruction fine-tuning and employs a novel multitask learning strategy based on task decoupling and grouping, which maximizes the benefit of joint learning within and across task groups while mitigating negative transfer. Numerical experiments on consolidated public datasets demonstrate that CrashChat consistently outperforms existing MLLMs across model scales and traditional vision-based methods, achieving state-of-the-art performance. It reaches near-perfect accuracy in crash recognition, a 176\% improvement in crash localization, and a 40\% improvement in the more challenging pre-crash localization. Compared to general MLLMs, it substantially enhances textual accuracy and content coverage in crash description and reasoning tasks, with 0.18-0.41 increases in BLEU scores and 0.18-0.42 increases in ROUGE scores. Beyond its strong performance, CrashChat is a convenient, end-to-end analytical tool ready for practical implementation. The dataset and implementation code for CrashChat are available at https://github.com/Liangkd/CrashChat. --- ### Psychometric Validation of the Sophotechnic Mediation Scale and a New Understanding of the Development of GenAI Mastery: Lessons from 3,932 Adult Brazilian Workers Authors: Bruno Campello de Souza Published: 2025-12-21 Categories: cs.CY, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2512.18871v3 The rapid diffusion of generative artificial intelligence (GenAI) systems has introduced new forms of human-technology interaction, raising the question of whether sustained engagement gives rise to stable, internalized modes of cognition rather than merely transient efficiency gains. Grounded in the Cognitive Mediation Networks Theory, this study investigates Sophotechnic Mediation, a mode of thinking and acting associated with prolonged interaction with GenAI, and presents a comprehensive psychometric validation of the Sophotechnic Mediation Scale. Data were collected between 2023 and 2025 from independent cross-sectional samples totaling 3,932 adult workers from public and private organizations in the Metropolitan Region of Pernambuco, Brazil. Results indicate excellent internal consistency, a robust unidimensional structure, and measurement invariance across cohorts. Ordinal-robust confirmatory factor analyses and residual diagnostics show that elevated absolute fit indices reflect minor local dependencies rather than incorrect dimensionality. Distributional analyses reveal a time-evolving pattern characterized by a declining mass of non-adopters and convergence toward approximate Gaussianity among adopters, with model comparisons favoring a two-process hurdle model over a censored Gaussian specification. Sophotechnic Mediation is empirically distinct from Hypercultural mediation and is primarily driven by cumulative GenAI experience, with age moderating the rate of initial acquisition and the depth of later integration. Together, the findings support Sophotechnia as a coherent, measurable, and emergent mode of cognitive mediation associated with the ongoing GenAI revolution. --- ### Application of deep learning approaches for medieval historical documents transcription Authors: Maksym Voloshchuk, Bohdana Zarembovska, Mykola Kozlenko Published: 2025-12-21 Categories: cs.CV, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2512.18865v1 Handwritten text recognition and optical character recognition solutions show excellent results with processing data of modern era, but efficiency drops with Latin documents of medieval times. This paper presents a deep learning method to extract text information from handwritten Latin-language documents of the 9th to 11th centuries. The approach takes into account the properties inherent in medieval documents. The paper provides a brief introduction to the field of historical document transcription, a first-sight analysis of the raw data, and the related works and studies. The paper presents the steps of dataset development for further training of the models. The explanatory data analysis of the processed data is provided as well. The paper explains the pipeline of deep learning models to extract text information from the document images, from detecting objects to word recognition using classification models and embedding word images. The paper reports the following results: recall, precision, F1 score, intersection over union, confusion matrix, and mean string distance. The plots of the metrics are also included. The implementation is published on the GitHub repository. --- ### Cross-modal Counterfactual Explanations: Uncovering Decision Factors and Dataset Biases in Subjective Classification Authors: Alina Elena Baia, Andrea Cavallaro Published: 2025-12-21 Categories: cs.CV, cs.MM arXiv: https://arxiv.org/abs/2512.18864v1 Concept-driven counterfactuals explain decisions of classifiers by altering the model predictions through semantic changes. In this paper, we present a novel approach that leverages cross-modal decompositionality and image-specific concepts to create counterfactual scenarios expressed in natural language. We apply the proposed interpretability framework, termed Decompose and Explain (DeX), to the challenging domain of image privacy decisions, which are contextual and subjective. This application enables the quantification of the differential contributions of key scene elements to the model prediction. We identify relevant decision factors via a multi-criterion selection mechanism that considers both image similarity for minimal perturbations and decision confidence to prioritize impactful changes. This approach evaluates and compares diverse explanations, and assesses the interdependency and mutual influence among explanatory properties. By leveraging image-specific concepts, DeX generates image-grounded, sparse explanations, yielding significant improvements over the state of the art. Importantly, DeX operates as a training-free framework, offering high flexibility. Results show that DeX not only uncovers the principal contributing factors influencing subjective decisions, but also identifies underlying dataset biases allowing for targeted mitigation strategies to improve fairness. --- ### Merge on workspaces as Hopf algebra Markov chain Authors: Matilde Marcolli, David Skigin Published: 2025-12-21 Categories: math.DS, cs.CL, math.QA, math.RA arXiv: https://arxiv.org/abs/2512.18861v1 We study the dynamical properties of a Hopf algebra Markov chain with state space the binary rooted forests with labelled leaves. This Markovian dynamical system describes the core computational process of structure formation and transformation in syntax via the Merge operation, according to Chomsky's Minimalism model of generative linguistics. The dynamics decomposes into an ergodic dynamical system with uniform stationary distribution, given by the action of Internal Merge, while the contributions of External Merge and (a minimal form of) Sideward Merge reduce to a simpler Markov chain with state space the set of partitions and with combinatorial weights. The Sideward Merge part of the dynamics prevents convergence to fully formed connected structures (trees), unless the different forms of Merge are weighted by a cost function, as predicted by linguistic theory. Results on the asymptotic behavior of the Perron-Frobenius eigenvalue and eigenvector in this weighted case, obtained in terms of an associated Perron-Frobenius problem in the tropical semiring, show that the usual cost functions (Minimal Search and Resource Restrictions) proposed in the linguistic literature do not suffice to obtain convergence to the tree structures, while an additional optimization property based on the Shannon entropy achieves the expected result for the dynamics. We also comment on the introduction of continuous parameters related to semantic embedding and other computational models, and also on some filtering of the dynamics by coloring rules that model the linguistic filtering by theta roles and phase structure, and on parametric variation and the process of parameter setting in Externalization. --- ### Toward Human-Centered AI-Assisted Terminology Work Authors: Antonio San Martin Published: 2025-12-21 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.18859v3 Generative AI is likely to transform terminology work by creating new opportunities for automation. At the same time, it raises concerns about the future of terminologists and terminological resources, as efficiency pressures may encourage excessive automation based on the perception that human expertise can be replaced by AI. However, large language models remain unreliable for terminological purposes due to errors, hallucinations, and various forms of bias, making terminologists indispensable for ensuring the accuracy and reliability of terminological data. This paper argues that human-centered AI, an approach that emphasizes that AI's primary goal should be to contribute to human well-being, provides a framework for maximizing the benefits of generative AI while mitigating its risks. It contends that high levels of automation and meaningful human control are compatible and desirable, and that AI should enhance terminologists' capabilities while preserving their agency and decision-making authority. The implications of AI-assisted terminology work are examined through three interrelated dimensions: the augmented terminologist, ethical AI, and human-centered design. In particular, the paper examines how AI integration reshapes the role of the terminologist, affects