# Research Papers — 2025-10 · bundle page 153 of 217

> Full-text markdown of papers #7601–7650 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/153
[← Previous bundle](https://aipod.dev/papers/archive/2025-10/page/152.md)
[Next bundle →](https://aipod.dev/papers/archive/2025-10/page/154.md)

### Kontinuous Kontext: Continuous Strength Control for Instruction-based Image Editing

Authors: Rishubh Parihar, Or Patashnik, Daniil Ostashev, R. Venkatesh Babu, Daniel Cohen-Or, Kuan-Chieh Wang
Published: 2025-10-09
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2510.08532v2

Instruction-based image editing offers a powerful and intuitive way to manipulate images through natural language. Yet, relying solely on text instructions limits fine-grained control over the extent of edits. We introduce Kontinuous Kontext, an instruction-driven editing model that provides a new dimension of control over edit strength, enabling users to adjust edits gradually from no change to a fully realized result in a smooth and continuous manner. Kontinuous Kontext extends a state-of-the-art image editing model to accept an additional input, a scalar edit strength which is then paired with the edit instruction, enabling explicit control over the extent of the edit. To inject this scalar information, we train a lightweight projector network that maps the input scalar and the edit instruction to coefficients in the model's modulation space. For training our model, we synthesize a diverse dataset of image-edit-instruction-strength quadruplets using existing generative models, followed by a filtering stage to ensure quality and consistency. Kontinuous Kontext provides a unified approach for fine-grained control over edit strength for instruction driven editing from subtle to strong across diverse operations such as stylization, attribute, material, background, and shape changes, without requiring attribute-specific training.

---

### SpatialLadder: Progressive Training for Spatial Reasoning in Vision-Language Models

Authors: Hongxing Li, Dingming Li, Zixuan Wang, Yuchen Yan, Hang Wu, Wenqi Zhang, Yongliang Shen, Weiming Lu, Jun Xiao, Yueting Zhuang
Published: 2025-10-09
Categories: cs.CV, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2510.08531v1

Spatial reasoning remains a fundamental challenge for Vision-Language Models (VLMs), with current approaches struggling to achieve robust performance despite recent advances. We identify that this limitation stems from a critical gap: existing methods attempt to learn spatial reasoning directly without establishing the hierarchical foundations of perception and understanding. To address this challenge, we present a comprehensive methodology for building spatial intelligence progressively. We introduce SpatialLadder-26k, a multimodal dataset containing 26,610 samples spanning object localization, single image, multi-view, and video spatial reasoning tasks, constructed through a standardized pipeline that ensures systematic coverage across modalities. Building on this dataset, we design a three-stage progressive training framework that (1) establishes spatial perception through object localization, (2) develops spatial understanding through multi-dimensional spatial tasks, and (3) strengthens complex reasoning via reinforcement learning with verifiable rewards. This approach yields SpatialLadder, a 3B-parameter model that achieves state-of-the-art performance on spatial reasoning benchmarks, with 23.4% average improvement over the base model, surpassing GPT-4o by 20.8% and Gemini-2.0-Flash by 10.1%. Notably, SpatialLadder maintains strong generalization with 7.2% improvement on out-of-domain benchmarks, demonstrating that progressive training from perception to reasoning is essential for robust spatial intelligence.

---

### X2Video: Adapting Diffusion Models for Multimodal Controllable Neural Video Rendering

Authors: Zhitong Huang, Mohan Zhang, Renhan Wang, Rui Tang, Hao Zhu, Jing Liao
Published: 2025-10-09
Categories: cs.GR, cs.CV
arXiv: https://arxiv.org/abs/2510.08530v1

We present X2Video, the first diffusion model for rendering photorealistic videos guided by intrinsic channels including albedo, normal, roughness, metallicity, and irradiance, while supporting intuitive multi-modal controls with reference images and text prompts for both global and local regions. The intrinsic guidance allows accurate manipulation of color, material, geometry, and lighting, while reference images and text prompts provide intuitive adjustments in the absence of intrinsic information. To enable these functionalities, we extend the intrinsic-guided image generation model XRGB to video generation by employing a novel and efficient Hybrid Self-Attention, which ensures temporal consistency across video frames and also enhances fidelity to reference images. We further develop a Masked Cross-Attention to disentangle global and local text prompts, applying them effectively onto respective local and global regions. For generating long videos, our novel Recursive Sampling method incorporates progressive frame sampling, combining keyframe prediction and frame interpolation to maintain long-range temporal consistency while preventing error accumulation. To support the training of X2Video, we assembled a video dataset named InteriorVideo, featuring 1,154 rooms from 295 interior scenes, complete with reliable ground-truth intrinsic channel sequences and smooth camera trajectories. Both qualitative and quantitative evaluations demonstrate that X2Video can produce long, temporally consistent, and photorealistic videos guided by intrinsic conditions. Additionally, X2Video effectively accommodates multi-modal controls with reference images, global and local text prompts, and simultaneously supports editing on color, material, geometry, and lighting through parametric tuning. Project page: https://luckyhzt.github.io/x2video

---

### CoMAS: Co-Evolving Multi-Agent Systems via Interaction Rewards

Authors: Xiangyuan Xue, Yifan Zhou, Guibin Zhang, Zaibin Zhang, Yijiang Li, Chen Zhang, Zhenfei Yin, Philip Torr, Wanli Ouyang, Lei Bai
Published: 2025-10-09
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.08529v2

Self-evolution is a central research topic in enabling large language model (LLM)-based agents to continually improve their capabilities after pretraining. Recent research has witnessed a transition from reinforcement learning (RL)-free to RL-based methods. Current RL-based methods either rely on dense external reward signals or extract intrinsic reward signals from LLMs themselves. However, these approaches diverge from the self-evolution mechanisms observed in human intelligence, where individuals learn and improve through mutual discussion and collaboration. In this work, we introduce Co-Evolving Multi-Agent Systems (CoMAS), a novel framework that enables agents to improve autonomously by learning from inter-agent interactions without external supervision. CoMAS generates intrinsic rewards from rich discussion dynamics, employs an LLM-as-a-judge mechanism to formulate these rewards, and optimizes each agent's policy through RL, thereby enabling decentralized and scalable co-evolution. Experimental results demonstrate that CoMAS consistently outperforms untrained agents and achieves state-of-the-art performance across most evaluation settings. Ablation studies confirm the necessity of interaction-based reward signals and reveal promising scalability as the number and diversity of agents increase. These findings establish CoMAS as a novel and effective paradigm for self-evolution in LLM-based agents.

---

### FlexTraj: Image-to-Video Generation with Flexible Point Trajectory Control

Authors: Zhiyuan Zhang, Can Wang, Dongdong Chen, Jing Liao
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.08527v1

We present FlexTraj, a framework for image-to-video generation with flexible point trajectory control. FlexTraj introduces a unified point-based motion representation that encodes each point with a segmentation ID, a temporally consistent trajectory ID, and an optional color channel for appearance cues, enabling both dense and sparse trajectory control. Instead of injecting trajectory conditions into the video generator through token concatenation or ControlNet, FlexTraj employs an efficient sequence-concatenation scheme that achieves faster convergence, stronger controllability, and more efficient inference, while maintaining robustness under unaligned conditions. To train such a unified point trajectory-controlled video generator, FlexTraj adopts an annealing training strategy that gradually reduces reliance on complete supervision and aligned condition. Experimental results demonstrate that FlexTraj enables multi-granularity, alignment-agnostic trajectory control for video generation, supporting various applications such as motion cloning, drag-based image-to-video, motion interpolation, camera redirection, flexible action control and mesh animations.

