# Research Papers — 2025-10 · bundle page 118 of 217

> Full-text markdown of papers #5851–5900 of 10843 published in 2025-10, newest first. AI-readable mirrors of the full 10843-paper 2025-10 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2025-10/page/118
[← Previous bundle](https://aipod.dev/papers/archive/2025-10/page/117.md)
[Next bundle →](https://aipod.dev/papers/archive/2025-10/page/119.md)

### Generation Space Size: Understanding and Calibrating Open-Endedness of LLM Generations

Authors: Sunny Yu, Ahmad Jabbar, Robert Hawkins, Dan Jurafsky, Myra Cheng
Published: 2025-10-14
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.12699v1

Different open-ended generation tasks require different degrees of output diversity. However, current LLMs are often miscalibrated. They collapse to overly homogeneous outputs for creative tasks and hallucinate diverse but incorrect responses for factual tasks. We argue that these two failure modes are unified by, and can both be addressed by, the notion of effective generation space size (GSS) -- the set of semantically distinct outputs a model considers for a prompt. We present GSSBench, a task suite of prompt pairs with ground-truth GSS relationships to assess different metrics and understand where models diverge from desired behavior. We find that hallucination detection metrics, particularly EigenScore, consistently outperform standard diversity and uncertainty quantification metrics, while using only model internals, providing interpretable insights into a model's internal task representations. We demonstrate three applications of GSS: (1) detecting prompt ambiguity and predicting clarification questions for better grounding, (2) interpreting overthinking and underthinking in reasoning models, and (3) steering models to expand their generation space to yield high-quality and diverse outputs.

---

### Multi-Agent Debate for LLM Judges with Adaptive Stability Detection

Authors: Tianyu Hu, Zhen Tan, Song Wang, Huaizhi Qu, Tianlong Chen
Published: 2025-10-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.12697v1

With advancements in reasoning capabilities, Large Language Models (LLMs) are increasingly employed for automated judgment tasks. While LLMs-as-Judges offer promise in automating evaluations, current approaches often rely on simplistic aggregation methods (e.g., majority voting), which can fail even when individual agents provide correct answers. To address this, we propose a multi-agent debate judge framework where agents collaboratively reason and iteratively refine their responses. We formalize the debate process mathematically, analyzing agent interactions and proving that debate amplifies correctness compared to static ensembles. To enhance efficiency, we introduce a stability detection mechanism that models judge consensus dynamics via a time-varying Beta-Binomial mixture, with adaptive stopping based on distributional similarity (Kolmogorov-Smirnov test). This mechanism models the judges' collective correct rate dynamics using a time-varying mixture of Beta-Binomial distributions and employs an adaptive stopping criterion based on distributional similarity (Kolmogorov-Smirnov statistic). Experiments across multiple benchmarks and models demonstrate that our framework improves judgment accuracy over majority voting while maintaining computational efficiency.

---

### ERA: Transforming VLMs into Embodied Agents via Embodied Prior Learning and Online Reinforcement Learning

Authors: Hanyang Chen, Mark Zhao, Rui Yang, Qinwei Ma, Ke Yang, Jiarui Yao, Kangrui Wang, Hao Bai, Zhenhailong Wang, Rui Pan, Mengchao Zhang, Jose Barreiros et al.
Published: 2025-10-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.12693v1

Recent advances in embodied AI highlight the potential of vision language models (VLMs) as agents capable of perception, reasoning, and interaction in complex environments. However, top-performing systems rely on large-scale models that are costly to deploy, while smaller VLMs lack the necessary knowledge and skills to succeed. To bridge this gap, we present \textit{Embodied Reasoning Agent (ERA)}, a two-stage framework that integrates prior knowledge learning and online reinforcement learning (RL). The first stage, \textit{Embodied Prior Learning}, distills foundational knowledge from three types of data: (1) Trajectory-Augmented Priors, which enrich existing trajectory data with structured reasoning generated by stronger models; (2) Environment-Anchored Priors, which provide in-environment knowledge and grounding supervision; and (3) External Knowledge Priors, which transfer general knowledge from out-of-environment datasets. In the second stage, we develop an online RL pipeline that builds on these priors to further enhance agent performance. To overcome the inherent challenges in agent RL, including long horizons, sparse rewards, and training instability, we introduce three key designs: self-summarization for context management, dense reward shaping, and turn-level policy optimization. Extensive experiments on both high-level planning (EB-ALFRED) and low-level control (EB-Manipulation) tasks demonstrate that ERA-3B surpasses both prompting-based large models and previous training-based baselines. Specifically, it achieves overall improvements of 8.4\% on EB-ALFRED and 19.4\% on EB-Manipulation over GPT-4o, and exhibits strong generalization to unseen tasks. Overall, ERA offers a practical path toward scalable embodied intelligence, providing methodological insights for future embodied AI systems.

---

### Who is a Better Matchmaker? Human vs. Algorithmic Judge Assignment in a High-Stakes Startup Competition

Authors: Sarina Xi, Orelia Pi, Miaomiao Zhang, Becca Xiong, Jacqueline Ng Lane, Nihar B. Shah
Published: 2025-10-14
Categories: cs.HC, cs.AI, cs.CL, cs.CY, cs.LG
arXiv: https://arxiv.org/abs/2510.12692v1

There is growing interest in applying artificial intelligence (AI) to automate and support complex decision-making tasks. However, it remains unclear how algorithms compare to human judgment in contexts requiring semantic understanding and domain expertise. We examine this in the context of the judge assignment problem, matching submissions to suitably qualified judges. Specifically, we tackled this problem at the Harvard President's Innovation Challenge, the university's premier venture competition awarding over \$500,000 to student and alumni startups. This represents a real-world environment where high-quality judge assignment is essential. We developed an AI-based judge-assignment algorithm, Hybrid Lexical-Semantic Similarity Ensemble (HLSE), and deployed it at the competition. We then evaluated its performance against human expert assignments using blinded match-quality scores from judges on $309$ judge-venture pairs. Using a Mann-Whitney U statistic based test, we found no statistically significant difference in assignment quality between the two approaches ($AUC=0.48, p=0.40$); on average, algorithmic matches are rated $3.90$ and manual matches $3.94$ on a 5-point scale, where 5 indicates an excellent match. Furthermore, manual assignments that previously required a full week could be automated in several hours by the algorithm during deployment. These results demonstrate that HLSE achieves human-expert-level matching quality while offering greater scalability and efficiency, underscoring the potential of AI-driven solutions to support and enhance human decision-making for judge assignment in high-stakes settings.

---

### DiffEM: Learning from Corrupted Data with Diffusion Models via Expectation Maximization

Authors: Danial Hosseintabar, Fan Chen, Giannis Daras, Antonio Torralba, Constantinos Daskalakis
Published: 2025-10-14
Categories: cs.LG, cs.AI, cs.CV
arXiv: https://arxiv.org/abs/2510.12691v3

Diffusion models have emerged as powerful generative priors for high-dimensional inverse problems, yet learning them when only corrupted or noisy observations are available remains challenging. In this work, we propose a new method for training diffusion models with Expectation-Maximization (EM) from corrupted data. Our proposed method, DiffEM, utilizes conditional diffusion models to reconstruct clean data from observations in the E-step, and then uses the reconstructed data to refine the conditional diffusion model in the M-step. Theoretically, we provide monotonic convergence guarantees for the DiffEM iteration, assuming appropriate statistical conditions. We demonstrate the effectiveness of our approach through experiments on various image reconstruction tasks.

---

### From Delegates to Trustees: How Optimizing for Long-Term Interests Shapes Bias and Alignment in LLM

Authors: Suyash Fulay, Jocelyn Zhu, Michiel Bakker
Published: 2025-10-14
Categories: cs.CY, cs.AI
arXiv: https://arxiv.org/abs/2510.12689v2

Large language models (LLMs) have shown promising accuracy in predicting survey responses and policy preferences, which has increased interest in their potential to represent human interests in various domains. Most existing research has focused on "behavioral cloning", effectively evaluating how well models reproduce individuals' expressed preferences. Drawing on theories of political representation, we highlight an underexplored design trade-off: whether AI systems should act as delegates, mirroring expressed preferences, or as trustees, exercising judgment about what best serves an individual's interests. This trade-off is closely related to issues of LLM sycophancy, where models can encourage behavior or validate beliefs that may be aligned with a user's short-term preferences, but is detrimental to their long-term interests. Through a series of experiments simulating votes on various policy issues in the U.S. context, we apply a temporal utility framework that weighs short and long-term interests (simulating a trustee role) and compare voting outcomes to behavior-cloning models (simulating a delegate). We find that trustee-style predictions weighted toward long-term interests produce policy decisions that align more closely with expert consensus on well-understood issues, but also show greater bias toward models' default stances on topics lacking clear agreement. These findings reveal a fundamental trade-off in designing AI systems to represent human interests. Delegate models better preserve user autonomy but may diverge from well-supported policy positions, while trustee models can promote welfare on well-understood issues yet risk paternalism and bias on subjective topics.