professional values and working conditions, requires the management of AI-generated bias, and calls for the design of AI tools around the terminologist's needs. The paper concludes that a human-centered orientation is necessary to ensure that AI strengthens, rather than undermines, the essential role of terminology work in supporting specialized communication and the accurate transmission of knowledge across languages and cultures. --- ### CORE: Concept-Oriented Reinforcement for Bridging the Definition-Application Gap in Mathematical Reasoning Authors: Zijun Gao, Zhikun Xu, Xiao Ye, Ben Zhou Published: 2025-12-21 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.18857v3 Large language models (LLMs) often solve challenging math exercises yet fail to apply the concept right when the problem requires genuine understanding. Popular Reinforcement Learning with Verifiable Rewards (RLVR) pipelines reinforce final answers but provide little fine-grained conceptual signal, so models improve at pattern reuse rather than conceptual applications. We introduce CORE (Concept-Oriented REinforcement), an RL training framework that turns explicit concepts into a controllable supervision signal. Starting from a high-quality, low-contamination textbook resource that links verifiable exercises to concise concept descriptions, we run a sanity probe showing LLMs can restate definitions but fail concept-linked quizzes, quantifying the conceptual reasoning gap. CORE then (i) synthesizes concept-aligned quizzes, (ii) injects brief concept snippets during rollouts to elicit concept-primed trajectories, and (iii) reinforces conceptual reasoning via trajectory replacement after group failures, a lightweight forward-KL constraint that aligns unguided with concept-primed policies, or standard GRPO directly on concept-aligned quizzes. Across several models, CORE delivers consistent gains over vanilla and SFT baselines on both in-domain concept-exercise suites and diverse out-of-domain math benchmarks. CORE unifies direct training on concept-aligned quizzes and concept-injected rollouts under outcome regularization. It provides fine-grained conceptual supervision that bridges problem-solving competence and genuine conceptual reasoning, while remaining algorithm- and verifier-agnostic. --- ### VizDefender: Unmasking Visualization Tampering through Proactive Localization and Intent Inference Authors: Sicheng Song, Yanjie Zhang, Zixin Chen, Huamin Qu, Changbo Wang, Chenhui Li Published: 2025-12-21 Categories: cs.CV, cs.HC arXiv: https://arxiv.org/abs/2512.18853v1 The integrity of data visualizations is increasingly threatened by image editing techniques that enable subtle yet deceptive tampering. Through a formative study, we define this challenge and categorize tampering techniques into two primary types: data manipulation and visual encoding manipulation. To address this, we present VizDefender, a framework for tampering detection and analysis. The framework integrates two core components: 1) a semi-fragile watermark module that protects the visualization by embedding a location map to images, which allows for the precise localization of tampered regions while preserving visual quality, and 2) an intent analysis module that leverages Multimodal Large Language Models (MLLMs) to interpret manipulation, inferring the attacker's intent and misleading effects. Extensive evaluations and user studies demonstrate the effectiveness of our methods. --- ### Brain-Gen: Towards Interpreting Neural Signals for Stimulus Reconstruction Using Transformers and Latent Diffusion Models Authors: Hasib Aslam, Muhammad Talal Faiz, Muhammad Imran Malik Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18843v1 Advances in neuroscience and artificial intelligence have enabled preliminary decoding of brain activity. However, despite the progress, the interpretability of neural representations remains limited. A significant challenge arises from the intrinsic properties of electroencephalography (EEG) signals, including high noise levels, spatial diffusion, and pronounced temporal variability. To interpret the neural mechanism underlying thoughts, we propose a transformers-based framework to extract spatial-temporal representations associated with observed visual stimuli from EEG recordings. These features are subsequently incorporated into the attention mechanisms of Latent Diffusion Models (LDMs) to facilitate the reconstruction of visual stimuli from brain activity. The quantitative evaluations on publicly available benchmark datasets demonstrate that the proposed method excels at modeling the semantic structures from EEG signals; achieving up to 6.5% increase in latent space clustering accuracy and 11.8% increase in zero shot generalization across unseen classes while having comparable Inception Score and Fréchet Inception Distance with existing baselines. Our work marks a significant step towards generalizable semantic interpretation of the EEG signals. --- ### MDToC: Metacognitive Dynamic Tree of Concepts for Boosting Mathematical Problem-Solving of Large Language Models Authors: Tung Duong Ta, Tim Oates, Thien Van Luong, Huan Vu, Tien Cuong Nguyen Published: 2025-12-21 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.18841v2 Despite advances in mathematical reasoning capabilities, Large Language Models (LLMs) still struggle with calculation verification when using established prompting techniques. We present MDToC (Metacognitive Dynamic Tree of Concepts), a three-phase approach that constructs a concept tree, develops accuracy-verified calculations for each concept, and employs majority voting to evaluate competing solutions. Evaluations across CHAMP, MATH, and Game-of-24 benchmarks demonstrate our MDToC's effectiveness, with GPT-4-Turbo achieving 58.1\% on CHAMP, 86.6\% on MATH, and 85\% on Game-of-24 - outperforming GoT by 5\%, 5.4\%, and 4\% on all these tasks, respectively, without hand-engineered hints. MDToC consistently surpasses existing prompting methods across all backbone models, yielding improvements of up to 7.6\% over ToT and 6.2\% over GoT, establishing metacognitive calculation verification as a promising direction for enhanced mathematical reasoning. --- ### Generative Modeling through Koopman Spectral Analysis: An Operator-Theoretic Perspective Authors: Yuanchao Xu, Fengyi Li, Masahiro Fujisawa, Xiaoyuan Cheng, Youssef Marzouk, Isao Ishikawa Published: 2025-12-21 Categories: cs.LG, math.DS arXiv: https://arxiv.org/abs/2512.18837v3 We propose Koopman Spectral Wasserstein Gradient Descent (KSWGD), a particle-based generative modeling framework that learns the Langevin generator via Koopman theory and integrates it with Wasserstein gradient descent. Our key insight is that this spectral structure of the underlying distribution can be directly estimated from trajectory data via the Koopman operator, eliminating the need for explicit knowledge of the target potential. Additionally, we prove that KSWGD maintains an approximately constant dissipation rate, thereby establishing linear convergence and overcoming the vanishing-gradient phenomenon that hinders existing kernel-based particle methods. We further provide a Feynman--Kac interpretation that clarifies the method's probabilistic foundation. Experiments on compact manifolds, metastable multi-well systems, and high-dimensional stochastic partial differential equations demonstrate that KSWGD consistently outperforms baselines in both convergence speed and sample quality. --- ### Mix, MinHash, and Match: Cross-Source Agreement for Multilingual Pretraining Datasets Authors: Sultan Alrashed, Francesco Orabona Published: 2025-12-21 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.18834v2 Multilingual data from the web is essential for LLM pretraining. Yet, scraping it is expensive, and research groups repeatedly crawl the same content. For example, we found that over 40\% of tokens across major Arabic web corpora are duplicated between sources. In this work, we propose to use this wasteful redundancy as a quality signal to create high-quality pretraining datasets. Our key insight is that cross-source agreement functions as a free, model-free quality filter: content retained by multiple independent pipelines is more likely to represent high-quality text. Crucially, this signal requires no additional computation beyond standard deduplication, which is already performed at scale when pretraining language models. So, we propose MixMinMatch, a method that combines multiple existing web corpora, performs cross-dataset MinHash deduplication, and identifies documents independently recovered by multiple sources. We apply MixMinMatch to Arabic, Turkish, and Hindi, producing corpora that match or exceed the quality of the best single-source baselines, while providing up to 4$\times$ more unique tokens. On Arabic, our matched subset achieves a 4.5\% relative improvement over ArabicWeb24, while on Turkish, we improve over FineWeb-2 by 5.5\%. We release the datasets at: https://huggingface.co/collections/AdaMLLab/mixminmatch --- ### From Word to World: Can Large Language Models be Implicit Text-based World Models? Authors: Yixia Li, Hongru Wang, Jiahao Qiu, Zhenfei Yin, Dongdong Zhang, Cheng Qian, Zeping Li, Pony Ma, Guanhua Chen, Heng Ji Published: 2025-12-21 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.18832v2 Agentic reinforcement learning increasingly relies on experience-driven scaling, yet real-world environments remain non-adaptive, limited in