---

### Convergence Theorems for Entropy-Regularized and Distributional Reinforcement Learning

Authors: Yash Jhaveri, Harley Wiltzer, Patrick Shafto, Marc G. Bellemare, David Meger
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.08526v1

In the pursuit of finding an optimal policy, reinforcement learning (RL) methods generally ignore the properties of learned policies apart from their expected return. Thus, even when successful, it is difficult to characterize which policies will be learned and what they will do. In this work, we present a theoretical framework for policy optimization that guarantees convergence to a particular optimal policy, via vanishing entropy regularization and a temperature decoupling gambit. Our approach realizes an interpretable, diversity-preserving optimal policy as the regularization temperature vanishes and ensures the convergence of policy derived objects--value functions and return distributions. In a particular instance of our method, for example, the realized policy samples all optimal actions uniformly. Leveraging our temperature decoupling gambit, we present an algorithm that estimates, to arbitrary accuracy, the return distribution associated to its interpretable, diversity-preserving optimal policy.

---

### Which Heads Matter for Reasoning? RL-Guided KV Cache Compression

Authors: Wenjie Du, Li Jiang, Keda Tao, Xue Liu, Huan Wang
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.08525v3

Reasoning large language models exhibit complex reasoning behaviors via extended chain-of-thought generation that are highly fragile to information loss during decoding, creating critical challenges for KV cache compression. Existing token-dropping methods directly disrupt reasoning chains by removing intermediate steps, while head-reallocation methods, designed for retrieval tasks, fail to preserve the heads essential for generative reasoning. However, no existing method can identify which attention heads genuinely maintain reasoning consistency and control generation termination. To address this, we propose RLKV, which uses reinforcement learning as a probe to discover which heads contribute to reasoning quality by directly optimizing their cache usage against actual generation outcomes. This discovery naturally leads to an efficient compression strategy: we allocate full KV cache to reasoning-critical heads while aggressively compressing others with constant-size KV cache. Experiments reveal that a fraction of heads proves essential for reasoning, enabling 20--60% cache reduction with near-lossless performance across diverse tasks and models, and up to 2.06x end-to-end speedup at 60% reduction.

---

### Efficient Prompt Optimisation for Legal Text Classification with Proxy Prompt Evaluator

Authors: Hyunji Lee, Kevin Chenhao Li, Matthias Grabmair, Shanshan Xu
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.08524v1

Prompt optimization aims to systematically refine prompts to enhance a language model's performance on specific tasks. Fairness detection in Terms of Service (ToS) clauses is a challenging legal NLP task that demands carefully crafted prompts to ensure reliable results. However, existing prompt optimization methods are often computationally expensive due to inefficient search strategies and costly prompt candidate scoring. In this paper, we propose a framework that combines Monte Carlo Tree Search (MCTS) with a proxy prompt evaluator to more effectively explore the prompt space while reducing evaluation costs. Experiments demonstrate that our approach achieves higher classification accuracy and efficiency than baseline methods under a constrained computation budget.

---

### DYNAMIX: RL-based Adaptive Batch Size Optimization in Distributed Machine Learning Systems

Authors: Yuanjun Dai, Keqiang He, An Wang
Published: 2025-10-09
Categories: cs.LG, cs.DC
arXiv: https://arxiv.org/abs/2510.08522v1

Existing batch size selection approaches in distributed machine learning rely on static allocation or simplistic heuristics that fail to adapt to heterogeneous, dynamic computing environments. We present DYNAMIX, a reinforcement learning framework that formulates batch size optimization as a sequential decision-making problem using Proximal Policy Optimization (PPO). Our approach employs a multi-dimensional state representation encompassing network-level metrics, system-level resource utilization, and training statistical efficiency indicators to enable informed decision-making across diverse computational resources. Our approach eliminates the need for explicit system modeling while integrating seamlessly with existing distributed training frameworks. Through evaluations across diverse workloads, hardware configurations, and network conditions, DYNAMIX achieves up to 6.3% improvement in the final model accuracy and 46% reduction in the total training time. Our scalability experiments demonstrate that DYNAMIX maintains the best performance as cluster size increases to 32 nodes, while policy transfer experiments show that learned policies generalize effectively across related model architectures.

---

### FlowSearch: Advancing deep research with dynamic structured knowledge flow

Authors: Yusong Hu, Runmin Ma, Yue Fan, Jinxin Shi, Zongsheng Cao, Yuhao Zhou, Jiakang Yuan, Shuaiyu Zhang, Shiyang Feng, Xiangchao Yan, Shufei Zhang, Wenlong Zhang et al.
Published: 2025-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.08521v2

Deep research is an inherently challenging task that demands both breadth and depth of thinking. It involves navigating diverse knowledge spaces and reasoning over complex, multi-step dependencies, which presents substantial challenges for agentic systems. To address this, we propose FlowSearch, a multi-agent framework that actively constructs and evolves a dynamic structured knowledge flow to drive subtask execution and reasoning. FlowSearch is capable of strategically planning and expanding the knowledge flow to enable parallel exploration and hierarchical task decomposition, while also adjusting the knowledge flow in real time based on feedback from intermediate reasoning outcomes and insights. FlowSearch achieves competitive performance on both general and scientific benchmarks, including GAIA, HLE, GPQA and TRQA, demonstrating its effectiveness in multi-disciplinary research scenarios and its potential to advance scientific discovery. The code is available at https://github.com/InternScience/InternAgent.

---

### CaRT: Teaching LLM Agents to Know When They Know Enough

Authors: Grace Liu, Yuxiao Qu, Jeff Schneider, Aarti Singh, Aviral Kumar
Published: 2025-10-09
Categories: cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2510.08517v1

Many tasks require learned models to strategically gather relevant information over multiple rounds of interaction before actually acting on a task. Strategic information gathering requires models to know not only how to effectively acquire information, but also when to stop gathering information and make a decision, in order to avoid overthinking or getting derailed when acting. In this paper, we formalize this problem and introduce Counterfactuals and Reasoning for Termination (CaRT), an approach for teaching LLMs when to stop seeking information. To appropriately learn when to terminate, CaRT fine-tunes LLMs using counterfactual pairs of trajectories, one where termination is appropriate and a minimally modified version of the same trajectory where it is not. It trains the LLM to explain the rationale for the termination decision in either case via verbal reasoning, and imbues this capability into the base LLM via fine-tuning. We instantiate CaRT in two domains: interactive medical diagnosis and math problem solving. In both domains, we find that CaRT improves the efficiency of information gathering and task success rate compared to other fine-tuning methods.

---

### SliceFine: The Universal Winning-Slice Hypothesis for Pretrained Networks

Authors: Md Kowsher, Ali O. Polat, Ehsan Mohammady Ardehaly, Mehrdad Salehi, Zia Ghiasi, Prasanth Murali, Chen Chen
Published: 2025-10-09
Categories: cs.CV, cs.CL
arXiv: https://arxiv.org/abs/2510.08513v2

This paper presents a theoretical framework explaining why fine tuning small, randomly selected subnetworks (slices) within pre trained models can be sufficient for downstream adaptation. We prove that pretrained networks exhibit a universal winning slice property arising from two phenomena: (1) spectral balance the eigenspectra of different weight matrix slices are remarkably similar; and (2) high task energy their backbone representations retain rich, task relevant features. This leads to the Universal Winning Slice Hypothesis, which provides a theoretical foundation for parameter efficient fine tuning (PEFT) in large scale models. Inspired by this, we propose SliceFine, a PEFT method that exploits this inherent redundancy by updating only selected slices of the original weights introducing zero new parameters, unlike adapter-based approaches. Empirically, SliceFine matches the performance of state of the art PEFT methods across language and vision tasks, while significantly improving training speed, memory efficiency, and model compactness. Our work bridges theory and practice, offering a theoretically grounded alternative to existing PEFT techniques.