---

### EReLiFM: Evidential Reliability-Aware Residual Flow Meta-Learning for Open-Set Domain Generalization under Noisy Labels

Authors: Kunyu Peng, Di Wen, Kailun Yang, Jia Fu, Yufan Chen, Ruiping Liu, Jiamin Wu, Junwei Zheng, M. Saquib Sarfraz, Luc Van Gool, Danda Pani Paudel, Rainer Stiefelhagen
Published: 2025-10-14
Categories: cs.CV, cs.LG, cs.RO
arXiv: https://arxiv.org/abs/2510.12687v2

Open-Set Domain Generalization (OSDG) aims to enable deep learning models to recognize unseen categories in new domains, which is crucial for real-world applications. Label noise hinders open-set domain generalization by corrupting source-domain knowledge, making it harder to recognize known classes and reject unseen ones. While existing methods address OSDG under Noisy Labels (OSDG-NL) using hyperbolic prototype-guided meta-learning, they struggle to bridge domain gaps, especially with limited clean labeled data. In this paper, we propose Evidential Reliability-Aware Residual Flow Meta-Learning (EReLiFM). We first introduce an unsupervised two-stage evidential loss clustering method to promote label reliability awareness. Then, we propose a residual flow matching mechanism that models structured domain- and category-conditioned residuals, enabling diverse and uncertainty-aware transfer paths beyond interpolation-based augmentation. During this meta-learning process, the model is optimized such that the update direction on the clean set maximizes the loss decrease on the noisy set, using pseudo labels derived from the most confident predicted class for supervision. Experimental results show that EReLiFM outperforms existing methods on OSDG-NL, achieving state-of-the-art performance. The source code is available at https://github.com/KPeng9510/ERELIFM.

---

### Few Shot Semi-Supervised Learning for Abnormal Stop Detection from Sparse GPS Trajectories

Authors: Muhammad Ayub Sabir, Junbiao Pang, Jiaqi Wu, Fatima Ashraf
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12686v1

Abnormal stop detection (ASD) in intercity coach transportation is critical for ensuring passenger safety, operational reliability, and regulatory compliance. However, two key challenges hinder ASD effectiveness: sparse GPS trajectories, which obscure short or unauthorized stops, and limited labeled data, which restricts supervised learning. Existing methods often assume dense sampling or regular movement patterns, limiting their applicability. To address data sparsity, we propose a Sparsity-Aware Segmentation (SAS) method that adaptively defines segment boundaries based on local spatial-temporal density. Building upon these segments, we introduce three domain-specific indicators to capture abnormal stop behaviors. To further mitigate the impact of sparsity, we develop Locally Temporal-Indicator Guided Adjustment (LTIGA), which smooths these indicators via local similarity graphs. To overcome label scarcity, we construct a spatial-temporal graph where each segment is a node with LTIGA-refined features. We apply label propagation to expand weak supervision across the graph, followed by a GCN to learn relational patterns. A final self-training module incorporates high-confidence pseudo-labels to iteratively improve predictions. Experiments on real-world coach data show an AUC of 0.854 and AP of 0.866 using only 10 labeled instances, outperforming prior methods. The code and dataset are publicly available at \href{https://github.com/pangjunbiao/Abnormal-Stop-Detection-SSL.git}

---

### CoRA: Covariate-Aware Adaptation of Time Series Foundation Models

Authors: Guo Qin, Zhi Chen, Yong Liu, Zhiyuan Shi, Haixuan Liu, Xiangdong Huang, Jianmin Wang, Mingsheng Long
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12681v1

Time Series Foundation Models (TSFMs) have shown significant impact through their model capacity, scalability, and zero-shot generalization. However, due to the heterogeneity of inter-variate dependencies and the backbone scalability on large-scale multivariate datasets, most TSFMs are typically pre-trained on univariate time series. This limitation renders them oblivious to crucial information from diverse covariates in real-world forecasting tasks. To further enhance the performance of TSFMs, we propose a general covariate-aware adaptation (CoRA) framework for TSFMs. It leverages pre-trained backbones of foundation models while effectively incorporating exogenous covariates from various modalities, including time series, language, and images, to improve the quality of predictions. Technically, CoRA maintains the equivalence of initialization and parameter consistency during adaptation. With preserved backbones of foundation models as frozen feature extractors, the outcome embeddings from foundation models are empirically demonstrated more informative than raw data. Further, CoRA employs a novel Granger Causality Embedding (GCE) to automatically evaluate covariates regarding their causal predictability with respect to the target variate. We incorporate these weighted embeddings with a zero-initialized condition-injection mechanism, avoiding catastrophic forgetting of pre-trained foundation models and gradually integrates exogenous information. Extensive experiments show that CoRA of TSFMs surpasses state-of-the-art covariate-aware deep forecasters with full or few-shot training samples, achieving 31.1% MSE reduction on covariate-aware forecasting. Compared to other adaptation methods, CoRA exhibits strong compatibility with various advanced TSFMs and extends the scope of covariates to other modalities, presenting a practical paradigm for the application of TSFMs.

---

### Demystifying Hybrid Thinking: Can LLMs Truly Switch Between Think and No-Think?

Authors: Shouren Wang, Wang Yang, Xianxuan Long, Debargha Ganguly, Qifan Wang, Vipin Chaudhary, Xiaotian Han
Published: 2025-10-14
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2510.12680v2

Hybrid thinking enables LLMs to switch between reasoning and direct answering, offering a balance between efficiency and reasoning capability. Yet our experiments reveal that current hybrid thinking LLMs only achieve partial mode separation: reasoning behaviors often leak into the no-think mode. To understand and mitigate this, we analyze the factors influencing controllability and identify four that matter most: (1) larger data scale, (2) using think and no-think answers from different questions rather than the same question, (3) a moderate increase in no-think data number, and (4) a two-phase strategy that first trains reasoning ability and then applies hybrid think training. Building on these findings, we propose a practical recipe that, compared to standard training, can maintain accuracy in both modes while significantly reducing no-think output length (from 1085 to 585 on MATH500) and occurrences of reasoning-supportive tokens such as "wait" (from 5917 to 522 on MATH500). Our findings highlight the limitations of current hybrid thinking and offer directions for strengthening its controllability. The code is available at: https://github.com/SR-A-W/demystifying-hybrid-thinking

---

### MCOP: Multi-UAV Collaborative Occupancy Prediction

Authors: Zefu Lin, Wenbo Chen, Xiaojuan Jin, Yuran Yang, Lue Fan, Yixin Zhang, Yufeng Zhang, Zhaoxiang Zhang
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12679v2

Unmanned Aerial Vehicle (UAV) swarm systems necessitate efficient collaborative perception mechanisms for diverse operational scenarios. Current Bird's Eye View (BEV)-based approaches exhibit two main limitations: bounding-box representations fail to capture complete semantic and geometric information of the scene, and their performance significantly degrades when encountering undefined or occluded objects. To address these limitations, we propose a novel multi-UAV collaborative occupancy prediction framework. Our framework effectively preserves 3D spatial structures and semantics through integrating a Spatial-Aware Feature Encoder and Cross-Agent Feature Integration. To enhance efficiency, we further introduce Altitude-Aware Feature Reduction to compactly represent scene information, along with a Dual-Mask Perceptual Guidance mechanism to adaptively select features and reduce communication overhead. Due to the absence of suitable benchmark datasets, we extend three datasets for evaluation: two virtual datasets (Air-to-Pred-Occ and UAV3D-Occ) and one real-world dataset (GauUScene-Occ). Experiments results demonstrate that our method achieves state-of-the-art accuracy, significantly outperforming existing collaborative methods while reducing communication overhead to only a fraction of previous approaches.