coverage, and difficult to scale. World models offer a potential way to improve learning efficiency through simulated experience, but it remains unclear whether large language models can reliably serve this role and under what conditions they meaningfully benefit agents. We study these questions in text-based environments, which provide a controlled setting to reinterpret language modeling as next-state prediction under interaction. We introduce a three-level framework for evaluating LLM-based world models: (i) fidelity and consistency, (ii) scalability and robustness, and (iii) agent utility. Across five representative environments, we find that sufficiently trained world models maintain coherent latent state, scale predictably with data and model size, and improve agent performance via action verification, synthetic trajectory generation, and warm-starting reinforcement learning. Meanwhile, these gains depend critically on behavioral coverage and environment complexity, delineating clear boundry on when world modeling effectively supports agent learning. --- ### HARBOR: Holistic Adaptive Risk assessment model for BehaviORal healthcare Authors: Aditya Siddhant Published: 2025-12-21 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.18829v3 Behavioral healthcare risk assessment remains a challenging problem due to the highly multimodal nature of patient data and the temporal dynamics of mood and affective disorders. While large language models (LLMs) have demonstrated strong reasoning capabilities, their effectiveness in structured clinical risk scoring remains unclear. In this work, we introduce HARBOR, a behavioral health aware language model designed to predict a discrete mood and risk score, termed the Harbor Risk Score (HRS), on an integer scale from -3 (severe depression) to +3 (mania). We also release PEARL, a longitudinal behavioral healthcare dataset spanning four years of monthly observations from three patients, containing physiological, behavioral, and self reported mental health signals. We benchmark traditional machine learning models, proprietary LLMs, and HARBOR across multiple evaluation settings and ablations. Our results show that HARBOR outperforms classical baselines and off the shelf LLMs, achieving 69 percent accuracy compared to 54 percent for logistic regression and 29 percent for the strongest proprietary LLM baseline. --- ### Hyperbolic Graph Embeddings: a Survey and an Evaluation on Anomaly Detection Authors: Souhail Abdelmouaiz Sadat, Mohamed Yacine Touahria Miliani, Khadidja Hab El Hames, Hamida Seba, Mohammed Haddad Published: 2025-12-21 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.18826v1 This survey reviews hyperbolic graph embedding models, and evaluate them on anomaly detection, highlighting their advantages over Euclidean methods in capturing complex structures. Evaluating models like \textit{HGCAE}, \textit{\(\mathcal{P}\)-VAE}, and \textit{HGCN} demonstrates high performance, with \textit{\(\mathcal{P}\)-VAE} achieving an F1-score of 94\% on the \textit{Elliptic} dataset and \textit{HGCAE} scoring 80\% on \textit{Cora}. In contrast, Euclidean methods like \textit{DOMINANT} and \textit{GraphSage} struggle with complex data. The study emphasizes the potential of hyperbolic spaces for improving anomaly detection, and provides an open-source library to foster further research in this field. --- ### Controllable Probabilistic Forecasting with Stochastic Decomposition Layers Authors: John S. Schreck, William E. Chapman, Charlie Becker, David John Gagne, Dhamma Kimpara, Nihanth Cherukuru, Judith Berner, Kirsten J. Mayer, Negin Sobhani Published: 2025-12-21 Categories: cs.LG, cs.AI, physics.ao-ph, physics.geo-ph arXiv: https://arxiv.org/abs/2512.18815v1 AI weather prediction ensembles with latent noise injection and optimized with the continuous ranked probability score (CRPS) have produced both accurate and well-calibrated predictions with far less computational cost compared with diffusion-based methods. However, current CRPS ensemble approaches vary in their training strategies and noise injection mechanisms, with most injecting noise globally throughout the network via conditional normalization. This structure increases training expense and limits the physical interpretability of the stochastic perturbations. We introduce Stochastic Decomposition Layers (SDL) for converting deterministic machine learning weather models into probabilistic ensemble systems. Adapted from StyleGAN's hierarchical noise injection, SDL applies learned perturbations at three decoder scales through latent-driven modulation, per-pixel noise, and channel scaling. When applied to WXFormer via transfer learning, SDL requires less than 2\% of the computational cost needed to train the baseline model. Each ensemble member is generated from a compact latent tensor (5 MB), enabling perfect reproducibility and post-inference spread adjustment through latent rescaling. Evaluation on 2022 ERA5 reanalysis shows ensembles with spread-skill ratios approaching unity and rank histograms that progressively flatten toward uniformity through medium-range forecasts, achieving calibration competitive with operational IFS-ENS. Multi-scale experiments reveal hierarchical uncertainty: coarse layers modulate synoptic patterns while fine layers control mesoscale variability. The explicit latent parameterization provides interpretable uncertainty quantification for operational forecasting and climate applications. --- ### EchoMotion: Unified Human Video and Motion Generation via Dual-Modality Diffusion Transformer Authors: Yuxiao Yang, Hualian Sheng, Sijia Cai, Jing Lin, Jiahao Wang, Bing Deng, Junzhe Lu, Haoqian Wang, Jieping Ye Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18814v1 Video generation models have advanced significantly, yet they still struggle to synthesize complex human movements due to the high degrees of freedom in human articulation. This limitation stems from the intrinsic constraints of pixel-only training objectives, which inherently bias models toward appearance fidelity at the expense of learning underlying kinematic principles. To address this, we introduce EchoMotion, a framework designed to model the joint distribution of appearance and human motion, thereby improving the quality of complex human action video generation. EchoMotion extends the DiT (Diffusion Transformer) framework with a dual-branch architecture that jointly processes tokens concatenated from different modalities. Furthermore, we propose MVS-RoPE (Motion-Video Syncronized RoPE), which offers unified 3D positional encoding for both video and motion tokens. By providing a synchronized coordinate system for the dual-modal latent sequence, MVS-RoPE establishes an inductive bias that fosters temporal alignment between the two modalities. We also propose a Motion-Video Two-Stage Training Strategy. This strategy enables the model to perform both the joint generation of complex human action videos and their corresponding motion sequences, as well as versatile cross-modal conditional generation tasks. To facilitate the training of a model with these capabilities, we construct HuMoVe, a large-scale dataset of approximately 80,000 high-quality, human-centric video-motion pairs. Our findings reveal that explicitly representing human motion is complementary to appearance, significantly boosting the coherence and plausibility of human-centric video generation. --- ### Revealing Perception and Generation Dynamics in LVLMs: Mitigating Hallucinations via Validated Dominance Correction Authors: Guangtao Lyu, Xinyi Cheng, Chenghao Xu, Qi Liu, Muli Yang, Fen Fang, Huilin Chen, Jiexi Yan, Xu Yang, Cheng Deng Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18813v1 Large Vision-Language Models (LVLMs) have shown remarkable capabilities, yet hallucinations remain a persistent challenge. This work presents a systematic analysis of the internal evolution of visual perception and token generation in LVLMs, revealing two key patterns. First, perception follows a three-stage GATE process: early layers perform a Global scan, intermediate layers Approach and Tighten on core content, and later layers Explore supplementary regions. Second, generation exhibits an SAD (Subdominant Accumulation to Dominant) pattern, where hallucinated tokens arise from the repeated accumulation of subdominant tokens lacking support from attention (visual perception) or feed-forward network (internal knowledge). Guided by these findings, we devise the VDC (Validated Dominance Correction) strategy, which detects unsupported tokens and replaces them with validated dominant ones to improve output reliability. Extensive experiments across multiple models and benchmarks confirm that VDC substantially mitigates hallucinations. --- ### FedVideoMAE: Efficient Federated Video Moderation with Differential Privacy and Secure Aggregation Authors: Ziyuan Tao, Chuanzhi Xu, Sandaru Jayawardana, Adnan Mahmood, Wei Bao, Kanchana Thilakarathna, Teng Joon Lim Published: 2025-12-21 Categories: cs.CV, cs.AI, cs.MM arXiv: https://arxiv.org/abs/2512.18809v3 Short-form video moderation is increasingly pushed toward edge and privacy-sensitive settings, where users may intend videos for a limited audience, such as friends or private groups, but sending raw clips to a central server can broaden exposure, consume bandwidth, and add moderation latency. Federated learning can keep videos on device, but unprotected model updates may still leak information, and