---

### Have We Scene It All? Scene Graph-Aware Deep Point Cloud Compression

Authors: Nikolaos Stathoulopoulos, Christoforos Kanellakis, George Nikolakopoulos
Published: 2025-10-09
Categories: cs.CV, cs.RO
arXiv: https://arxiv.org/abs/2510.08512v2

Efficient transmission of 3D point cloud data is critical for advanced perception in centralized and decentralized multi-agent robotic systems, especially nowadays with the growing reliance on edge and cloud-based processing. However, the large and complex nature of point clouds creates challenges under bandwidth constraints and intermittent connectivity, often degrading system performance. We propose a deep compression framework based on semantic scene graphs. The method decomposes point clouds into semantically coherent patches and encodes them into compact latent representations with semantic-aware encoders conditioned by Feature-wise Linear Modulation (FiLM). A folding-based decoder, guided by latent features and graph node attributes, enables structurally accurate reconstruction. Experiments on the SemanticKITTI and nuScenes datasets show that the framework achieves state-of-the-art compression rates, reducing data size by up to 98% while preserving both structural and semantic fidelity. In addition, it supports downstream applications such as multi-robot pose graph optimization and map merging, achieving trajectory accuracy and map alignment comparable to those obtained with raw LiDAR scans.

---

### AutoMLGen: Navigating Fine-Grained Optimization for Coding Agents

Authors: Shangheng Du, Xiangchao Yan, Dengyang Jiang, Jiakang Yuan, Yusong Hu, Xin Li, Liang He, Bo Zhang, Lei Bai
Published: 2025-10-09
Categories: cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2510.08511v1

Large language models (LLMs) have shown impressive performance in general programming tasks. However, in Machine Learning Engineering (MLE) scenarios such as AutoML and Kaggle competitions, achieving high performance depends heavily on expert intervention and repeated adjustments rather than simply generating correct code. When applied directly to these tasks, LLMs often lack fine-grained domain priors, and existing MLE approaches that use linear or tree-structured searches limit knowledge transfer to adjacent hierarchical links. As a result, they cannot leverage past full trajectories or share information across branches, limiting self-evolving ability and search space diversity. To address these limitations, we introduce AutoMLGen, an LLM-based coding agent that integrates a domain knowledge base for high-quality prior guidance and Monte Carlo Graph Search (MCGS) for efficient exploration. MCGS retains the tree-guided exploration of MCTS while embedding a graph structure into the expansion stage to enable dynamic path reorganization, historical trajectory reuse, and multi-solution fusion to support both self-evolution and collaborative learning. Combined with fine-grained operator sets, this design improves stability and accelerates convergence. Evaluation on the MLE-Bench shows that AutoMLGen achieves state-of-the-art performance in numerous dimensions, such as the average medal rate and the valid submission rate, under a 12-hour budget (half the standard runtime). The code is available at https://github.com/Alpha-Innovator/InternAgent.

---

### To Sink or Not to Sink: Visual Information Pathways in Large Vision-Language Models

Authors: Jiayun Luo, Wan-Cyuan Fan, Lyuyang Wang, Xiangteng He, Tanzila Rahman, Purang Abolmaesumi, Leonid Sigal
Published: 2025-10-09
Categories: cs.CV, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2510.08510v1

Large Vision Language Models (LVLMs) have recently emerged as powerful architectures capable of understanding and reasoning over both visual and textual information. These models typically rely on two key components: a Vision Transformer (ViT) and a Large Language Model (LLM). ViT encodes visual content into a sequence of image tokens and serves as the perceptual front-end -- the eyes of the model. In contrast, the LLM interprets these tokens to perform high-level reasoning, generates responses, and functions as the cognitive core -- the brain of the model. However, it remains unclear which visual tokens contribute most significantly to understanding and reasoning, and how effectively these signals are propagated from ViT to the LLM. While most existing works have focused on identifying attention sinks, low-semantic tokens receiving disproportionately high attention, within the LLM, we shift the focus to the vision encoder by identifying a class of high-norm visual tokens from ViT, referred to as ViT attention sinks -- a problem that has been rarely studied but is indeed very important for LVLMs. Our findings show that these ViT sinks encapsulate high-level semantic concepts from images, allowing the LLM to perform more effective understanding and reasoning. Despite their importance, these sink tokens are often overlooked in existing LVLM architectures. To explore their contribution, we present both qualitative and quantitative analyses of the information embedded in these sink tokens. We also propose both training-free and training-based approaches to better leverage how this information is interpreted by the LLM, and to what extent. By explicitly utilizing these tokens, we demonstrate substantial improvements across a range of LVLMs and visual reasoning tasks, highlighting the untapped potential of ViT attention sinks in enhancing visual reasoning.

---

### MoA-VR: A Mixture-of-Agents System Towards All-in-One Video Restoration

Authors: Lu Liu, Chunlei Cai, Shaocheng Shen, Jianfeng Liang, Weimin Ouyang, Tianxiao Ye, Jian Mao, Huiyu Duan, Jiangchao Yao, Xiaoyun Zhang, Qiang Hu, Guangtao Zhai
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.08508v1

Real-world videos often suffer from complex degradations, such as noise, compression artifacts, and low-light distortions, due to diverse acquisition and transmission conditions. Existing restoration methods typically require professional manual selection of specialized models or rely on monolithic architectures that fail to generalize across varying degradations. Inspired by expert experience, we propose MoA-VR, the first \underline{M}ixture-\underline{o}f-\underline{A}gents \underline{V}ideo \underline{R}estoration system that mimics the reasoning and processing procedures of human professionals through three coordinated agents: Degradation Identification, Routing and Restoration, and Restoration Quality Assessment. Specifically, we construct a large-scale and high-resolution video degradation recognition benchmark and build a vision-language model (VLM) driven degradation identifier. We further introduce a self-adaptive router powered by large language models (LLMs), which autonomously learns effective restoration strategies by observing tool usage patterns. To assess intermediate and final processed video quality, we construct the \underline{Res}tored \underline{V}ideo \underline{Q}uality (Res-VQ) dataset and design a dedicated VLM-based video quality assessment (VQA) model tailored for restoration tasks. Extensive experiments demonstrate that MoA-VR effectively handles diverse and compound degradations, consistently outperforming existing baselines in terms of both objective metrics and perceptual quality. These results highlight the potential of integrating multimodal intelligence and modular reasoning in general-purpose video restoration systems.