---

### Keep Calm and Avoid Harmful Content: Concept Alignment and Latent Manipulation Towards Safer Answers

Authors: Ruben Belo, Marta Guimaraes, Claudia Soares
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12672v2

Large Language Models are susceptible to jailbreak attacks that bypass built-in safety guardrails (e.g., by tricking the model with adversarial prompts). We propose Concept Alignment and Concept Manipulation CALM, an inference-time method that suppresses harmful concepts by modifying latent representations of the last layer of the model, without retraining. Leveraging concept whitening technique from Computer Vision combined with orthogonal projection, CALM removes unwanted latent directions associated with harmful content while preserving model performance. Experiments show that CALM reduces harmful outputs and outperforms baseline methods in most metrics, offering a lightweight approach to AI safety with no additional training data or model fine-tuning, while incurring only a small computational overhead at inference.

---

### TerraCodec: Compressing Optical Earth Observation Data

Authors: Julen Costa-Watanabe, Isabelle Wittmann, Benedikt Blumenstiel, Konrad Schindler
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12670v3

Earth observation (EO) satellites produce massive streams of multispectral image time series, posing pressing challenges for storage and transmission. Yet, learned EO compression remains fragmented and lacks publicly available, large-scale pretrained codecs. Moreover, prior work has largely focused on image compression, leaving temporal redundancy and EO video codecs underexplored. To address these gaps, we introduce TerraCodec (TEC), a family of learned codecs pretrained on Sentinel-2 EO data. TEC includes efficient multispectral image variants and a Temporal Transformer model (TEC-TT) that leverages dependencies across time. To overcome the fixed-rate setting of today's neural codecs, we present Latent Repacking, a novel method for training flexible-rate transformer models that operate on varying rate-distortion settings. TerraCodec outperforms classical codecs, achieving 3-10x higher compression at equivalent image quality. Beyond compression, TEC-TT enables zero-shot cloud inpainting, surpassing state-of-the-art methods on the AllClear benchmark. Our results establish neural codecs as a promising direction for Earth observation. Our code and models are publically available at https://github.com/IBM/TerraCodec.

---

### Structure-Aware Spectral Sparsification via Uniform Edge Sampling

Authors: Kaiwen He, Petros Drineas, Rajiv Khanna
Published: 2025-10-14
Categories: cs.LG, cs.DS
arXiv: https://arxiv.org/abs/2510.12669v1

Spectral clustering is a fundamental method for graph partitioning, but its reliance on eigenvector computation limits scalability to massive graphs. Classical sparsification methods preserve spectral properties by sampling edges proportionally to their effective resistances, but require expensive preprocessing to estimate these resistances. We study whether uniform edge sampling-a simple, structure-agnostic strategy-can suffice for spectral clustering. Our main result shows that for graphs admitting a well-separated $k$-clustering, characterized by a large structure ratio $Υ(k) = λ_{k+1} / ρ_G(k)$, uniform sampling preserves the spectral subspace used for clustering. Specifically, we prove that uniformly sampling $O(γ^2 n \log n / ε^2)$ edges, where $γ$ is the Laplacian condition number, yields a sparsifier whose top $(n-k)$-dimensional eigenspace is approximately orthogonal to the cluster indicators. This ensures that the spectral embedding remains faithful, and clustering quality is preserved. Our analysis introduces new resistance bounds for intra-cluster edges, a rank-$(n-k)$ effective resistance formulation, and a matrix Chernoff bound adapted to the dominant eigenspace. These tools allow us to bypass importance sampling entirely. Conceptually, our result connects recent coreset-based clustering theory to spectral sparsification, showing that under strong clusterability, even uniform sampling is structure-aware. This provides the first provable guarantee that uniform edge sampling suffices for structure-preserving spectral clustering.

---

### Understanding Parametric Knowledge Injection in Retrieval-Augmented Generation

Authors: Minghao Tang, Shiyu Ni, Jingtong Wu, Zengxin Han, Keping Bi
Published: 2025-10-14
Categories: cs.IR, cs.CL
arXiv: https://arxiv.org/abs/2510.12668v2

Context-grounded generation underpins many LLM applications, including long-document question answering (QA), conversational personalization, and retrieval-augmented generation (RAG). However, classic token-based context concatenation is costly for long inputs and can be lost in the middle at extreme context lengths. Recent work explores context parameterization, which encodes context into lightweight trainable parameters (e.g., LoRA adapters) injected into a frozen LLM. Extending this idea to retrieved evidence yields parametric RAG (P-RAG), which incorporates knowledge via parameter updates rather than token-level attention. In this paper, we present a systematic study of this emerging RAG paradigm-parametric knowledge injection. First, we reassess P-RAG under answer-presence accuracy and show that it does not consistently outperform standard token-based RAG (T-RAG), while combining both (PT-RAG) achieves the best overall performance. Second, we introduce a QA benchmark with up-to-date knowledge beyond the LLM's internal memory to enable controlled analysis. Our representational and mechanistic results indicate that parametric representations capture document-level semantics and primarily influence deeper feed-forward computations, providing high-level guidance but limited evidence consolidation. Finally, we evaluate parametric injection under key RAG challenges, demonstrating improved faithfulness under knowledge conflicts, stronger robustness to retrieval noise, and solid generalization to tasks beyond QA. Our findings clarify the strengths and limitations of parametric RAG and provide practical guidance for future retrieval-augmented LLM systems.

---

### Structured Sparsity and Weight-adaptive Pruning for Memory and Compute efficient Whisper models

Authors: Prasenjit K Mudi, Anshi Sachan, Dahlia Devapriya, Sheetal Kalyani
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12666v1

Whisper models have achieved remarkable progress in speech recognition; yet their large size remains a bottleneck for deployment on resource-constrained edge devices. This paper proposes a framework to design fine-tuned variants of Whisper which address the above problem. Structured sparsity is enforced via the Sparse Group LASSO penalty as a loss regularizer, to reduce the number of FLOating Point operations (FLOPs). Further, a weight statistics aware pruning algorithm is proposed. We also design our custom text normalizer for WER evaluation. On Common Voice 11.0 Hindi dataset, we obtain, without degrading WER, (a) 35.4% reduction in model parameters, 14.25% lower memory consumption and 18.5% fewer FLOPs on Whisper-small, and (b) 31% reduction in model parameters, 15.29% lower memory consumption and 16.95% fewer FLOPs on Whisper-medium; and, (c) substantially outperform the state-of-the-art Iterative Magnitude Pruning based method by pruning 18.7% more parameters along with a 12.31 reduction in WER.

---

### On the Use of Hierarchical Vision Foundation Models for Low-Cost Human Mesh Recovery and Pose Estimation

Authors: Shuhei Tarashima, Yushan Wang, Norio Tagawa
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12660v2

In this work, we aim to develop simple and efficient models for human mesh recovery (HMR) and its predecessor task, human pose estimation (HPE). State-of-the-art HMR methods, such as HMR2.0 and its successors, rely on large, non-hierarchical vision transformers as encoders, which are inherited from the corresponding HPE models like ViTPose. To establish baselines across varying computational budgets, we first construct three lightweight HMR2.0 variants by adapting the corresponding ViTPose models. In addition, we propose leveraging the early stages of hierarchical vision foundation models (VFMs), including Swin Transformer, GroupMixFormer, and VMamba, as encoders. This design is motivated by the observation that intermediate stages of hierarchical VFMs produce feature maps with resolutions comparable to or higher than those of non-hierarchical counterparts. We conduct a comprehensive evaluation of 27 hierarchical-VFM-based HMR and HPE models, demonstrating that using only the first two or three stages achieves performance on par with full-stage models. Moreover, we show that the resulting truncated models exhibit better trade-offs between accuracy and computational efficiency compared to existing lightweight alternatives. The source code is available at https://github.com/nttcom/TruncHierVFM.

---

### SG-XDEAT: Sparsity-Guided Cross-Dimensional and Cross-Encoding Attention with Target-Aware Conditioning in Tabular Learning

Authors: Chih-Chuan Cheng, Yi-Ju Tseng
Published: 2025-10-14
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.12659v1

We propose SG-XDEAT (Sparsity-Guided Cross Dimensional and Cross-Encoding Attention with Target Aware Conditioning), a novel framework designed for supervised learning on tabular data. At its core, SG-XDEAT employs a dual-stream encoder that decomposes each input feature into two parallel representations: a raw value stream and a target-conditioned (label-aware) stream. These dual representations are then propagated through a hierarchical stack of attention-based modules. SG-XDEAT integrates three key components: (i) Cross-Dimensional self-attention, which captures intra-view dependencies among features within each stream; (ii) Cross-Encoding self-attention, which enables bidirectional interaction between raw and target-aware representations; and (iii) an Adaptive Sparse Self-Attention (ASSA) mechanism, which dynamically suppresses low-utility tokens by driving their attention weights toward zero--thereby mitigating the impact of noise. Empirical results on multiple public benchmarks show consistent gains over strong baselines, confirming that jointly modeling raw and target-aware views--while adaptively filtering noise--yields a more robust deep tabular learner.