full-video backbones are expensive to communicate. We present FedVideoMAE, a privacy-preserving federated framework for violence detection that adapts a frozen VideoMAE backbone with lightweight LoRA and prompt parameters. Each training round combines self-supervised masked video reconstruction with client-side differential privacy and pairwise masked aggregation (SA) of adapter updates. Violence labels are held out from federation and used only for downstream evaluation, separating private representation learning from supervised assessment. On RWF-2000, exchanging 5,518,848 trainable parameters instead of the 156,371,328-parameter instantiated pretraining state gives a 28.3x model-state payload ratio. FedVideoMAE reaches 77.25% test accuracy without DP or SA, while accuracy under DP+SA remains in the 65.25-66.00% range. Transfer experiments on RLVS and binary UCF-Crime show similar behavior. These results characterize the privacy-utility trade-off for edge video moderation. Code is available at: https://github.com/zyt-599/FedVideoMAE --- ### Tempo as the Stable Cue: Hierarchical Mixture of Tempo and Beat Experts for Music to 3D Dance Generation Authors: Guangtao Lyu, Chenghao Xu, Qi Liu, Jiexi Yan, Muli Yang, Fen Fang, Cheng Deng Published: 2025-12-21 Categories: cs.CV, cs.MM, cs.SD arXiv: https://arxiv.org/abs/2512.18804v1 Music to 3D dance generation aims to synthesize realistic and rhythmically synchronized human dance from music. While existing methods often rely on additional genre labels to further improve dance generation, such labels are typically noisy, coarse, unavailable, or insufficient to capture the diversity of real-world music, which can result in rhythm misalignment or stylistic drift. In contrast, we observe that tempo, a core property reflecting musical rhythm and pace, remains relatively consistent across datasets and genres, typically ranging from 60 to 200 BPM. Based on this finding, we propose TempoMoE, a hierarchical tempo-aware Mixture-of-Experts module that enhances the diffusion model and its rhythm perception. TempoMoE organizes motion experts into tempo-structured groups for different tempo ranges, with multi-scale beat experts capturing fine- and long-range rhythmic dynamics. A Hierarchical Rhythm-Adaptive Routing dynamically selects and fuses experts from music features, enabling flexible, rhythm-aligned generation without manual genre labels. Extensive experiments demonstrate that TempoMoE achieves state-of-the-art results in dance quality and rhythm alignment. --- ### Reliable Audio Deepfake Detection in Variable Conditions via Quantum-Kernel SVMs Authors: Lisan Al Amin, Vandana P. Janeja Published: 2025-12-21 Categories: cs.SD, cs.AI arXiv: https://arxiv.org/abs/2512.18797v1 Detecting synthetic speech is challenging when labeled data are scarce and recording conditions vary. Existing end-to-end deep models often overfit or fail to generalize, and while kernel methods can remain competitive, their performance heavily depends on the chosen kernel. Here, we show that using a quantum kernel in audio deepfake detection reduces falsepositive rates without increasing model size. Quantum feature maps embed data into high-dimensional Hilbert spaces, enabling the use of expressive similarity measures and compact classifiers. Building on this motivation, we compare quantum-kernel SVMs (QSVMs) with classical SVMs using identical mel-spectrogram preprocessing and stratified 5-fold cross-validation across four corpora (ASVspoof 2019 LA, ASVspoof 5 (2024), ADD23, and an In-the-Wild set). QSVMs achieve consistently lower equalerror rates (EER): 0.183 vs. 0.299 on ASVspoof 5 (2024), 0.081 vs. 0.188 on ADD23, 0.346 vs. 0.399 on ASVspoof 2019, and 0.355 vs. 0.413 In-the-Wild. At the EER operating point (where FPR equals FNR), these correspond to absolute false-positiverate reductions of 0.116 (38.8%), 0.107 (56.9%), 0.053 (13.3%), and 0.058 (14.0%), respectively. We also report how consistent the results are across cross-validation folds and margin-based measures of class separation, using identical settings for both models. The only modification is the kernel; the features and SVM remain unchanged, no additional trainable parameters are introduced, and the quantum kernel is computed on a conventional computer. --- ### The Dead Salmons of AI Interpretability Authors: Maxime Méloux, Giada Dirupo, François Portet, Maxime Peyrard Published: 2025-12-21 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.18792v1 In a striking neuroscience study, the authors placed a dead salmon in an MRI scanner and showed it images of humans in social situations. Astonishingly, standard analyses of the time reported brain regions predictive of social emotions. The explanation, of course, was not supernatural cognition but a cautionary tale about misapplied statistical inference. In AI interpretability, reports of similar ''dead salmon'' artifacts abound: feature attribution, probing, sparse auto-encoding, and even causal analyses can produce plausible-looking explanations for randomly initialized neural networks. In this work, we examine this phenomenon and argue for a pragmatic statistical-causal reframing: explanations of computational systems should be treated as parameters of a (statistical) model, inferred from computational traces. This perspective goes beyond simply measuring statistical variability of explanations due to finite sampling of input data; interpretability methods become statistical estimators, and findings should be tested against explicit and meaningful alternative computational hypotheses, with uncertainty quantified with respect to the postulated statistical model. It also highlights important theoretical issues, such as the identifiability of common interpretability queries, which we argue is critical to understand the field's susceptibility to false discoveries, poor generalizability, and high variance. More broadly, situating interpretability within the standard toolkit of statistical inference opens promising avenues for future work aimed at turning AI interpretability into a pragmatic and rigorous science. --- ### Smark: A Watermark for Text-to-Speech Diffusion Models via Discrete Wavelet Transform Authors: Yichuan Zhang, Chengxin Li, Yujie Gu Published: 2025-12-21 Categories: cs.SD, cs.AI, cs.CR arXiv: https://arxiv.org/abs/2512.18791v1 Text-to-Speech (TTS) diffusion models generate high-quality speech, which raises challenges for the model intellectual property protection and speech tracing for legal use. Audio watermarking is a promising solution. However, due to the structural differences among various TTS diffusion models, existing watermarking methods are often designed for a specific model and degrade audio quality, which limits their practical applicability. To address this dilemma, this paper proposes a universal watermarking scheme for TTS diffusion models, termed Smark. This is achieved by designing a lightweight watermark embedding framework that operates in the common reverse diffusion paradigm shared by all TTS diffusion models. To mitigate the impact on audio quality, Smark utilizes the discrete wavelet transform (DWT) to embed watermarks into the relatively stable low-frequency regions of the audio, which ensures seamless watermark-audio integration and is resistant to removal during the reverse diffusion process. Extensive experiments are conducted to evaluate the audio quality and watermark performance in various simulated real-world attack scenarios. The experimental results show that Smark achieves superior performance in both audio quality and watermark extraction accuracy. --- ### RIS-Enabled Smart Wireless Environments: Fundamentals and Distributed Optimization Authors: George C. Alexandropoulos, Kostantinos D. Katsanos, George Stamatelis, Ioannis Gavras Published: 2025-12-21 Categories: eess.SP, cs.IT, cs.LG arXiv: https://arxiv.org/abs/2512.18788v1 This chapter overviews the concept of Smart Wireless Environments (SWEs) motivated by the emerging technology of Reconfigurable Intelligent Surfaces (RISs). The operating principles and state-of-the-art hardware architectures of programmable metasurfaces are first introduced. Subsequently, key performance objectives and use cases of RIS-enabled SWEs, including spectral and energy efficiency, physical-layer security, integrated sensing and communications, as well as the emerging paradigm of over-the-air computing, are discussed. Focusing on the recent trend of Beyond-Diagonal (BD) RISs, two distributed designs of respective SWEs are presented. The first deals with a multi-user Multiple-Input Single-Output (MISO) system operating within the area of influence of a SWE comprising multiple BD-RISs. A hybrid distributed and fusion machine learning framework based on multi-branch attention-based convolutional Neural Networks (NNs), NN parameter sharing, and neuroevolutionary training is presented, which enables online mapping of channel realizations to the BD-RIS configurations as well as the multi-user transmit precoder. Performance evaluation results showcase that the distributedly optimized RIS-enabled SWE achieves near-optimal sum-rate performance with low online computational complexity. The second design focuses on the wideband interference MISO broadcast channel, where each base station exclusively controls one BD-RIS to serve its assigned group of users. A cooperative optimization framework that jointly designs the base station transmit precoders as well as the tunable capacitances