---

### Neologism Learning for Controllability and Self-Verbalization

Authors: John Hewitt, Oyvind Tafjord, Robert Geirhos, Been Kim
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.08506v1

Humans invent new words when there is a rising demand for a new useful concept (e.g., doomscrolling). We explore and validate a similar idea in our communication with LLMs: introducing new words to better understand and control the models, expanding on the recently introduced neologism learning. This method introduces a new word by adding a new word embedding and training with examples that exhibit the concept with no other changes in model parameters. We show that adding a new word allows for control of concepts such as flattery, incorrect answers, text length, as well as more complex concepts in AxBench. We discover that neologisms can also further our understanding of the model via self-verbalization: models can describe what each new word means to them in natural language, like explaining that a word that represents a concept of incorrect answers means ``a lack of complete, coherent, or meaningful answers...'' To validate self-verbalizations, we introduce plug-in evaluation: we insert the verbalization into the context of a model and measure whether it controls the target concept. In some self-verbalizations, we find machine-only synonyms: words that seem unrelated to humans but cause similar behavior in machines. Finally, we show how neologism learning can jointly learn multiple concepts in multiple words.

---

### AI-Driven Radiology Report Generation for Traumatic Brain Injuries

Authors: Riadh Bouslimi, Houda Trabelsi, Wahiba Ben Abdssalem Karaa, Hana Hedhli
Published: 2025-10-09
Categories: eess.IV, cs.AI, cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2510.08498v1

Traumatic brain injuries present significant diagnostic challenges in emergency medicine, where the timely interpretation of medical images is crucial for patient outcomes. In this paper, we propose a novel AI-based approach for automatic radiology report generation tailored to cranial trauma cases. Our model integrates an AC-BiFPN with a Transformer architecture to capture and process complex medical imaging data such as CT and MRI scans. The AC-BiFPN extracts multi-scale features, enabling the detection of intricate anomalies like intracranial hemorrhages, while the Transformer generates coherent, contextually relevant diagnostic reports by modeling long-range dependencies. We evaluate the performance of our model on the RSNA Intracranial Hemorrhage Detection dataset, where it outperforms traditional CNN-based models in both diagnostic accuracy and report generation. This solution not only supports radiologists in high-pressure environments but also provides a powerful educational tool for trainee physicians, offering real-time feedback and enhancing their learning experience. Our findings demonstrate the potential of combining advanced feature extraction with transformer-based text generation to improve clinical decision-making in the diagnosis of traumatic brain injuries.

---

### Better Together: Leveraging Unpaired Multimodal Data for Stronger Unimodal Models

Authors: Sharut Gupta, Shobhita Sundaram, Chenyu Wang, Stefanie Jegelka, Phillip Isola
Published: 2025-10-09
Categories: cs.LG, cs.CV
arXiv: https://arxiv.org/abs/2510.08492v1

Traditional multimodal learners find unified representations for tasks like visual question answering, but rely heavily on paired datasets. However, an overlooked yet potentially powerful question is: can one leverage auxiliary unpaired multimodal data to directly enhance representation learning in a target modality? We introduce UML: Unpaired Multimodal Learner, a modality-agnostic training paradigm in which a single model alternately processes inputs from different modalities while sharing parameters across them. This design exploits the assumption that different modalities are projections of a shared underlying reality, allowing the model to benefit from cross-modal structure without requiring explicit pairs. Theoretically, under linear data-generating assumptions, we show that unpaired auxiliary data can yield representations strictly more informative about the data-generating process than unimodal training. Empirically, we show that using unpaired data from auxiliary modalities -- such as text, audio, or images -- consistently improves downstream performance across diverse unimodal targets such as image and audio. Our project page: https://unpaired-multimodal.github.io/

---

### Splat the Net: Radiance Fields with Splattable Neural Primitives

Authors: Xilong Zhou, Bao-Huy Nguyen, Loïc Magne, Vladislav Golyanik, Thomas Leimkühler, Christian Theobalt
Published: 2025-10-09
Categories: cs.GR, cs.CV
arXiv: https://arxiv.org/abs/2510.08491v2

Radiance fields have emerged as a predominant representation for modeling 3D scene appearance. Neural formulations such as Neural Radiance Fields provide high expressivity but require costly ray marching for rendering, whereas primitive-based methods such as 3D Gaussian Splatting offer real-time efficiency through splatting, yet at the expense of representational power. Inspired by advances in both these directions, we introduce splattable neural primitives, a new volumetric representation that reconciles the expressivity of neural models with the efficiency of primitive-based splatting. Each primitive encodes a bounded neural density field parameterized by a shallow neural network. Our formulation admits an exact analytical solution for line integrals, enabling efficient computation of perspectively accurate splatting kernels. As a result, our representation supports integration along view rays without the need for costly ray marching. The primitives flexibly adapt to scene geometry and, being larger than prior analytic primitives, reduce the number required per scene. On novel-view synthesis benchmarks, our approach matches the quality and speed of 3D Gaussian Splatting while using $10\times$ fewer primitives and $6\times$ fewer parameters. These advantages arise directly from the representation itself, without reliance on complex control or adaptation frameworks. The project page is https://vcai.mpi-inf.mpg.de/projects/SplatNet/.

---

### Implementing Semantic Join Operators Efficiently

Authors: Immanuel Trummer
Published: 2025-10-09
Categories: cs.DB, cs.LG
arXiv: https://arxiv.org/abs/2510.08489v1

Semantic query processing engines often support semantic joins, enabling users to match rows that satisfy conditions specified in natural language. Such join conditions can be evaluated using large language models (LLMs) that solve novel tasks without task-specific training. Currently, many semantic query processing engines implement semantic joins via nested loops, invoking the LLM to evaluate the join condition on row pairs. Instead, this paper proposes a novel algorithm, inspired by the block nested loops join operator implementation in traditional database systems. The proposed algorithm integrates batches of rows from both input tables into a single prompt. The goal of the LLM invocation is to identify all matching row pairs in the current input. The paper introduces formulas that can be used to optimize the size of the row batches, taking into account constraints on the size of the LLM context window (limiting both input and output size). An adaptive variant of the proposed algorithm refers to cases in which the size of the output is difficult to estimate. A formal analysis of asymptotic processing costs, as well as empirical results, demonstrates that the proposed approach reduces costs significantly and performs well compared to join implementations used by recent semantic query processing engines.

---

### InstructX: Towards Unified Visual Editing with MLLM Guidance

Authors: Chong Mou, Qichao Sun, Yanze Wu, Pengze Zhang, Xinghui Li, Fulong Ye, Songtao Zhao, Qian He
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.08485v1

With recent advances in Multimodal Large Language Models (MLLMs) showing strong visual understanding and reasoning, interest is growing in using them to improve the editing performance of diffusion models. Despite rapid progress, most studies lack an in-depth analysis of MLLM design choices. Moreover, the integration of MLLMs and diffusion models remains an open challenge in some difficult tasks, such as video editing. In this paper, we present InstructX, a unified framework for image and video editing. Specifically, we conduct a comprehensive study on integrating MLLMs and diffusion models for instruction-driven editing across diverse tasks. Building on this study, we analyze the cooperation and distinction between images and videos in unified modeling. (1) We show that training on image data can lead to emergent video editing capabilities without explicit supervision, thereby alleviating the constraints imposed by scarce video training data. (2) By incorporating modality-specific MLLM features, our approach effectively unifies image and video editing tasks within a single model. Extensive experiments demonstrate that our method can handle a broad range of image and video editing tasks and achieves state-of-the-art performance.