---

### Towards Foundation Inference Models that Learn ODEs In-Context

Authors: Maximilian Mauel, Manuel Hinz, Patrick Seifner, David Berghaus, Ramses J. Sanchez
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12650v1

Ordinary differential equations (ODEs) describe dynamical systems evolving deterministically in continuous time. Accurate data-driven modeling of systems as ODEs, a central problem across the natural sciences, remains challenging, especially if the data is sparse or noisy. We introduce FIM-ODE (Foundation Inference Model for ODEs), a pretrained neural model designed to estimate ODEs zero-shot (i.e., in context) from sparse and noisy observations. Trained on synthetic data, the model utilizes a flexible neural operator for robust ODE inference, even from corrupted data. We empirically verify that FIM-ODE provides accurate estimates, on par with a neural state-of-the-art method, and qualitatively compare the structure of their estimated vector fields.

---

### Zero-Shot CFC: Fast Real-World Image Denoising based on Cross-Frequency Consistency

Authors: Yanlin Jiang, Yuchen Liu, Mingren Liu
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12646v1

Zero-shot denoisers address the dataset dependency of deep-learning-based denoisers, enabling the denoising of unseen single images. Nonetheless, existing zero-shot methods suffer from long training times and rely on the assumption of noise independence and a zero-mean property, limiting their effectiveness in real-world denoising scenarios where noise characteristics are more complicated. This paper proposes an efficient and effective method for real-world denoising, the Zero-Shot denoiser based on Cross-Frequency Consistency (ZSCFC), which enables training and denoising with a single noisy image and does not rely on assumptions about noise distribution. Specifically, image textures exhibit position similarity and content consistency across different frequency bands, while noise does not. Based on this property, we developed cross-frequency consistency loss and an ultralight network to realize image denoising. Experiments on various real-world image datasets demonstrate that our ZSCFC outperforms other state-of-the-art zero-shot methods in terms of computational efficiency and denoising performance.

---

### Reasoning Pattern Matters: Learning to Reason without Human Rationales

Authors: Chaoxu Pang, Yixuan Cao, Ping Luo
Published: 2025-10-14
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.12643v1

Large Language Models (LLMs) have demonstrated remarkable reasoning capabilities under the widely adopted SFT+RLVR paradigm, which first performs Supervised Fine-Tuning (SFT) on human-annotated reasoning trajectories (rationales) to establish initial reasoning behaviors, then applies Reinforcement Learning with Verifiable Rewards (RLVR) to optimize the model using verifiable signals without golden rationales. However, annotating high-quality rationales for the SFT stage remains prohibitively expensive. This paper investigates when and how rationale annotation costs can be substantially reduced without compromising reasoning performance. We identify a broad class of problems, termed patterned reasoning tasks, where reasoning follows a fixed, procedural strategy consistent across instances. Although instances vary in content such as domain knowledge, factual information, or numeric values, the solution derives from applying a shared reasoning pattern. We argue that the success of SFT+RLVR on such tasks primarily stems from its ability to enable models to internalize these reasoning patterns. Using numerical semantic matching as a representative task, we provide both causal and behavioral evidence showing that reasoning patterns rather than the quantity or quality of rationales are the key determinant of performance. Building on these insights, we propose Pattern-Aware LLMs as Rationale AnnOtators (PARO), a simple yet effective framework that enables LLMs to generate rationales aligned with task-specific reasoning patterns without requiring human rationale annotations. Experiments show that PARO-generated rationales achieve comparable SFT+RLVR performance to human rationales that are 10 times larger. These results suggest that large-scale human rationale annotations can be replaced with LLM-based automatic annotations requiring only limited human supervision over reasoning patterns.

---

### Aixel: A Unified, Adaptive and Extensible System for AI-powered Data Analysis

Authors: Meihui Zhang, Liming Wang, Chi Zhang, Zhaojing Luo
Published: 2025-10-14
Categories: cs.DB, cs.AI
arXiv: https://arxiv.org/abs/2510.12642v1

A growing trend in modern data analysis is the integration of data management with learning, guided by accuracy, latency, and cost requirements. In practice, applications draw data of different formats from many sources. In the meanwhile, the objectives and budgets change over time. Existing systems handle these applications across databases, analysis libraries, and tuning services. Such fragmentation leads to complex user interaction, limited adaptability, suboptimal performance, and poor extensibility across components. To address these challenges, we present Aixel, a unified, adaptive, and extensible system for AI-powered data analysis. The system organizes work across four layers: application, task, model, and data. The task layer provides a declarative interface to capture user intent, which is parsed into an executable operator plan. An optimizer compiles and schedules this plan to meet specified goals in accuracy, latency, and cost. The task layer coordinates the execution of data and model operators, with built-in support for reuse and caching to improve efficiency. The model layer offers versioned storage for index, metadata, tensors, and model artifacts. It supports adaptive construction, task-aligned drift detection, and safe updates that reuse shared components. The data layer provides unified data management capabilities, including indexing, constraint-aware discovery, task-aligned selection, and comprehensive feature management. With the above designed layers, Aixel delivers a user friendly, adaptive, efficient, and extensible system.

---

### On Foundation Models for Temporal Point Processes to Accelerate Scientific Discovery

Authors: David Berghaus, Patrick Seifner, Kostadin Cvejoski, Ramses J. Sanchez
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12640v2

Many scientific fields, from medicine to seismology, rely on analyzing sequences of events over time to understand complex systems. Traditionally, machine learning models must be built and trained from scratch for each new dataset, which is a slow and costly process. We introduce a new approach: a single, powerful model that learns the underlying patterns of event data in context. We trained this "foundation model" on millions of simulated event sequences, teaching it a general-purpose understanding of how events can unfold. As a result, our model can analyze new scientific data instantly, without retraining, simply by looking at a few examples from the dataset. It can also be quickly fine-tuned for even higher accuracy. This approach makes sophisticated event analysis more accessible and accelerates the pace of scientific discovery.

---

### Thermodynamic structure of the Sinkhorn flow

Authors: Anand Srinivasan, Jean-Jacques Slotine
Published: 2025-10-14
Categories: stat.ML, cs.LG, math.PR
arXiv: https://arxiv.org/abs/2510.12639v2

Entropy-regularized optimal transport, which has strong links to the Schrödinger bridge problem in statistical mechanics, enjoys a variety of applications from trajectory inference to generative modeling. A major driver of renewed interest in this problem is the recent development of fast matrix-scaling algorithms\textemdash known as iterative proportional fitting or the Sinkhorn algorithm\textemdash for entropic optimal transport, which have favorable complexity over traditional approaches to the unregularized problem. Here, we take a perspective on this algorithm rooted in the thermodynamic origins of Schrödinger's problem and inspired by the modern geometric theory of diffusion: is the Sinkhorn flow (viewed in continuous-time as a mirror descent by recent results) the gradient flow of entropy in a formal Riemannian geometry? We answer this question affirmatively, finding a nonlocal Wasserstein gradient structure in the dynamics of its free marginal. This offers a physical interpretation of the Sinkhorn flow as the stochastic dynamics of a particle with law evolving by the nonlocal diffusion of a chemical potential. Simultaneously, it brings a standard suite of functional inequalities characterizing Markov diffusion processes to bear upon its geometry and convergence. We prove an entropy-energy (de Bruijn) identity, a Poincaré inequality, and a Bakry-Émery-type condition under which a logarithmic Sobolev inequality (LSI) holds and implies exponential convergence of the Sinkhorn flow in entropy. We lastly discuss computational applications such as stopping heuristics and latent-space design criteria leveraging the LSI and, returning to the physical interpretation, the possibility of natural systems whose relaxation to equilibrium inherently solves entropic optimal transport or Schrödinger bridge problems.