and switch matrices of all metasurfaces is presented. Numerical results demonstrating the superior sum-rate performance of the designed RIS-enabled SWE for multi-cell MISO networks over benchmark schemes, considering non-cooperative configuration and conventional diagonal metasurfaces, are presented. --- ### Eff-GRot: Efficient and Generalizable Rotation Estimation with Transformers Authors: Fanis Mathioulakis, Gorjan Radevski, Tinne Tuytelaars Published: 2025-12-21 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2512.18784v1 We introduce Eff-GRot, an approach for efficient and generalizable rotation estimation from RGB images. Given a query image and a set of reference images with known orientations, our method directly predicts the object's rotation in a single forward pass, without requiring object- or category-specific training. At the core of our framework is a transformer that performs a comparison in the latent space, jointly processing rotation-aware representations from multiple references alongside a query. This design enables a favorable balance between accuracy and computational efficiency while remaining simple, scalable, and fully end-to-end. Experimental results show that Eff-GRot offers a promising direction toward more efficient rotation estimation, particularly in latency-sensitive applications. --- ### From Natural Language to Control Signals: A Conceptual Framework for Semantic Channel Finding in Complex Experimental Infrastructure Authors: Thorsten Hellert, Nikolay Agladze, Alex Giovannone, Jan Jug, Frank Mayet, Mark Sherwin, Antonin Sulc, Chris Tennant Published: 2025-12-21 Categories: cs.CL, physics.acc-ph arXiv: https://arxiv.org/abs/2512.18779v1 Modern experimental platforms such as particle accelerators, fusion devices, telescopes, and industrial process control systems expose tens to hundreds of thousands of control and diagnostic channels accumulated over decades of evolution. Operators and AI systems rely on informal expert knowledge, inconsistent naming conventions, and fragmented documentation to locate signals for monitoring, troubleshooting, and automated control, creating a persistent bottleneck for reliability, scalability, and language-model-driven interfaces. We formalize semantic channel finding-mapping natural-language intent to concrete control-system signals-as a general problem in complex experimental infrastructure, and introduce a four-paradigm framework to guide architecture selection across facility-specific data regimes. The paradigms span (i) direct in-context lookup over curated channel dictionaries, (ii) constrained hierarchical navigation through structured trees, (iii) interactive agent exploration using iterative reasoning and tool-based database queries, and (iv) ontology-grounded semantic search that decouples channel meaning from facility-specific naming conventions. We demonstrate each paradigm through proof-of-concept implementations at four operational facilities spanning two orders of magnitude in scale-from compact free-electron lasers to large synchrotron light sources-and diverse control-system architectures, from clean hierarchies to legacy environments. These implementations achieve 90-97% accuracy on expert-curated operational queries. --- ### Label-Informed Outlier Detection Based on Granule Density Authors: Baiyang Chen, Zhong Yuan, Dezhong Peng, Hongmei Chen, Xiaomin Song, Huiming Zheng Published: 2025-12-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.18774v1 Outlier detection, crucial for identifying unusual patterns with significant implications across numerous applications, has drawn considerable research interest. Existing semi-supervised methods typically treat data as purely numerical and} in a deterministic manner, thereby neglecting the heterogeneity and uncertainty inherent in complex, real-world datasets. This paper introduces a label-informed outlier detection method for heterogeneous data based on Granular Computing and Fuzzy Sets, namely Granule Density-based Outlier Factor (GDOF). Specifically, GDOF first employs label-informed fuzzy granulation to effectively represent various data types and develops granule density for precise density estimation. Subsequently, granule densities from individual attributes are integrated for outlier scoring by assessing attribute relevance with a limited number of labeled outliers. Experimental results on various real-world datasets show that GDOF stands out in detecting outliers in heterogeneous data with a minimal number of labeled outliers. The integration of Fuzzy Sets and Granular Computing in GDOF offers a practical framework for outlier detection in complex and diverse data types. All relevant datasets and source codes are publicly available for further research. This is the author's accepted manuscript of a paper published in IEEE Transactions on Fuzzy Systems. The final version is available at https://doi.org/10.1109/TFUZZ.2024.3514853 --- ### In-Context Audio Control of Video Diffusion Transformers Authors: Wenze Liu, Weicai Ye, Minghong Cai, Quande Liu, Xintao Wang, Xiangyu Yue Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18772v1 Recent advancements in video generation have seen a shift towards unified, transformer-based foundation models that can handle multiple conditional inputs in-context. However, these models have primarily focused on modalities like text, images, and depth maps, while strictly time-synchronous signals like audio have been underexplored. This paper introduces In-Context Audio Control of video diffusion transformers (ICAC), a framework that investigates the integration of audio signals for speech-driven video generation within a unified full-attention architecture, akin to FullDiT. We systematically explore three distinct mechanisms for injecting audio conditions: standard cross-attention, 2D self-attention, and unified 3D self-attention. Our findings reveal that while 3D attention offers the highest potential for capturing spatio-temporal audio-visual correlations, it presents significant training challenges. To overcome this, we propose a Masked 3D Attention mechanism that constrains the attention pattern to enforce temporal alignment, enabling stable training and superior performance. Our experiments demonstrate that this approach achieves strong lip synchronization and video quality, conditioned on an audio stream and reference images. --- ### MaskFocus: Focusing Policy Optimization on Critical Steps for Masked Image Generation Authors: Guohui Zhang, Hu Yu, Xiaoxiao Ma, Yaning Pan, Hang Xu, Feng Zhao Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18766v1 Reinforcement learning (RL) has demonstrated significant potential for post-training language models and autoregressive visual generative models, but adapting RL to masked generative models remains challenging. The core factor is that policy optimization requires accounting for the probability likelihood of each step due to its multi-step and iterative refinement process. This reliance on entire sampling trajectories introduces high computational cost, whereas natively optimizing random steps often yields suboptimal results. In this paper, we present MaskFocus, a novel RL framework that achieves effective policy optimization for masked generative models by focusing on critical steps. Specifically, we determine the step-level information gain by measuring the similarity between the intermediate images at each sampling step and the final generated image. Crucially, we leverage this to identify the most critical and valuable steps and execute focused policy optimization on them. Furthermore, we design a dynamic routing sampling mechanism based on entropy to encourage the model to explore more valuable masking strategies for samples with low entropy. Extensive experiments on multiple Text-to-Image benchmarks validate the effectiveness of our method. --- ### Gaussian-Mixture-Model Q-Functions for Policy Iteration in Reinforcement Learning Authors: Minh Vu, Konstantinos Slavakis Published: 2025-12-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.18763v2 Unlike their conventional use as estimators of probability density functions in reinforcement learning (RL), this paper introduces a novel function-approximation role for Gaussian mixture models (GMMs) as direct surrogates for Q-function losses. The proposed estimators, termed GMM-QFs, possess substantial representational capacity, as they are shown to be universal approximators over a broad class of functions. They are further embedded within Bellman residuals, where their learnable parameters--a fixed number of mixing weights, together with Gaussian mean vectors and covariance matrices--are inferred from data via optimization on a product Riemannian manifold. This geometric perspective on the parameter space naturally introduces Riemannian optimization into the policy-evaluation step of standard policy-iteration (PI) frameworks. Moreover, rigorous theoretical analysis establishes performance bounds on Q-function estimation error under the proposed PI scheme. Supporting numerical tests show that GMM-QFs deliver competitive performance and, in some cases, outperform state-of-the-art approaches across a range of benchmark RL tasks, all while maintaining a significantly smaller computational footprint than deep-learning methods. --- ### MEEA: Mere Exposure Effect-Driven Confrontational Optimization for LLM Jailbreaking Authors: Jianyi Zhang, Shizhao Liu, Ziyin Zhou, Zhen Li Published: 2025-12-21 