---

### DeepPrune: Parallel Scaling without Inter-trace Redundancy

Authors: Shangqing Tu, Yaxuan Li, Yushi Bai, Lei Hou, Juanzi Li
Published: 2025-10-09
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.08483v2

Parallel scaling has emerged as a powerful paradigm to enhance reasoning capabilities in large language models (LLMs) by generating multiple Chain-of-Thought (CoT) traces simultaneously. However, this approach introduces significant computational inefficiency due to inter-trace redundancy -- our analysis reveals that over 80% of parallel reasoning traces yield identical final answers, representing substantial wasted computation. To address this critical efficiency bottleneck, we propose DeepPrune, a novel framework that enables efficient parallel scaling through dynamic pruning. Our method features a specialized judge model trained with out-of-distribution data (AIME 2022, AIME 2023, and MATH 500) using oversampling techniques to accurately predict answer equivalence from partial reasoning traces, achieving 0.7072 AUROC on unseen reasoning models. Combined with an online greedy clustering algorithm that dynamically prunes redundant paths while preserving answer diversity. Comprehensive evaluations across three challenging benchmarks (AIME 2024, AIME 2025, and GPQA) and multiple reasoning models demonstrate that DeepPrune achieves remarkable token reduction of 65.73%--88.50% compared to conventional consensus sampling, while maintaining competitive accuracy within 3 percentage points. Our work establishes a new standard for efficient parallel reasoning, making high-performance reasoning more efficient. Our code and data are here: https://deepprune.github.io/.

---

### The Visual Iconicity Challenge: Evaluating Vision-Language Models on Sign Language Form-Meaning Mapping

Authors: Onur Keleş, Aslı Özyürek, Gerardo Ortega, Kadir Gökgöz, Esam Ghaleb
Published: 2025-10-09
Categories: cs.CV, cs.CL
arXiv: https://arxiv.org/abs/2510.08482v3

Iconicity, the resemblance between linguistic form and meaning, is pervasive in signed languages, offering a natural testbed for visual grounding. For vision-language models (VLMs), the challenge is to recover such essential mappings from dynamic human motion rather than static context. We introduce the Visual Iconicity Challenge, a novel video-based benchmark that adapts psycholinguistic measures to evaluate VLMs on three tasks: (i) phonological sign-form prediction (e.g., handshape, location), (ii) transparency (inferring meaning from visual form), and (iii) graded iconicity ratings. We assess 13 state-of-the-art VLMs in zero- and few-shot settings on Sign Language of the Netherlands and compare them to human baselines. On phonological form prediction, VLMs recover some handshape and location detail but remain below human performance; on transparency, they are far from human baselines; and only top models correlate moderately with human iconicity ratings. Interestingly, models with stronger phonological form prediction correlate better with human iconicity judgment, indicating shared sensitivity to visually grounded structure. Our findings validate these diagnostic tasks and motivate human-centric signals and embodied learning methods for modelling iconicity and improving visual grounding in multimodal models.

---

### Video-STAR: Reinforcing Open-Vocabulary Action Recognition with Tools

Authors: Zhenlong Yuan, Xiangyan Qu, Chengxuan Qian, Rui Chen, Jing Tang, Lei Sun, Xiangxiang Chu, Dapeng Zhang, Yiwei Wang, Yujun Cai, Shuo Li
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.08480v2

Multimodal large language models (MLLMs) have demonstrated remarkable potential in bridging visual and textual reasoning, yet their reliance on text-centric priors often limits their ability to disentangle semantically similar actions in open-vocabulary scenarios. To address this, we propose Video-STAR, a framework that harmonizes contextual sub-motion decomposition with tool-augmented reinforcement learning for open-vocabulary action recognition (OVAR). Unlike prior methods that treat actions as monolithic entities, our approach innovatively decomposes actions into discriminative sub-motions for fine-grained matching while dynamically invoking domain-specific tools for cross-modal interleaving, thereby enabling category-specific reasoning capacity and reducing cross-modal hallucination. Moreover, by designing a hierarchical reward that balances tool-usage efficiency, sub-motion relevance, and structural coherence in reasoning, our method autonomously leverages external tools to prioritize sub-motion patterns without explicit supervision, transmitting from text-centric reasoning to visually grounded inference. Extensive evaluations on HMDB-51, UCF-101, SSv2, Kinetics-400, and Kinetics-600 datasets demonstrate our state-of-the-art performance, outperforming existing methods in distinguishing fine-grained actions and handling cross-modal hallucination, validating our excellent robustness and generalization.

---

### DexMan: Learning Bimanual Dexterous Manipulation from Human and Generated Videos

Authors: Jhen Hsieh, Kuan-Hsun Tu, Kuo-Han Hung, Tsung-Wei Ke
Published: 2025-10-09
Categories: cs.RO, cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2510.08475v1

We present DexMan, an automated framework that converts human visual demonstrations into bimanual dexterous manipulation skills for humanoid robots in simulation. Operating directly on third-person videos of humans manipulating rigid objects, DexMan eliminates the need for camera calibration, depth sensors, scanned 3D object assets, or ground-truth hand and object motion annotations. Unlike prior approaches that consider only simplified floating hands, it directly controls a humanoid robot and leverages novel contact-based rewards to improve policy learning from noisy hand-object poses estimated from in-the-wild videos. DexMan achieves state-of-the-art performance in object pose estimation on the TACO benchmark, with absolute gains of 0.08 and 0.12 in ADD-S and VSD. Meanwhile, its reinforcement learning policy surpasses previous methods by 19% in success rate on OakInk-v2. Furthermore, DexMan can generate skills from both real and synthetic videos, without the need for manual data collection and costly motion capture, and enabling the creation of large-scale, diverse datasets for training generalist dexterous manipulation.

---

### Looking to Learn: Token-wise Dynamic Gating for Low-Resource Vision-Language Modelling

Authors: Bianca-Mihaela Ganescu, Suchir Salhan, Andrew Caines, Paula Buttery
Published: 2025-10-09
Categories: cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2510.08470v1

Training vision-language models on cognitively-plausible amounts of data requires rethinking how models integrate multimodal information. Within the constraints of the Vision track for the BabyLM Challenge 2025, we propose a lightweight decoder-based architecture with (1) token-wise dynamic gating for adaptive fusion of linguistic and visual cues, (2) feature modulation and channel attention to maximise the utility of limited visual information and (3) auxiliary contrastive objectives for visual grounding. Evaluation on five benchmarks (BLiMP, BLiMP Supplement, EWoK, Winoground and VQA) shows competitive or superior performance to multimodal baselines. More notably, our dynamic gate discovers interpretable patterns without explicit supervision, favouring visual cues for content words and linguistic cues for function words. While we identify limitations in the Challenge constraints, such as the information bottleneck created by global image embeddings and training instability from the dataset split, our findings establish dynamic gating as a powerful tool for efficient multimodal learning, offering both interpretability and performance even under severe constraints.

---

### Platform-Agnostic Modular Architecture for Quantum Benchmarking

Authors: Neer Patel, Anish Giri, Hrushikesh Pramod Patil, Noah Siekierski, Avimita Chatterjee, Sonika Johri, Timothy Proctor, Thomas Lubinski, Siyuan Niu
Published: 2025-10-09
Categories: quant-ph, cs.AI, cs.SE
arXiv: https://arxiv.org/abs/2510.08469v1

We present a platform-agnostic modular architecture that addresses the increasingly fragmented landscape of quantum computing benchmarking by decoupling problem generation, circuit execution, and results analysis into independent, interoperable components. Supporting over 20 benchmark variants ranging from simple algorithmic tests like Bernstein-Vazirani to complex Hamiltonian simulation with observable calculations, the system integrates with multiple circuit generation APIs (Qiskit, CUDA-Q, Cirq) and enables diverse workflows. We validate the architecture through successful integration with Sandia's $\textit{pyGSTi}$ for advanced circuit analysis and CUDA-Q for multi-GPU HPC simulations. Extensibility of the system is demonstrated by implementing dynamic circuit variants of existing benchmarks and a new quantum reinforcement learning benchmark, which become readily available across multiple execution and analysis modes. Our primary contribution is identifying and formalizing modular interfaces that enable interoperability between incompatible benchmarking frameworks, demonstrating that standardized interfaces reduce ecosystem fragmentation while preserving optimization flexibility. This architecture has been developed as a key enhancement to the continually evolving QED-C Application-Oriented Performance Benchmarks for Quantum Computing suite.