---

### Expert or not? assessing data quality in offline reinforcement learning

Authors: Arip Asadulaev, Fakhri Karray, Martin Takac
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12638v1

Offline reinforcement learning (RL) learns exclusively from static datasets, without further interaction with the environment. In practice, such datasets vary widely in quality, often mixing expert, suboptimal, and even random trajectories. The choice of algorithm therefore depends on dataset fidelity. Behavior cloning can suffice on high-quality data, whereas mixed- or low-quality data typically benefits from offline RL methods that stitch useful behavior across trajectories. Yet in the wild it is difficult to assess dataset quality a priori because the data's provenance and skill composition are unknown. We address the problem of estimating offline dataset quality without training an agent. We study a spectrum of proxies from simple cumulative rewards to learned value based estimators, and introduce the Bellman Wasserstein distance (BWD), a value aware optimal transport score that measures how dissimilar a dataset's behavioral policy is from a random reference policy. BWD is computed from a behavioral critic and a state conditional OT formulation, requiring no environment interaction or full policy optimization. Across D4RL MuJoCo tasks, BWD strongly correlates with an oracle performance score that aggregates multiple offline RL algorithms, enabling efficient prediction of how well standard agents will perform on a given dataset. Beyond prediction, integrating BWD as a regularizer during policy optimization explicitly pushes the learned policy away from random behavior and improves returns. These results indicate that value aware, distributional signals such as BWD are practical tools for triaging offline RL datasets and policy optimization.

---

### COSTAR-A: A prompting framework for enhancing Large Language Model performance on Point-of-View questions

Authors: Nzubechukwu C. Ohalete, Kevin B. Gittner, Lauren M. Matheny
Published: 2025-10-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.12637v1

Large Language Models (LLMs) are highly sensitive to prompt design, and making optimized prompting techniques is crucial for generating consistent, high-quality outputs. In this study, we introduce COSTAR-A, a novel prompt engineering framework that enhances the existing COSTAR method, which stands for Context, Objective, Style, Tone, Audience, and Response, by adding the 'Answer' component at the end. We demonstrate that while the original COSTAR framework improves prompt clarity and aligns outputs for larger LLMs, its performance is less consistent with smaller, locally optimized models, particularly in tasks that require more directive or constrained outputs. Through a series of controlled prompt-output assessments with smaller (at most 8 billion parameters), fine-tuned models, we found that COSTAR-A can enhance the output structure and decisiveness of localized LLMs for certain tasks, although its effectiveness varies across models and use cases. Notably, the Llama 3.1-8B model exhibited performance improvements when prompted with COSTAR-A compared to COSTAR alone. These findings emphasize the adaptability and scalability of COSTAR-A as a prompting framework, particularly in computationally efficient AI deployments on resource-constrained hardware.

---

### Adapting Noise to Data: Generative Flows from 1D Processes

Authors: Jannis Chemseddine, Gregor Kornhardt, Richard Duong, Gabriele Steidl
Published: 2025-10-14
Categories: stat.ML, cs.LG, math.AP
arXiv: https://arxiv.org/abs/2510.12636v5

The default Gaussian latent in flow-based generative models poses challenges when learning certain distributions such as heavy-tailed ones. We introduce a general framework for learning data-adaptive parametric prior distributions (latent noise) using one-dimensional quantile functions, optimized via the Wasserstein distance between noise and data. The quantile-based prior parameterization naturally adapts to both heavy-tailed and compactly supported distributions and shortens transport paths. Numerical results on heavy-tailed weather and image datasets confirm the method's flexibility and effectiveness achieved with negligible computational overhead.

---

### Memory as Action: Autonomous Context Curation for Long-Horizon Agentic Tasks

Authors: Yuxiang Zhang, Jiangming Shu, Ye Ma, Xueyuan Lin, Shangxi Wu, Jitao Sang
Published: 2025-10-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.12635v3

Long-context Large Language Models, despite their expanded capacity, require careful working memory management to mitigate attention dilution during long-horizon tasks. Yet existing approaches rely on external mechanisms that lack awareness of the agent's reasoning state, leading to suboptimal decisions. We propose Memory-as-Action (MemAct), a framework that treats working memory management as learnable policy actions. By formulating context management as in-place editing operations (deletion, insertion), MemAct enables joint optimization of information retention and task performance through end-to-end reinforcement learning. To address the computational challenges of dynamic context updates, we introduce Dynamic Context Policy Optimization, which restores training efficiency without compromising reasoning integrity. Experiments show that MemAct-RL-14B matches the accuracy of models $16\times$ larger while reducing average context length by 51\%, with learned strategies that adapt to model capabilities and generalize across task complexities.

---

### Laminar: A Scalable Asynchronous RL Post-Training Framework

Authors: Guangming Sheng, Yuxuan Tong, Borui Wan, Wang Zhang, Chaobo Jia, Xibin Wu, Yuqi Wu, Xiang Li, Chi Zhang, Yanghua Peng, Haibin Lin, Xin Liu et al.
Published: 2025-10-14
Categories: cs.LG, cs.AI, cs.DC
arXiv: https://arxiv.org/abs/2510.12633v1

Reinforcement learning (RL) post-training for Large Language Models (LLMs) is now scaling to large clusters and running for extended durations to enhance model reasoning performance. However, the scalability of existing RL frameworks is limited, as extreme long-tail skewness in RL trajectory generation causes severe GPU underutilization. Current asynchronous RL systems attempt to mitigate this, but they rely on global weight synchronization between the actor and all rollouts, which creates a rigid model update schedule. This global synchronization is ill-suited for the highly skewed and evolving distribution of trajectory generation latency in RL training, crippling training efficiency. Our key insight is that efficient scaling requires breaking this lockstep through trajectory-level asynchrony, which generates and consumes each trajectory independently. We propose Laminar, a scalable and robust RL post-training system built on a fully decoupled architecture. First, we replace global updates with a tier of relay workers acting as a distributed parameter service. This enables asynchronous and fine-grained weight synchronization, allowing rollouts to pull the latest weight anytime without stalling the actor's training loop. Second, a dynamic repack mechanism consolidates long-tail trajectories onto a few dedicated rollouts, maximizing generation throughput. The fully decoupled design also isolates failures, ensuring robustness for long-running jobs. Our evaluation on a 1024-GPU cluster shows that Laminar achieves up to 5.48$\times$ training throughput speedup over state-of-the-art systems, while reducing model convergence time.

---

### Designing Tools with Control Confidence

Authors: Ajith Anil Meera, Abian Torres, Pablo Lanillos
Published: 2025-10-14
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2510.12630v1

Prehistoric humans invented stone tools for specialized tasks by not just maximizing the tool's immediate goal-completion accuracy, but also increasing their confidence in the tool for later use under similar settings. This factor contributed to the increased robustness of the tool, i.e., the least performance deviations under environmental uncertainties. However, the current autonomous tool design frameworks solely rely on performance optimization, without considering the agent's confidence in tool use for repeated use. Here, we take a step towards filling this gap by i) defining an optimization framework for task-conditioned autonomous hand tool design for robots, where ii) we introduce a neuro-inspired control confidence term into the optimization routine that helps the agent to design tools with higher robustness. Through rigorous simulations using a robotic arm, we show that tools designed with control confidence as the objective function are more robust to environmental uncertainties during tool use than a pure accuracy-driven objective. We further show that adding control confidence to the objective function for tool design provides a balance between the robustness and goal accuracy of the designed tools under control perturbations. Finally, we show that our CMAES-based evolutionary optimization strategy for autonomous tool design outperforms other state-of-the-art optimizers by designing the optimal tool within the fewest iterations. Code: https://github.com/ajitham123/Tool_design_control_confidence.

---

### Learning-To-Measure: In-Context Active Feature Acquisition

Authors: Yuta Kobayashi, Zilin Jing, Jiayu Yao, Hongseok Namkoong, Shalmali Joshi
Published: 2025-10-14
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.12624v2

Active feature acquisition (AFA) is a sequential decision-making problem where the goal is to improve model performance for test instances by adaptively selecting which features to acquire. In practice, AFA methods often learn from retrospective data with systematic missingness in the features and limited task-specific labels. Most prior work addresses acquisition for a single predetermined task, limiting scalability. To address this limitation, we formalize the meta-AFA problem, where the goal is to learn acquisition policies across various tasks. We introduce Learning-to-Measure (L2M), which consists of i) reliable uncertainty quantification over unseen tasks, and ii) an uncertainty-guided greedy feature acquisition agent that maximizes conditional mutual information. We demonstrate a sequence-modeling or autoregressive pre-training approach that underpins reliable uncertainty quantification for tasks with arbitrary missingness. L2M operates directly on datasets with retrospective missingness and performs the meta-AFA task in-context, eliminating per-task retraining. Across synthetic and real-world tabular benchmarks, L2M matches or surpasses task-specific baselines, particularly under scarce labels and high missingness.