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.18755v1 The rapid advancement of large language models (LLMs) has intensified concerns about the robustness of their safety alignment. While existing jailbreak studies explore both single-turn and multi-turn strategies, most implicitly assume a static safety boundary and fail to account for how contextual interactions dynamically influence model behavior, leading to limited stability and generalization. Motivated by this gap, we propose MEEA (Mere Exposure Effect Attack), a psychology-inspired, fully automated black-box framework for evaluating multi-turn safety robustness, grounded in the mere exposure effect. MEEA leverages repeated low-toxicity semantic exposure to induce a gradual shift in a model's effective safety threshold, enabling progressive erosion of alignment constraints over sustained interactions. Concretely, MEEA constructs semantically progressive prompt chains and optimizes them using a simulated annealing strategy guided by semantic similarity, toxicity, and jailbreak effectiveness. Extensive experiments on both closed-source and open-source models, including GPT-4, Claude-3.5, and DeepSeek-R1, demonstrate that MEEA consistently achieves higher attack success rates than seven representative baselines, with an average Attack Success Rate (ASR) improvement exceeding 20%. Ablation studies further validate the necessity of both annealing-based optimization and contextual exposure mechanisms. Beyond improved attack effectiveness, our findings indicate that LLM safety behavior is inherently dynamic and history-dependent, challenging the common assumption of static alignment boundaries and highlighting the need for interaction-aware safety evaluation and defense mechanisms. Our code is available at: https://github.com/Carney-lsz/MEEA --- ### Context-Aware Network Based on Multi-scale Spatio-temporal Attention for Action Recognition in Videos Authors: Xiaoyang Li, Wenzhu Yang, Kanglin Wang, Tiebiao Wang, Qingsong Fei Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18750v1 Action recognition is a critical task in video understanding, requiring the comprehensive capture of spatio-temporal cues across various scales. However, existing methods often overlook the multi-granularity nature of actions. To address this limitation, we introduce the Context-Aware Network (CAN). CAN consists of two core modules: the Multi-scale Temporal Cue Module (MTCM) and the Group Spatial Cue Module (GSCM). MTCM effectively extracts temporal cues at multiple scales, capturing both fast-changing motion details and overall action flow. GSCM, on the other hand, extracts spatial cues at different scales by grouping feature maps and applying specialized extraction methods to each group. Experiments conducted on five benchmark datasets (Something-Something V1 and V2, Diving48, Kinetics-400, and UCF101) demonstrate the effectiveness of CAN. Our approach achieves competitive performance, outperforming most mainstream methods, with accuracies of 50.4% on Something-Something V1, 63.9% on Something-Something V2, 88.4% on Diving48, 74.9% on Kinetics-400, and 86.9% on UCF101. These results highlight the importance of capturing multi-scale spatio-temporal cues for robust action recognition. --- ### Code2Doc: A Quality-First Curated Dataset for Code Documentation Authors: Recep Kaan Karaman, Meftun Akarsu Published: 2025-12-21 Categories: cs.SE, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.18748v2 The performance of automatic code documentation generation models depends critically on the quality of the training data used for supervision. However, most existing code documentation datasets are constructed through large scale scraping of public repositories with limited quality control. As a result, they often contain noisy documentation, extensive duplication, and increasing contamination from AI generated content. These issues weaken the supervision signal available to learning-based models and complicate evaluation. We introduce Code2Doc, a quality-first curated dataset for function-level code documentation generation. Code2Doc consists of 13,358 high-quality function-documentation pairs extracted from widely used open-source repositories spanning five programming languages: Python, Java, TypeScript, JavaScript, and C++. The dataset is constructed using a four-stage curation pipeline that enforces documentation completeness and clarity, filters functions based on structural and complexity criteria, removes exact and near-duplicate code, and identifies documentation likely to be AI generated. Starting from 52,069 extracted candidates, only 25.6% satisfy all quality constraints. We provide a detailed analysis of the resulting dataset, which achieves a mean documentation quality score of 6.93 out of 10. Overall, 86.9% of samples contain explicit type annotations, and only 2.9% are flagged as potentially AI generated. Baseline experiments show that fine-tuning a large language model on Code2Doc yields relative improvements of 29.47% in BLEU and 24.04% in ROUGE-L over zero shot performance, despite the modest dataset size. We release both the dataset and the full curation pipeline to support reproducible research on automatic code documentation generation. --- ### IPCV: Information-Preserving Compression for MLLM Visual Encoders Authors: Yuan Chen, Zichen Wen, Yuzhou Wu, Xuyang Liu, Shuang Chen, Junpeng Ma, Weijia Li, Conghui He, Linfeng Zhang Published: 2025-12-21 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.18747v1 Multimodal Large Language Models (MLLMs) deliver strong vision-language performance but at high computational cost, driven by numerous visual tokens processed by the Vision Transformer (ViT) encoder. Existing token pruning strategies are inadequate: LLM-stage token pruning overlooks the ViT's overhead, while conventional ViT token pruning, without language guidance, risks discarding textually critical visual cues and introduces feature distortions amplified by the ViT's bidirectional attention. To meet these challenges, we propose IPCV, a training-free, information-preserving compression framework for MLLM visual encoders. IPCV enables aggressive token pruning inside the ViT via Neighbor-Guided Reconstruction (NGR) that temporarily reconstructs pruned tokens to participate in attention with minimal overhead, then fully restores them before passing to the LLM. Besides, we introduce Attention Stabilization (AS) to further alleviate the negative influence from token pruning by approximating the K/V of pruned tokens. It can be directly applied to previous LLM-side token pruning methods to enhance their performance. Extensive experiments show that IPCV substantially reduces end-to-end computation and outperforms state-of-the-art training-free token compression methods across diverse image and video benchmarks. Our code is available at https://github.com/Perkzi/IPCV. --- ### MemEvolve: Meta-Evolution of Agent Memory Systems Authors: Guibin Zhang, Haotian Ren, Chong Zhan, Zhenhong Zhou, Junhao Wang, He Zhu, Wangchunshu Zhou, Shuicheng Yan Published: 2025-12-21 Categories: cs.CL, cs.MA arXiv: https://arxiv.org/abs/2512.18746v1 Self-evolving memory systems are unprecedentedly reshaping the evolutionary paradigm of large language model (LLM)-based agents. Prior work has predominantly relied on manually engineered memory architectures to store trajectories, distill experience, and synthesize reusable tools, enabling agents to evolve on the fly within environment interactions. However, this paradigm is fundamentally constrained by the staticity of the memory system itself: while memory facilitates agent-level evolving, the underlying memory architecture cannot be meta-adapted to diverse task contexts. To address this gap, we propose MemEvolve, a meta-evolutionary framework that jointly evolves agents' experiential knowledge and their memory architecture, allowing agent systems not only to accumulate experience but also to progressively refine how they learn from it. To ground MemEvolve in prior research and foster openness in future self-evolving systems, we introduce EvolveLab, a unified self-evolving memory codebase that distills twelve representative memory systems into a modular design space (encode, store, retrieve, manage), providing both a standardized implementation substrate and a fair experimental arena. Extensive evaluations on four challenging agentic benchmarks demonstrate that MemEvolve achieves (I) substantial performance gains, improving frameworks such as SmolAgent and Flash-Searcher by up to $17.06\%$; and (II) strong cross-task and cross-LLM generalization, designing memory architectures that transfer effectively across diverse benchmarks and backbone models. --- ### InSight-o3: Empowering Multimodal Foundation Models with Generalized Visual Search Authors: Kaican Li, Lewei Yao, Jiannan Wu, Tiezheng Yu, Jierun Chen, Haoli Bai, Lu Hou, Lanqing Hong, Wei Zhang, Nevin L. Zhang Published: 2025-12-21 Categories: cs.CV, cs.CL, cs.LG arXiv: https://arxiv.org/abs/2512.18745v1 The ability for AI agents to "think with images" requires a sophisticated blend of reasoning and perception. However, current open multimodal agents still largely fall short on the reasoning aspect crucial for real-world tasks like analyzing documents with dense charts/diagrams and navigating maps. To address this gap, we introduce O3-Bench, a new benchmark designed to evaluate multimodal reasoning with interleaved attention to visual details. O3-Bench