---

### In-Context Clustering with Large Language Models

Authors: Ying Wang, Mengye Ren, Andrew Gordon Wilson
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.08466v1

We propose In-Context Clustering (ICC), a flexible LLM-based procedure for clustering data from diverse distributions. Unlike traditional clustering algorithms constrained by predefined similarity measures, ICC flexibly captures complex relationships among inputs through an attention mechanism. We show that pretrained LLMs exhibit impressive zero-shot clustering capabilities on text-encoded numeric data, with attention matrices showing salient cluster patterns. Spectral clustering using attention matrices offers surprisingly competitive performance. We further enhance the clustering capabilities of LLMs on numeric and image data through fine-tuning using the Next Token Prediction (NTP) loss. Moreover, the flexibility of LLM prompting enables text-conditioned image clustering, a capability that classical clustering methods lack. Our work extends in-context learning to an unsupervised setting, showcasing the effectiveness and flexibility of LLMs for clustering. Our code is available at https://agenticlearning.ai/icc.

---

### Accumulated Aggregated D-Optimal Designs for Estimating Main Effects in Black-Box Models

Authors: Chih-Yu Chang, Ming-Chung Chang
Published: 2025-10-09
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2510.08465v2

Estimating how individual input variables affect the output of a black-box model is a central task in explainable machine learning. However, existing methods suffer from two key limitations: sensitivity to out-of-distribution (OOD) evaluations, which arises when query points are placed far from the data manifold, and instability under feature correlation, which can lead to unreliable effect estimates in practice. We introduce a unified view of main effect estimation as a design problem, which reveals that all existing methods differ only in their choice of evaluation locations. Building on this formulation, we propose A2D2E, an Estimator based on Accumulated Aggregated D-Optimal Designs, which replaces evaluations with a D-optimal hypercube design to minimize the variance of main effect estimation. A2D2E is model-agnostic, requires no differentiability of the predictor, and admits a closed-form estimator with complexity comparable to existing approaches. We establish that A2D2E is consistent to the same population target as ALE, and extend this result to the realistic setting where only a surrogate model is available. Through extensive simulations across multiple predictive models and dependence settings, we demonstrate that A2D2E outperforms ALE-based methods, with the largest gains under high feature correlation.

---

### Don't Run with Scissors: Pruning Breaks VLA Models but They Can Be Recovered

Authors: Jason Jabbour, Dong-Ki Kim, Max Smith, Jay Patrikar, Radhika Ghosal, Youhui Wang, Ali Agha, Vijay Janapa Reddi, Shayegan Omidshafiei
Published: 2025-10-09
Categories: cs.RO, cs.LG
arXiv: https://arxiv.org/abs/2510.08464v2

Vision-Language-Action (VLA) models are a promising paradigm for field robotics, but their deployment is bottlenecked by strict edge SWaP-C constraints. Compression techniques such as quantization, distillation, and pruning are well-established for LLMs and VLMs. Pruning, however, remains particularly underexplored for VLA models. We surprisingly find that pruning VLA models yields drastic performance degradation and increased safety violations. Through spectral analysis, we identify structural properties of VLA weights that help explain this fragility. Based on this insight, we introduce GLUESTICK, a post-pruning recovery method that restores much of the original model's capability while preserving sparsity benefits. GLUESTICK computes a one-time corrective term by interpolating between the dense and pruned models in weight space. At inference, each pruned layer applies this correction to recover lost performance with minimal overhead. Our approach requires no additional training and is agnostic to the pruning algorithm. Across diverse VLA architectures and domains, GLUESTICK recovers up to 100% of the success rate and restores safety to near-dense levels, while saving up to 5 GB of memory and achieving up to 1.75x speedup. Additional material can be found at: https://gluestick-vla.github.io/

---

### Wavefunction Flows: Efficient Quantum Simulation of Continuous Flow Models

Authors: David Layden, Ryan Sweke, Vojtěch Havlíček, Anirban Chowdhury, Kirill Neklyudov
Published: 2025-10-09
Categories: quant-ph, cs.LG, stat.ML
arXiv: https://arxiv.org/abs/2510.08462v1

Flow models are a cornerstone of modern machine learning. They are generative models that progressively transform probability distributions according to learned dynamics. Specifically, they learn a continuous-time Markov process that efficiently maps samples from a simple source distribution into samples from a complex target distribution. We show that these models are naturally related to the Schrödinger equation, for an unusual Hamiltonian on continuous variables. Moreover, we prove that the dynamics generated by this Hamiltonian can be efficiently simulated on a quantum computer. Together, these results give a quantum algorithm for preparing coherent encodings (a.k.a., qsamples) for a vast family of probability distributions--namely, those expressible by flow models--by reducing the task to an existing classical learning problem, plus Hamiltonian simulation. For statistical problems defined by flow models, such as mean estimation and property testing, this enables the use of quantum algorithms tailored to qsamples, which may offer advantages over classical algorithms based only on samples from a flow model. More broadly, these results reveal a close connection between state-of-the-art machine learning models, such as flow matching and diffusion models, and one of the main expected capabilities of quantum computers: simulating quantum dynamics.

---

### LeWiDi-2025 at NLPerspectives: Third Edition of the Learning with Disagreements Shared Task

Authors: Elisa Leonardelli, Silvia Casola, Siyao Peng, Giulia Rizzi, Valerio Basile, Elisabetta Fersini, Diego Frassinelli, Hyewon Jang, Maja Pavlovic, Barbara Plank, Massimo Poesio
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.08460v2

Many researchers have reached the conclusion that AI models should be trained to be aware of the possibility of variation and disagreement in human judgments, and evaluated as per their ability to recognize such variation. The LEWIDI series of shared tasks on Learning With Disagreements was established to promote this approach to training and evaluating AI models, by making suitable datasets more accessible and by developing evaluation methods. The third edition of the task builds on this goal by extending the LEWIDI benchmark to four datasets spanning paraphrase identification, irony detection, sarcasm detection, and natural language inference, with labeling schemes that include not only categorical judgments as in previous editions, but ordinal judgments as well. Another novelty is that we adopt two complementary paradigms to evaluate disagreement-aware systems: the soft-label approach, in which models predict population-level distributions of judgments, and the perspectivist approach, in which models predict the interpretations of individual annotators. Crucially, we moved beyond standard metrics such as cross-entropy, and tested new evaluation metrics for the two paradigms. The task attracted diverse participation, and the results provide insights into the strengths and limitations of methods to modeling variation. Together, these contributions strengthen LEWIDI as a framework and provide new resources, benchmarks, and findings to support the development of disagreement-aware technologies.