---

### ACADATA: Parallel Dataset of Academic Data for Machine Translation

Authors: Iñaki Lacunza, Javier Garcia Gilabert, Francesca De Luca Fornaciari, Javier Aula-Blasco, Aitor Gonzalez-Agirre, Maite Melero, Marta Villegas
Published: 2025-10-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.12621v2

We present ACADATA, a high-quality parallel dataset for academic translation, that consists of two subsets: ACAD-TRAIN, which contains approximately 1.5 million author-generated paragraph pairs across 96 language directions and ACAD-BENCH, a curated evaluation set of almost 6,000 translations covering 12 directions. To validate its utility, we fine-tune two Large Language Models (LLMs) on ACAD-TRAIN and benchmark them on ACAD-BENCH against specialized machine-translation systems, general-purpose, open-weight LLMs, and several large-scale proprietary models. Experimental results demonstrate that fine-tuning on ACAD-TRAIN leads to improvements in academic translation quality by +6.1 and +12.4 d-BLEU points on average for 7B and 2B models respectively, while also improving long-context translation in a general domain by up to 24.9% when translating out of English. The fine-tuned top-performing model surpasses the best propietary and open-weight models on academic translation domain. By releasing ACAD-TRAIN, ACAD-BENCH and the fine-tuned models, we provide the community with a valuable resource to advance research in academic domain and long-context translation.

---

### Towards Fast Coarse-graining and Equation Discovery with Foundation Inference Models

Authors: Manuel Hinz, Maximilian Mauel, Patrick Seifner, David Berghaus, Kostadin Cvejoski, Ramses J. Sanchez
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12618v2

High-dimensional recordings of dynamical processes are often characterized by a much smaller set of effective variables, evolving on low-dimensional manifolds. Identifying these latent dynamics requires solving two intertwined problems: discovering appropriate coarse-grained variables and simultaneously fitting the governing equations. Most machine learning approaches tackle these tasks jointly by training autoencoders together with models that enforce dynamical consistency. We propose to decouple the two problems by leveraging the recently introduced Foundation Inference Models (FIMs). FIMs are pretrained models that estimate the infinitesimal generators of dynamical systems (e.g., the drift and diffusion of a stochastic differential equation) in zero-shot mode. By amortizing the inference of the dynamics through a FIM with frozen weights, and training only the encoder-decoder map, we define a simple, simulation-consistent loss that stabilizes representation learning. A proof of concept on a stochastic double-well system with semicircle diffusion, embedded into synthetic video data, illustrates the potential of this approach for fast and reusable coarse-graining pipelines.

---

### Same model, better performance: the impact of shuffling on DNA Language Models benchmarking

Authors: Davide Greco, Konrad Rawlik
Published: 2025-10-14
Categories: q-bio.GN, cs.LG
arXiv: https://arxiv.org/abs/2510.12617v2

Large Language Models are increasingly popular in genomics due to their potential to decode complex biological sequences. Hence, researchers require a standardized benchmark to evaluate DNA Language Models (DNA LMs) capabilities. However, evaluating DNA LMs is a complex task that intersects genomic's domain-specific challenges and machine learning methodologies, where seemingly minor implementation details can significantly compromise benchmark validity. We demonstrate this through BEND (Benchmarking DNA Language Models), where hardware-dependent hyperparameters -- number of data loading workers and buffer sizes -- create spurious performance variations of up to 4% for identical models. The problem stems from inadequate data shuffling interacting with domain specific data characteristics. Experiments with three DNA language models (HyenaDNA, DNABERT-2, ResNet-LM) show these artifacts affect both absolute performance and relative model rankings. We propose a simple solution: pre-shuffling data before storage eliminates hardware dependencies while maintaining efficiency. This work highlights how standard ML practices can interact unexpectedly with domain-specific data characteristics, with broader implications for benchmark design in specialized domains.

---

### A Functional Perspective on Knowledge Distillation in Neural Networks

Authors: Israel Mason-Williams, Gabryel Mason-Williams, Helen Yannakoudakis
Published: 2025-10-14
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.12615v2

Knowledge distillation is considered a compression mechanism when judged on the resulting student's accuracy and loss, yet its functional impact is poorly understood. We quantify the compression capacity of knowledge distillation and the resulting knowledge transfer from a functional perspective, decoupling compression from architectural reduction to provide an improved understanding of knowledge distillation. We employ a control-driven experimental protocol with hypothesis testing and random control distillation to isolate and understand knowledge transfer mechanisms across data modalities. To test the breadth and limits of our analyses, we study self-distillation, standard distillation, feature-map matching variants, distillation scaling laws across model sizes, and the impact of temperature on knowledge transfer. We find statistically supported knowledge transfer in some modalities and architectures; however, the extent of this transfer is less pronounced than anticipated, even under conditions that maximise knowledge sharing. Notably, in cases of significant functional transfer, we identify a consistent and severe asymmetric transfer of negative knowledge to the student, raising safety concerns for knowledge distillation. Across 22 experimental setups, 9 architectures, and 7 datasets, our results suggest that knowledge distillation functions less as a robust compression-by-transfer mechanism and more as a data-dependent regulariser whose transfer component is biased towards negative asymmetric transfer.

---

### StyleDecipher: Robust and Explainable Detection of LLM-Generated Texts with Stylistic Analysis

Authors: Siyuan Li, Aodu Wulianghai, Xi Lin, Guangyan Li, Xiang Chen, Jun Wu, Jianhua Li
Published: 2025-10-14
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.12608v1

With the increasing integration of large language models (LLMs) into open-domain writing, detecting machine-generated text has become a critical task for ensuring content authenticity and trust. Existing approaches rely on statistical discrepancies or model-specific heuristics to distinguish between LLM-generated and human-written text. However, these methods struggle in real-world scenarios due to limited generalization, vulnerability to paraphrasing, and lack of explainability, particularly when facing stylistic diversity or hybrid human-AI authorship. In this work, we propose StyleDecipher, a robust and explainable detection framework that revisits LLM-generated text detection using combined feature extractors to quantify stylistic differences. By jointly modeling discrete stylistic indicators and continuous stylistic representations derived from semantic embeddings, StyleDecipher captures distinctive style-level divergences between human and LLM outputs within a unified representation space. This framework enables accurate, explainable, and domain-agnostic detection without requiring access to model internals or labeled segments. Extensive experiments across five diverse domains, including news, code, essays, reviews, and academic abstracts, demonstrate that StyleDecipher consistently achieves state-of-the-art in-domain accuracy. Moreover, in cross-domain evaluations, it surpasses existing baselines by up to 36.30%, while maintaining robustness against adversarial perturbations and mixed human-AI content. Further qualitative and quantitative analysis confirms that stylistic signals provide explainable evidence for distinguishing machine-generated text. Our source code can be accessed at https://github.com/SiyuanLi00/StyleDecipher.

---

### WaterFlow: Explicit Physics-Prior Rectified Flow for Underwater Saliency Mask Generation

Authors: Runting Li, Shijie Lian, Hua Li, Yutong Li, Wenhui Wu, Sam Kwong
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12605v2

Underwater Salient Object Detection (USOD) faces significant challenges, including underwater image quality degradation and domain gaps. Existing methods tend to ignore the physical principles of underwater imaging or simply treat degradation phenomena in underwater images as interference factors that must be eliminated, failing to fully exploit the valuable information they contain. We propose WaterFlow, a rectified flow-based framework for underwater salient object detection that innovatively incorporates underwater physical imaging information as explicit priors directly into the network training process and introduces temporal dimension modeling, significantly enhancing the model's capability for salient object identification. On the USOD10K dataset, WaterFlow achieves a 0.072 gain in S_m, demonstrating the effectiveness and superiority of our method. https://github.com/Theo-polis/WaterFlow.