features challenging problems that require agents to piece together subtle visual information from distinct image areas through multi-step reasoning. The problems are highly challenging even for frontier systems like OpenAI o3, which only obtains 40.8% accuracy on O3-Bench. To make progress, we propose InSight-o3, a multi-agent framework consisting of a visual reasoning agent (vReasoner) and a visual search agent (vSearcher) for which we introduce the task of generalized visual search -- locating relational, fuzzy, or conceptual regions described in free-form language, beyond just simple objects or figures in natural images. We then present a multimodal LLM purpose-trained for this task via reinforcement learning. As a plug-and-play agent, our vSearcher empowers frontier multimodal models (as vReasoners), significantly improving their performance on a wide range of benchmarks. This marks a concrete step towards powerful o3-like open systems. Our code and dataset can be found at https://github.com/m-Just/InSight-o3 . --- ### Memorize-and-Generate: Towards Long-Term Consistency in Real-Time Video Generation Authors: Tianrui Zhu, Shiyi Zhang, Zhirui Sun, Jingqi Tian, Yansong Tang Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18741v2 Frame-level autoregressive (frame-AR) models have achieved significant progress, enabling real-time video generation comparable to bidirectional diffusion models and serving as a foundation for interactive world models and game engines. However, current approaches in long video generation typically rely on window attention, which naively discards historical context outside the window, leading to catastrophic forgetting and scene inconsistency; conversely, retaining full history incurs prohibitive memory costs. To address this trade-off, we propose Memorize-and-Generate (MAG), a framework that decouples memory compression and frame generation into distinct tasks. Specifically, we train a memory model to compress historical information into a compact KV cache, and a separate generator model to synthesize subsequent frames utilizing this compressed representation. Furthermore, we introduce MAG-Bench to strictly evaluate historical memory retention. Extensive experiments demonstrate that MAG achieves superior historical scene consistency while maintaining competitive performance on standard video generation benchmarks. --- ### AMLID: An Adaptive Multispectral Landmine Identification Dataset for Drone-Based Detection Authors: James E. Gallagher, Edward J. Oughton Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18738v1 Landmines remain a persistent humanitarian threat, with an estimated 110 million mines deployed across 60 countries, claiming approximately 26,000 casualties annually. Current detection methods are hazardous, inefficient, and prohibitively expensive. We present the Adaptive Multispectral Landmine Identification Dataset (AMLID), the first open-source dataset combining Red-Green-Blue (RGB) and Long-Wave Infrared (LWIR) imagery for Unmanned Aerial Systems (UAS)-based landmine detection. AMLID comprises of 12,078 labeled images featuring 21 globally deployed landmine types across anti-personnel and anti-tank categories in both metal and plastic compositions. The dataset spans 11 RGB-LWIR fusion levels, four sensor altitudes, two seasonal periods, and three daily illumination conditions. By providing comprehensive multispectral coverage across diverse environmental variables, AMLID enables researchers to develop and benchmark adaptive detection algorithms without requiring access to live ordnance or expensive data collection infrastructure, thereby democratizing humanitarian demining research. --- ### PIPCFR: Pseudo-outcome Imputation with Post-treatment Variables for Individual Treatment Effect Estimation Authors: Zichuan Lin, Xiaokai Huang, Jiate Liu, Yuxuan Han, Jia Chen, Xiapeng Wu, Deheng Ye Published: 2025-12-21 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.18737v1 The estimation of individual treatment effects (ITE) focuses on predicting the outcome changes that result from a change in treatment. A fundamental challenge in observational data is that while we need to infer outcome differences under alternative treatments, we can only observe each individual's outcome under a single treatment. Existing approaches address this limitation either by training with inferred pseudo-outcomes or by creating matched instance pairs. However, recent work has largely overlooked the potential impact of post-treatment variables on the outcome. This oversight prevents existing methods from fully capturing outcome variability, resulting in increased variance in counterfactual predictions. This paper introduces Pseudo-outcome Imputation with Post-treatment Variables for Counterfactual Regression (PIPCFR), a novel approach that incorporates post-treatment variables to improve pseudo-outcome imputation. We analyze the challenges inherent in utilizing post-treatment variables and establish a novel theoretical bound for ITE risk that explicitly connects post-treatment variables to ITE estimation accuracy. Unlike existing methods that ignore these variables or impose restrictive assumptions, PIPCFR learns effective representations that preserve informative components while mitigating bias. Empirical evaluations on both real-world and simulated datasets demonstrate that PIPCFR achieves significantly lower ITE errors compared to existing methods. --- ### Is Your Conditional Diffusion Model Actually Denoising? Authors: Daniel Pfrommer, Zehao Dou, Christopher Scarvelis, Max Simchowitz, Ali Jadbabaie Published: 2025-12-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.18736v1 We study the inductive biases of diffusion models with a conditioning-variable, which have seen widespread application as both text-conditioned generative image models and observation-conditioned continuous control policies. We observe that when these models are queried conditionally, their generations consistently deviate from the idealized "denoising" process upon which diffusion models are formulated, inducing disagreement between popular sampling algorithms (e.g. DDPM, DDIM). We introduce Schedule Deviation, a rigorous measure which captures the rate of deviation from a standard denoising process, and provide a methodology to compute it. Crucially, we demonstrate that the deviation from an idealized denoising process occurs irrespective of the model capacity or amount of training data. We posit that this phenomenon occurs due to the difficulty of bridging distinct denoising flows across different parts of the conditioning space and show theoretically how such a phenomenon can arise through an inductive bias towards smoothness. --- ### $M^3-Verse$: A "Spot the Difference" Challenge for Large Multimodal Models Authors: Kewei Wei, Bocheng Hu, Jie Cao, Xiaohan Chen, Zhengxi Lu, Wubing Xia, Weili Xu, Jiaao Wu, Junchen He, Mingyu Jia, Ciyun Zhao, Ye Sun et al. Published: 2025-12-21 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.18735v2 Modern Large Multimodal Models (LMMs) have demonstrated extraordinary ability in static image and single-state spatial-temporal understanding. However, their capacity to comprehend the dynamic changes of objects within a shared spatial context between two distinct video observations, remains largely unexplored. This ability to reason about transformations within a consistent environment is particularly crucial for advancements in the field of spatial intelligence. In this paper, we introduce $M^3-Verse$, a Multi-Modal, Multi-State, Multi-Dimensional benchmark, to formally evaluate this capability. It is built upon paired videos that provide multi-perspective observations of an indoor scene before and after a state change. The benchmark contains a total of 270 scenes and 2,932 questions, which are categorized into over 50 subtasks that probe 4 core capabilities. We evaluate 16 state-of-the-art LMMs and observe their limitations in tracking state transitions. To address these challenges, we further propose a simple yet effective baseline that achieves significant performance improvements in multi-state perception. $M^3-Verse$ thus provides a challenging new testbed to catalyze the development of next-generation models with a more holistic understanding of our dynamic visual world. You can get the construction pipeline from https://github.com/Wal-K-aWay/M3-Verse_pipeline and full benchmark data from https://www.modelscope.cn/datasets/WalKaWay/M3-Verse. --- ### Breast Cancer Recurrence Risk Prediction Based on Multiple Instance Learning Authors: Jinqiu Chen, Huyan Xu Published: 2025-12-21 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.18734v1 Predicting breast cancer recurrence risk is a critical clinical challenge. This study investigates the potential of computational pathology to stratify patients using deep learning on routine Hematoxylin and Eosin (H&E) stained whole-slide images (WSIs). We developed and compared three Multiple Instance Learning (MIL) frameworks -- CLAM-SB, ABMIL, and ConvNeXt-MIL-XGBoost -- on an in-house dataset of 210 patient cases. The models were trained to predict 5-year recurrence risk, categorized into three tiers (low, medium, high), with ground truth labels established by the 21-gene Recurrence Score. Features were extracted using the UNI and CONCH pre-trained models. In a 5-fold cross-validation, the modified CLAM-SB model demonstrated the strongest performance, achieving a mean Area Under the Curve (AUC) of 0.836 and a classification accuracy of 76.2%. Our findings demonstrate the feasibility of using deep learning on standard histology slides for automated, genomics-correlated risk stratification, highlighting a promising pathway toward rapid and cost-effective clinical decision support. --- ### Explainable and Fine-Grained Safeguarding of LLM Multi-Agent Systems via Bi-Level Graph Anomaly Detection Authors: Junjun Pan, Yixin Liu, Rui Miao, Kaize Ding, Yu Zheng, Quoc Viet Hung Nguyen, Alan Wee-Chung Liew, Shirui Pan Published: 2025-12-21 Categories: cs.CR, cs.AI, cs.MA arXiv: https://arxiv.org/abs/2512.18733v1 Large language model (LLM)-based multi-agent systems (MAS) have shown strong capabilities in solving complex tasks. As MAS become increasingly autonomous in various safety-critical tasks, detecting malicious agents has become a critical security concern. Although existing graph anomaly detection (GAD)-based defenses can identify anomalous agents, they mainly rely on coarse sentence-level information and overlook fine-grained lexical cues, leading to suboptimal performance. Moreover, the lack of interpretability in these methods limits their reliability and real-world applicability. To address these limitations, we propose XG-Guard, an explainable and fine-grained safeguarding framework for detecting malicious agents in MAS. To incorporate both coarse and fine-grained textual information for anomalous agent identification, we utilize a bi-level agent encoder to jointly model the sentence- and token-level representations of each agent. A theme-based anomaly detector further captures the evolving discussion focus in MAS dialogues, while a bi-level score fusion mechanism quantifies token-level contributions for explanation. Extensive experiments across diverse MAS topologies and attack scenarios demonstrate robust detection performance and strong interpretability of XG-Guard. --- ### Counterfactual Basis Extension and Representational Geometry: An MDL-Constrained Model of Conceptual Growth Authors: Chainarong Amornbunchornvej Published: 2025-12-21 Categories: cs.AI, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2512.18732v1 Concept learning becomes possible only when existing representations fail to account for experience. Most models of learning and inference, however, presuppose a fixed representational basis within which belief updating occurs. In this paper, I address a prior question: under what structural conditions can the representational basis itself expand in a principled and selective way? I propose a geometric framework in which conceptual growth is modeled as admissible basis extension evaluated under a Minimum Description Length (MDL) criterion. Experience, whether externally observed or internally simulated, is represented as vectors relative to a current conceptual subspace. Residual components capture systematic representational failure, and candidate conceptual extensions are restricted to low-rank, admissible transformations. I show that any MDL-accepted extension can be chosen so that its novel directions lie entirely within the residual span induced by experience, while extensions orthogonal to this span strictly increase description length and are therefore rejected. This yields a conservative account of imagination and conceptual innovation. Internally generated counterfactual representations contribute to learning only insofar as they expose or amplify structured residual error, and cannot introduce arbitrary novelty. I further distinguish representational counterfactuals--counterfactuals over an agent's conceptual basis--from causal or value-level counterfactuals, and show how MDL provides a normative selection principle governing representational change. Overall, the framework characterizes conceptual development as an error-driven, geometry-constrained process of basis extension, clarifying both the role and the limits of imagination in learning and theory change. --- ### A Theoretical Lens for RL-Tuned Language Models via Energy-Based Models Authors: Zhiquan Tan, Yinrong Hong Published: 2025-12-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.18730v1 Large language models (LLMs) trained via KL-regularized reinforcement learning demonstrate strong instruction following, self-correction, and reasoning abilities. Yet their theoretical underpinnings remain limited. We exploit the closed-form energy-based model (EBM) structure of the optimal KL-regularized policy to provide a unified variational analysis of LLMs. For instruction-tuned models, under natural assumptions on reward potentials and pretraining symmetry, we prove that the transition kernel satisfies detailed balance with respect to a scalar potential encoding response quality. This yields monotonic KL convergence to a high-quality stationary distribution, bounded hitting times to superior states, and exponential mixing governed by the spectral gap. For reasoning models trained with verifiable rewards (RLVR), we show the objective is equivalent to expected KL minimization toward an optimal reasoning distribution, with the suboptimality gap reducing to the Bernoulli KL between target and current accuracies along the natural gradient flow. This helps explain empirical entropy-accuracy trade-offs. --- ### ML Inference Scheduling with Predictable Latency Authors: Haidong Zhao, Nikolaos Georgantas Published: 2025-12-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.18725v3 Machine learning (ML) inference serving systems can schedule requests to improve GPU utilization and to meet service level objectives (SLOs) or deadlines. However, improving GPU utilization may compromise latency-sensitive scheduling, as concurrent tasks contend for GPU resources and thereby introduce interference. Given that interference effects introduce unpredictability in scheduling, neglecting them may compromise SLO or deadline satisfaction. Nevertheless, existing interference prediction approaches remain limited in several respects, which may restrict their usefulness for scheduling. First, they are often coarse-grained, which ignores runtime co-location dynamics and thus restricts their accuracy in interference prediction. Second, they tend to use a static prediction model, which may not effectively cope with different workload characteristics. In this paper, we evaluate the potential limitations of existing interference prediction approaches, finding that coarse-grained methods can lead to noticeable deviations in prediction accuracy and that static models degrade considerably under changing workloads. --- ### Generating Risky Samples with Conformity Constraints via Diffusion Models Authors: Han Yu, Hao Zou, Xingxuan Zhang, Zhengyi Wang, Yue He, Kehan Li, Peng Cui Published: 2025-12-21 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.18722v1 Although neural networks achieve promising performance in many tasks, they may still fail when encountering some examples and bring about risks to applications. To discover risky samples, previous literature attempts to search for patterns of risky samples within existing datasets or inject perturbation into them. Yet in this way the diversity of risky samples is limited by the coverage of existing datasets. To overcome this limitation, recent works adopt diffusion models to produce new risky samples beyond the coverage of existing datasets. However, these methods struggle in the conformity between generated samples and expected categories, which could introduce label noise and severely limit their effectiveness in applications. To address this issue, we propose RiskyDiff that incorporates the embeddings of both texts and images as implicit constraints of category conformity. We also design a conformity score to further explicitly strengthen the category conformity, as well as introduce the mechanisms of embedding screening and risky gradient guidance to boost the risk of generated samples. Extensive experiments reveal that RiskyDiff greatly outperforms existing methods in terms of the degree of risk, generation quality, and conformity with conditioned categories. We also empirically show the generalization ability of the models can be enhanced by augmenting training data with generated samples of high conformity. --- ### Unsupervised Feature Selection via Robust Autoencoder and Adaptive Graph Learning Authors: Feng Yu, MD Saifur Rahman Mazumder, Ying Su, Oscar Contreras Velasco Published: 2025-12-21 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2512.18720v2 Effective feature selection is essential for high-dimensional data analysis and machine learning. Unsupervised feature selection (UFS) aims to simultaneously cluster data and identify the most discriminative features. Most existing UFS methods linearly project features into a pseudo-label space for clustering, but they suffer from two critical limitations: (1) an oversimplified linear mapping that fails to capture complex feature relationships, and (2) an assumption of uniform cluster distributions, ignoring outliers prevalent in real-world data. To address these issues, we propose the Robust Autoencoder-based Unsupervised Feature Selection (RAEUFS) model, which leverages a deep autoencoder to learn nonlinear feature representations while inherently improving robustness to outliers. We further develop an efficient optimization algorithm for RAEUFS. Extensive experiments demonstrate that our method outperforms state-of-the-art UFS approaches in both clean and outlier-contaminated data settings. ---