---

### SummDiff: Generative Modeling of Video Summarization with Diffusion

Authors: Kwanseok Kim, Jaehoon Hahm, Sumin Kim, Jinhwan Sul, Byunghak Kim, Joonseok Lee
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.08458v1

Video summarization is a task of shortening a video by choosing a subset of frames while preserving its essential moments. Despite the innate subjectivity of the task, previous works have deterministically regressed to an averaged frame score over multiple raters, ignoring the inherent subjectivity of what constitutes a good summary. We propose a novel problem formulation by framing video summarization as a conditional generation task, allowing a model to learn the distribution of good summaries and to generate multiple plausible summaries that better reflect varying human perspectives. Adopting diffusion models for the first time in video summarization, our proposed method, SummDiff, dynamically adapts to visual contexts and generates multiple candidate summaries conditioned on the input video. Extensive experiments demonstrate that SummDiff not only achieves the state-of-the-art performance on various benchmarks but also produces summaries that closely align with individual annotator preferences. Moreover, we provide a deeper insight with novel metrics from an analysis of the knapsack, which is an important last step of generating summaries but has been overlooked in evaluation.

---

### ARES: Multimodal Adaptive Reasoning via Difficulty-Aware Token-Level Entropy Shaping

Authors: Shuang Chen, Yue Guo, Yimeng Ye, Shijue Huang, Wenbo Hu, Haoxi Li, Manyuan Zhang, Jiayu Chen, Song Guo, Nanyun Peng
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.08457v1

Recent advances in multimodal large reasoning models (MLRMs) have substantially improved their ability to solve complex textual and visual tasks. However, these models tend to overthink on simple problems, producing unnecessarily lengthy reasoning traces, while under-exploring on challenging ones, leading to missed solutions. To address this imbalance, we propose ARES, a unified open-source framework for adaptive reasoning that dynamically allocates exploration effort based on task difficulty. Our approach is motivated by two key empirical findings: (i) while single-token entropy is noisy, high window-entropy (HWE) tokens (token-level entropies averaged under a sliding window) can reliably capture reasoning-critical moments; and (ii) reducing HWE usage benefits easy problems, while increasing it is essential for solving hard ones. Building on these insights, ARES introduces a two-stage training pipeline. In the Adaptive Cold-Start stage, we curate multimodal and textual data paired with reasoning traces of length proportional to problem difficulty, equipping the model with initial difficulty awareness. In the second stage, we develop Adaptive Entropy Policy Optimization (AEPO), which uses HWE tokens as exploration triggers to decide when to explore, and a hierarchical entropy reward with dynamic KL control to decide how much to explore. Extensive experiments demonstrate that ARES achieves superior performance and reasoning efficiency across diverse mathematical, logical, and multimodal benchmarks, while closing the gap to leading commercial systems under significantly lower inference costs.

---

### Integral Signatures of Activation Functions: A 9-Dimensional Taxonomy and Stability Theory for Deep Learning

Authors: Ankur Mali, Lawrence Hall, Jake Williams, Gordon Richards
Published: 2025-10-09
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.08456v1

Activation functions govern the expressivity and stability of neural networks, yet existing comparisons remain largely heuristic. We propose a rigorous framework for their classification via a nine-dimensional integral signature S_sigma(phi), combining Gaussian propagation statistics (m1, g1, g2, m2, eta), asymptotic slopes (alpha_plus, alpha_minus), and regularity measures (TV(phi'), C(phi)). This taxonomy establishes well-posedness, affine reparameterization laws with bias, and closure under bounded slope variation. Dynamical analysis yields Lyapunov theorems with explicit descent constants and identifies variance stability regions through (m2', g2). From a kernel perspective, we derive dimension-free Hessian bounds and connect smoothness to bounded variation of phi'. Applying the framework, we classify eight standard activations (ReLU, leaky-ReLU, tanh, sigmoid, Swish, GELU, Mish, TeLU), proving sharp distinctions between saturating, linear-growth, and smooth families. Numerical Gauss-Hermite and Monte Carlo validation confirms theoretical predictions. Our framework provides principled design guidance, moving activation choice from trial-and-error to provable stability and kernel conditioning.

---

### gLSTM: Mitigating Over-Squashing by Increasing Storage Capacity

Authors: Hugh Blayney, Álvaro Arroyo, Xiaowen Dong, Michael M. Bronstein
Published: 2025-10-09
Categories: cs.LG, cs.AI, stat.ML
arXiv: https://arxiv.org/abs/2510.08450v2

Graph Neural Networks (GNNs) leverage the graph structure to transmit information between nodes, typically through the message-passing mechanism. While these models have found a wide variety of applications, they are known to suffer from over-squashing, where information from a large receptive field of node representations is collapsed into a single fixed sized vector, resulting in an information bottleneck. In this paper, we re-examine the over-squashing phenomenon through the lens of model storage and retrieval capacity, which we define as the amount of information that can be stored in a node's representation for later use. We study some of the limitations of existing tasks used to measure over-squashing and introduce a new synthetic task to demonstrate that an information bottleneck can saturate this capacity. Furthermore, we adapt ideas from the sequence modeling literature on associative memories, fast weight programmers, and the xLSTM model to develop a novel GNN architecture with improved capacity. We demonstrate strong performance of this architecture both on our capacity synthetic task, as well as a range of real-world graph benchmarks.

---

### Hierarchical Spatial Algorithms for High-Resolution Image Quantization and Feature Extraction

Authors: Noor Islam S. Mohammad
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.08449v1

This study introduces a modular framework for spatial image processing, integrating grayscale quantization, color and brightness enhancement, image sharpening, bidirectional transformation pipelines, and geometric feature extraction. A stepwise intensity transformation quantizes grayscale images into eight discrete levels, producing a posterization effect that simplifies representation while preserving structural detail. Color enhancement is achieved via histogram equalization in both RGB and YCrCb color spaces, with the latter improving contrast while maintaining chrominance fidelity. Brightness adjustment is implemented through HSV value-channel manipulation, and image sharpening is performed using a 3 * 3 convolution kernel to enhance high-frequency details. A bidirectional transformation pipeline that integrates unsharp masking, gamma correction, and noise amplification achieved accuracy levels of 76.10% and 74.80% for the forward and reverse processes, respectively. Geometric feature extraction employed Canny edge detection, Hough-based line estimation (e.g., 51.50° for billiard cue alignment), Harris corner detection, and morphological window localization. Cue isolation further yielded 81.87\% similarity against ground truth images. Experimental evaluation across diverse datasets demonstrates robust and deterministic performance, highlighting its potential for real-time image analysis and computer vision.

---

### Synthetic Series-Symbol Data Generation for Time Series Foundation Models

Authors: Wenxuan Wang, Kai Wu, Yujian Betterest Li, Dan Wang, Xiaoyu Zhang
Published: 2025-10-09
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.08445v3

Foundation models for time series analysis (TSA) have attracted significant attention. However, challenges such as training data scarcity and imbalance continue to hinder their development. Inspired by complex dynamic system theories, we design a series-symbol data generation mechanism, enabling the unrestricted creation of high-quality time series data paired with corresponding symbolic expressions. To leverage series-symbol data pairs with strong correlations, we develop SymTime, a pre-trained foundation model for enhancing time series representation using symbolic information. SymTime demonstrates competitive performance across five major TSA tasks when fine-tunes with downstream tasks, rivaling foundation models pre-trained on real-world datasets. This approach underscores the potential of series-symbol data generation and pretraining mechanisms in overcoming data scarcity and enhancing task performance. The code is available at https://github.com/wwhenxuan/SymTime.