---

### COINS: SemantiC Ids Enhanced COLd Item RepresentatioN for Click-through Rate Prediction in E-commerce Search

Authors: Qihang Zhao, Zhongbo Sun, Xiaoyang Zheng, Xian Guo, Siyuan Wang, Zihan Liang, Mingcan Peng, Ben Chen, Chenyi Lei
Published: 2025-10-14
Categories: cs.IR, cs.AI
arXiv: https://arxiv.org/abs/2510.12604v4

With the rise of modern search and recommendation platforms, insufficient collaborative information of cold-start items exacerbates the Matthew effect of existing platform items, challenging platform diversity and becoming a longstanding issue. Existing methods align items' side content with collaborative information to transfer collaborative signals from high-popularity items to cold-start items. However, these methods fail to account for the asymmetry between collaboration and content, nor the fine-grained differences among items. To address these issues, we propose COINS, an item representation enhancement approach based on fused alignment of semantic IDs. Specifically, we use RQ-OPQ encoding to quantize item content and collaborative information, followed by a two-step alignment: RQ encoding transfers shared collaborative signals across items, while OPQ encoding learns differentiated information of items. Comprehensive offline experiments on large-scale industrial datasets demonstrate superiority of COINS, and rigorous online A/B tests confirm statistically significant improvements: item CTR +1.66%, buyers +1.57%, and order volume +2.17%.

---

### Reasoning in the Dark: Interleaved Vision-Text Reasoning in Latent Space

Authors: Chao Chen, Zhixin Ma, Yongqi Li, Yupeng Hu, Yinwei Wei, Wenjie Li, Liqiang Nie
Published: 2025-10-14
Categories: cs.CV, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2510.12603v2

Multimodal reasoning aims to enhance the capabilities of MLLMs by incorporating intermediate reasoning steps before reaching the final answer. It has evolved from text-only reasoning to the integration of visual information, enabling the thought process to be conveyed through both images and text. Despite its effectiveness, current multimodal reasoning methods depend on explicit reasoning steps that require labor-intensive vision-text annotations and inherently introduce significant inference latency. To address these issues, we introduce multimodal latent reasoning with the advantages of multimodal representation, reduced annotation, and inference efficiency. To facilitate it, we propose Interleaved Vision-Text Latent Reasoning (IVT-LR), which injects both visual and textual information in the reasoning process within the latent space. Specifically, IVT-LR represents each reasoning step by combining two implicit parts: latent text (the hidden states from the previous step) and latent vision (a set of selected image embeddings). We further introduce a progressive multi-stage training strategy to enable MLLMs to perform the above multimodal latent reasoning steps. Experiments on M$^3$CoT and ScienceQA demonstrate that our IVT-LR method achieves an average performance increase of 5.45\% in accuracy, while simultaneously achieving a speed increase of over 5 times compared to existing approaches.

---

### Research in Collaborative Learning Does Not Serve Cross-Silo Federated Learning in Practice

Authors: Kevin Kuo, Chhavi Yadav, Virginia Smith
Published: 2025-10-14
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.12595v1

Cross-silo federated learning (FL) is a promising approach to enable cross-organization collaboration in machine learning model development without directly sharing private data. Despite growing organizational interest driven by data protection regulations such as GDPR and HIPAA, the adoption of cross-silo FL remains limited in practice. In this paper, we conduct an interview study to understand the practical challenges associated with cross-silo FL adoption. With interviews spanning a diverse set of stakeholders such as user organizations, software providers, and academic researchers, we uncover various barriers, from concerns about model performance to questions of incentives and trust between participating organizations. Our study shows that cross-silo FL faces a set of challenges that have yet to be well-captured by existing research in the area and are quite distinct from other forms of federated learning such as cross-device FL. We end with a discussion on future research directions that can help overcome these challenges.

---

### Teaching Language Models to Faithfully Express their Uncertainty

Authors: Bryan Eikema, Evgenia Ilia, José G. C. de Souza, Chrysoula Zerva, Wilker Aziz
Published: 2025-10-14
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.12587v1

Large language models (LLMs) often miscommunicate their uncertainty: repeated queries can produce divergent answers, yet generated responses are typically unhedged or hedged in ways that do not reflect this variability. This conveys unfaithful information about the uncertain state of the LLMs' knowledge, creating a faithfulness gap that affects even strong LLMs. We introduce Faithful Uncertainty Tuning (FUT): a fine-tuning approach that teaches instruction-tuned LLMs to express uncertainty faithfully without altering their underlying answer distribution. We construct training data by augmenting model samples with uncertainty hedges (i.e. verbal cues such as 'possibly' or 'likely') aligned with sample consistency, requiring no supervision beyond the model and a set of prompts. We evaluate FUT on open-domain question answering (QA) across multiple models and datasets. Our results show that FUT substantially reduces the faithfulness gap, while preserving QA accuracy and introducing minimal semantic distribution shift. Further analyses demonstrate robustness across decoding strategies, choice of hedgers, and other forms of uncertainty expression (i.e. numerical). These findings establish FUT as a simple and effective way to teach LLMs to communicate uncertainty faithfully.

---

### There is No VAE: End-to-End Pixel-Space Generative Modeling via Self-Supervised Pre-training

Authors: Jiachen Lei, Keli Liu, Julius Berner, Haiming Yu, Hongkai Zheng, Jiahong Wu, Xiangxiang Chu
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12586v3

Pixel-space generative models are often more difficult to train and generally underperform compared to their latent-space counterparts, leaving a persistent performance and efficiency gap. In this paper, we introduce a novel two-stage training framework that closes this gap for pixel-space diffusion and consistency models. In the first stage, we pre-train encoders to capture meaningful semantics from clean images while aligning them with points along the same deterministic sampling trajectory, which evolves points from the prior to the data distribution. In the second stage, we integrate the encoder with a randomly initialized decoder and fine-tune the complete model end-to-end for both diffusion and consistency models. Our framework achieves state-of-the-art (SOTA) performance on ImageNet. Specifically, our diffusion model reaches an FID of 1.58 on ImageNet-256 and 2.35 on ImageNet-512 with 75 number of function evaluations (NFE) surpassing prior pixel-space methods and VAE-based counterparts by a large margin in both generation quality and training efficiency. In a direct comparison, our model significantly outperforms DiT while using only around 30\% of its training compute. Furthermore, our consistency model achieves an impressive FID of 8.82 on ImageNet-256, significantly outperforming its latent-space counterparts. This marks the first successful training of a consistency model directly on high-resolution images without relying on pre-trained VAEs or diffusion models. Our codes are available at: \href{https://github.com/AMAP-ML/EPG}{https://github.com/AMAP-ML/EPG}

---

### LayerSync: Self-aligning Intermediate Layers

Authors: Yasaman Haghighi, Bastien van Delft, Mariam Hassan, Alexandre Alahi
Published: 2025-10-14
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2510.12581v2

We propose LayerSync, a domain-agnostic approach for improving the generation quality and the training efficiency of diffusion models. Prior studies have highlighted the connection between the quality of generation and the representations learned by diffusion models, showing that external guidance on model intermediate representations accelerates training. We reconceptualize this paradigm by regularizing diffusion models with their own intermediate representations. Building on the observation that representation quality varies across diffusion model layers, we show that the most semantically rich representations can act as an intrinsic guidance for weaker ones, reducing the need for external supervision. Our approach, LayerSync, is a self-sufficient, plug-and-play regularizer term with no overhead on diffusion model training and generalizes beyond the visual domain to other modalities. LayerSync requires no pretrained models nor additional data. We extensively evaluate the method on image generation and demonstrate its applicability to other domains such as audio, video, and motion generation. We show that it consistently improves the generation quality and the training efficiency. For example, we speed up the training of flow-based transformer by over 8.75x on ImageNet dataset and improved the generation quality by 23.6%. The code is available at https://github.com/vita-epfl/LayerSync.

---

### Unlocking Zero-Shot Plant Segmentation with Pl@ntNet Intelligence

Authors: Simon Ravé, Jean-Christophe Lombardo, Pejman Rasti, Alexis Joly, David Rousseau
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12579v1

We present a zero-shot segmentation approach for agricultural imagery that leverages Plantnet, a large-scale plant classification model, in conjunction with its DinoV2 backbone and the Segment Anything Model (SAM). Rather than collecting and annotating new datasets, our method exploits Plantnet's specialized plant representations to identify plant regions and produce coarse segmentation masks. These masks are then refined by SAM to yield detailed segmentations. We evaluate on four publicly available datasets of various complexity in terms of contrast including some where the limited size of the training data and complex field conditions often hinder purely supervised methods. Our results show consistent performance gains when using Plantnet-fine-tuned DinoV2 over the base DinoV2 model, as measured by the Jaccard Index (IoU). These findings highlight the potential of combining foundation models with specialized plant-centric models to alleviate the annotation bottleneck and enable effective segmentation in diverse agricultural scenarios.

---

### Learning Human Motion with Temporally Conditional Mamba

Authors: Quang Nguyen, Tri Le, Baoru Huang, Minh Nhat Vu, Ngan Le, Thieu Vo, Anh Nguyen
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12573v1

Learning human motion based on a time-dependent input signal presents a challenging yet impactful task with various applications. The goal of this task is to generate or estimate human movement that consistently reflects the temporal patterns of conditioning inputs. Existing methods typically rely on cross-attention mechanisms to fuse the condition with motion. However, this approach primarily captures global interactions and struggles to maintain step-by-step temporal alignment. To address this limitation, we introduce Temporally Conditional Mamba, a new mamba-based model for human motion generation. Our approach integrates conditional information into the recurrent dynamics of the Mamba block, enabling better temporally aligned motion. To validate the effectiveness of our method, we evaluate it on a variety of human motion tasks. Extensive experiments demonstrate that our model significantly improves temporal alignment, motion realism, and condition consistency over state-of-the-art approaches. Our project page is available at https://zquang2202.github.io/TCM.

---

### MMOT: The First Challenging Benchmark for Drone-based Multispectral Multi-Object Tracking

Authors: Tianhao Li, Tingfa Xu, Ying Wang, Haolin Qin, Xu Lin, Jianan Li
Published: 2025-10-14
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.12565v1

Drone-based multi-object tracking is essential yet highly challenging due to small targets, severe occlusions, and cluttered backgrounds. Existing RGB-based tracking algorithms heavily depend on spatial appearance cues such as color and texture, which often degrade in aerial views, compromising reliability. Multispectral imagery, capturing pixel-level spectral reflectance, provides crucial cues that enhance object discriminability under degraded spatial conditions. However, the lack of dedicated multispectral UAV datasets has hindered progress in this domain. To bridge this gap, we introduce MMOT, the first challenging benchmark for drone-based multispectral multi-object tracking. It features three key characteristics: (i) Large Scale - 125 video sequences with over 488.8K annotations across eight categories; (ii) Comprehensive Challenges - covering diverse conditions such as extreme small targets, high-density scenarios, severe occlusions, and complex motion; and (iii) Precise Oriented Annotations - enabling accurate localization and reduced ambiguity under aerial perspectives. To better extract spectral features and leverage oriented annotations, we further present a multispectral and orientation-aware MOT scheme adapting existing methods, featuring: (i) a lightweight Spectral 3D-Stem integrating spectral features while preserving compatibility with RGB pretraining; (ii) an orientation-aware Kalman filter for precise state estimation; and (iii) an end-to-end orientation-adaptive transformer. Extensive experiments across representative trackers consistently show that multispectral input markedly improves tracking performance over RGB baselines, particularly for small and densely packed objects. We believe our work will advance drone-based multispectral multi-object tracking research. Our MMOT, code, and benchmarks are publicly available at https://github.com/Annzstbl/MMOT.

---

### HardcoreLogic: Challenging Large Reasoning Models with Long-tail Logic Puzzle Games

Authors: Jingcong Liang, Shijun Wan, Xuehai Wu, Yitong Li, Qianglong Chen, Duyu Tang, Siyuan Wang, Zhongyu Wei
Published: 2025-10-14
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.12563v3

Large Reasoning Models (LRMs) have demonstrated impressive performance on complex tasks, including logical puzzle games that require deriving solutions satisfying all constraints. However, whether they can flexibly apply appropriate rules to varying conditions, particularly when faced with non-canonical game variants, remains an open question. Existing corpora focus on popular puzzles like 9x9 Sudoku, risking overfitting to canonical formats and memorization of solution patterns, which can mask deficiencies in understanding novel rules or adapting strategies to new variants. To address this, we introduce HardcoreLogic, a challenging benchmark of over 5,000 puzzles across 10 games, designed to test the robustness of LRMs on the "long-tail" of logical games. HardcoreLogic systematically transforms canonical puzzles through three dimensions: Increased Complexity (IC), Uncommon Elements (UE), and Unsolvable Puzzles (UP), reducing reliance on shortcut memorization. Evaluations on a diverse set of LRMs reveal significant performance drops, even for models achieving top scores on existing benchmarks, indicating heavy reliance on memorized stereotypes. While increased complexity is the dominant source of difficulty, models also struggle with subtle rule variations that do not necessarily increase puzzle difficulty. Our systematic error analysis on solvable and unsolvable puzzles further highlights gaps in genuine reasoning. Overall, HardcoreLogic exposes the limitations of current LRMs and establishes a benchmark for advancing high-level logical reasoning.

---

### CoIRL-AD: Collaborative-Competitive Imitation-Reinforcement Learning in Latent World Models for Autonomous Driving

Authors: Xiaoji Zheng, Ziyuan Yang, Yanhao Chen, Yuhang Peng, Yuanrong Tang, Gengyuan Liu, Bokui Chen, Jiangtao Gong
Published: 2025-10-14
Categories: cs.CV, cs.LG, cs.RO
arXiv: https://arxiv.org/abs/2510.12560v2

End-to-end autonomous driving models trained with imitation learning (IL) often generalize poorly, particularly in long-tail scenarios where expert demonstrations are sparse. Reinforcement learning (RL) can provide complementary task-level supervision, but applying RL to real-world autonomous driving is challenging in offline settings without interactive simulators, where datasets are dominated by expert actions and provide limited behavioral diversity. We propose CoIRL-AD, a competitive dual-policy framework that integrates IL and RL under a unified offline training regime. CoIRL-AD decouples imitation and reward optimization into separate actors to alleviate objective conflicts, uses imagined future rollouts for long-horizon reward estimation, and introduces a competition mechanism that selectively transfers beneficial behaviors while keeping RL anchored to expert-like driving. Experiments on the nuScenes benchmark show that CoIRL-AD consistently improves robustness over strong IL-based baselines, with especially large gains in cross-city generalization and long-tail scenarios. Code is available at: https://github.com/SEU-zxj/CoIRL-AD.

---

### Inclusive Fitness as a Key Step Towards More Advanced Social Behaviors in Multi-Agent Reinforcement Learning Settings

Authors: Andries Rosseau, Raphaël Avalos, Ann Nowé
Published: 2025-10-14
Categories: cs.AI, cs.MA, cs.SI
arXiv: https://arxiv.org/abs/2510.12555v1

The competitive and cooperative forces of natural selection have driven the evolution of intelligence for millions of years, culminating in nature's vast biodiversity and the complexity of human minds. Inspired by this process, we propose a novel multi-agent reinforcement learning framework where each agent is assigned a genotype and where reward functions are modelled after the concept of inclusive fitness. An agent's genetic material may be shared with other agents, and our inclusive reward function naturally accounts for this. We study the resulting social dynamics in two types of network games with prisoner's dilemmas and find that our results align with well-established principles from biology, such as Hamilton's rule. Furthermore, we outline how this framework can extend to more open-ended environments with spatial and temporal structure, finite resources, and evolving populations. We hypothesize the emergence of an arms race of strategies, where each new strategy is a gradual improvement over earlier adaptations of other agents, effectively producing a multi-agent autocurriculum analogous to biological evolution. In contrast to the binary team-based structures prevalent in earlier research, our gene-based reward structure introduces a spectrum of cooperation ranging from full adversity to full cooperativeness based on genetic similarity, enabling unique non team-based social dynamics. For example, one agent having a mutual cooperative relationship with two other agents, while the two other agents behave adversarially towards each other. We argue that incorporating inclusive fitness in agents provides a foundation for the emergence of more strategically advanced and socially intelligent agents.

---

### VISaGE: Understanding Visual Generics and Exceptions

Authors: Stella Frank, Emily Allaway
Published: 2025-10-14
Categories: cs.CL, cs.CV
arXiv: https://arxiv.org/abs/2510.12548v1

While Vision Language Models (VLMs) learn conceptual representations, in the form of generalized knowledge, during training, they are typically used to analyze individual instances. When evaluation instances are atypical, this paradigm results in tension between two priors in the model. The first is a pragmatic prior that the textual and visual input are both relevant, arising from VLM finetuning on congruent inputs; the second is a semantic prior that the conceptual representation is generally true for instances of the category. In order to understand how VLMs trade off these priors, we introduce a new evaluation dataset, VISaGE, consisting of both typical and exceptional images. In carefully balanced experiments, we show that conceptual understanding degrades when the assumption of congruency underlying the pragmatic prior is violated with incongruent images. This effect is stronger than the effect of the semantic prior when querying about individual instances.

---