---

### Gaze on the Prize: Shaping Visual Attention with Return-Guided Contrastive Learning

Authors: Andrew Lee, Ian Chuang, Dechen Gao, Kai Fukazawa, Iman Soltani
Published: 2025-10-09
Categories: cs.CV, cs.AI, cs.RO
arXiv: https://arxiv.org/abs/2510.08442v2

Visual Reinforcement Learning (RL) agents must learn to act based on high-dimensional image data where only a small fraction of the pixels is task-relevant. This forces agents to waste exploration and computational resources on irrelevant features, leading to sample-inefficient and unstable learning. To address this, inspired by human visual foveation, we introduce Gaze on the Prize. This framework augments visual RL with a learnable foveal attention mechanism (Gaze), guided by a self-supervised signal derived from the agent's experience pursuing higher returns (the Prize). Our key insight is that return differences reveal what matters most: If two similar representations produce different outcomes, their distinguishing features are likely task-relevant, and the gaze should focus on them accordingly. This is realized through return-guided contrastive learning that trains the attention to distinguish between the features relevant to success and failure. We group similar visual representations into positives and negatives based on their return differences and use the resulting labels to construct contrastive triplets. These triplets provide the training signal that teaches the attention mechanism to produce distinguishable representations for states associated with different outcomes. Our method achieves up to 2.52x improvement in sample efficiency and can solve challenging tasks from the ManiSkill3 benchmark that the baseline fails to learn, without modifying the underlying algorithm or hyperparameters.

---

### xRouter: Training Cost-Aware LLMs Orchestration System via Reinforcement Learning

Authors: Cheng Qian, Zuxin Liu, Shirley Kokane, Akshara Prabhakar, Jielin Qiu, Haolin Chen, Zhiwei Liu, Heng Ji, Weiran Yao, Shelby Heinecke, Silvio Savarese, Caiming Xiong et al.
Published: 2025-10-09
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2510.08439v1

Modern LLM deployments confront a widening cost-performance spectrum: premium models deliver strong reasoning but are expensive, while lightweight models are economical yet brittle on complex tasks. Static escalation rules and keyword heuristics under-utilize this spectrum and fail to adapt across task types. We present xRouter, a tool-calling-based routing system in which a learned router can either answer directly or invoke one or more external models. The router is trained end-to-end with reinforcement learning using an explicit, cost-aware reward that encodes cost-performance trade-offs, eliminating the need for hand-engineered routing rules. Our implementation encompasses the full reinforcement learning framework, including reward and cost accounting, as well as the deployment and evaluation pipelines. Across diverse benchmarks, xRouter achieves strong cost-performance trade-offs (e.g., substantial cost reductions at comparable task completion rates), and provides empirical insights into what reliably helps learned routing and what does not, ranging from model trainability to the difficulty of eliciting sophisticated orchestration behaviors in small open models. We hope these findings and our open implementation will serve as a practical substrate for advancing learned, cost-aware LLM orchestration.

---

### Navigating Sparsities in High-Dimensional Linear Contextual Bandits

Authors: Rui Zhao, Zihan Chen, Zemin Zheng
Published: 2025-10-09
Categories: math.ST, cs.LG
arXiv: https://arxiv.org/abs/2510.08435v1

High-dimensional linear contextual bandit problems remain a significant challenge due to the curse of dimensionality. Existing methods typically consider either the model parameters to be sparse or the eigenvalues of context covariance matrices to be (approximately) sparse, lacking general applicability due to the rigidity of conventional reward estimators. To overcome this limitation, a powerful pointwise estimator is introduced in this work that adaptively navigates both kinds of sparsity. Based on this pointwise estimator, a novel algorithm, termed HOPE, is proposed. Theoretical analyses demonstrate that HOPE not only achieves improved regret bounds in previously discussed homogeneous settings (i.e., considering only one type of sparsity) but also, for the first time, efficiently handles two new challenging heterogeneous settings (i.e., considering a mixture of two types of sparsity), highlighting its flexibility and generality. Experiments corroborate the superiority of HOPE over existing methods across various scenarios.

---

### Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency

Authors: Kaiwen Zheng, Yuji Wang, Qianli Ma, Huayu Chen, Jintao Zhang, Yogesh Balaji, Jianfei Chen, Ming-Yu Liu, Jun Zhu, Qinsheng Zhang
Published: 2025-10-09
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2510.08431v3

Although continuous-time consistency models (e.g., sCM, MeanFlow) are theoretically principled and empirically powerful for fast academic-scale diffusion, its applicability to large-scale text-to-image and video tasks remains unclear due to infrastructure challenges in Jacobian-vector product (JVP) computation and the limitations of evaluation benchmarks like FID. This work represents the first effort to scale up continuous-time consistency to general application-level image and video diffusion models, and to make JVP-based distillation effective at large scale. We first develop a parallelism-compatible FlashAttention-2 JVP kernel, enabling sCM training on models with over 10 billion parameters and high-dimensional video tasks. Our investigation reveals fundamental quality limitations of sCM in fine-detail generation, which we attribute to error accumulation and the "mode-covering" nature of its forward-divergence objective. To remedy this, we propose the score-regularized continuous-time consistency model (rCM), which incorporates score distillation as a long-skip regularizer. This integration complements sCM with the "mode-seeking" reverse divergence, effectively improving visual quality while maintaining high generation diversity. Validated on large-scale models (Cosmos-Predict2, Wan2.1) up to 14B parameters and 5-second videos, rCM generally matches the state-of-the-art distillation method DMD2 on quality metrics while mitigating mode collapse and offering notable advantages in diversity, all without GAN tuning or extensive hyperparameter searches. The distilled models generate high-fidelity samples in only $1\sim4$ steps, accelerating diffusion sampling by $15\times\sim50\times$. These results position rCM as a practical and theoretically grounded framework for advancing large-scale diffusion distillation. Code is available at https://github.com/NVlabs/rcm.

---

### ClauseLens: Clause-Grounded, CVaR-Constrained Reinforcement Learning for Trustworthy Reinsurance Pricing

Authors: Stella C. Dong, James R. Finlay
Published: 2025-10-09
Categories: cs.LG, cs.AI, stat.ML
arXiv: https://arxiv.org/abs/2510.08429v1

Reinsurance treaty pricing must satisfy stringent regulatory standards, yet current quoting practices remain opaque and difficult to audit. We introduce ClauseLens, a clause-grounded reinforcement learning framework that produces transparent, regulation-compliant, and risk-aware treaty quotes. ClauseLens models the quoting task as a Risk-Aware Constrained Markov Decision Process (RA-CMDP). Statutory and policy clauses are retrieved from legal and underwriting corpora, embedded into the agent's observations, and used both to constrain feasible actions and to generate clause-grounded natural language justifications. Evaluated in a multi-agent treaty simulator calibrated to industry data, ClauseLens reduces solvency violations by 51%, improves tail-risk performance by 27.9% (CVaR_0.10), and achieves 88.2% accuracy in clause-grounded explanations with retrieval precision of 87.4% and recall of 91.1%. These findings demonstrate that embedding legal context into both decision and explanation pathways yields interpretable, auditable, and regulation-aligned quoting behavior consistent with Solvency II, NAIC RBC, and the EU AI Act.

---

### Reinforcing Diffusion Models by Direct Group Preference Optimization

Authors: Yihong Luo, Tianyang Hu, Jing Tang
Published: 2025-10-09
Categories: cs.LG, cs.CV
arXiv: https://arxiv.org/abs/2510.08425v1

While reinforcement learning methods such as Group Relative Preference Optimization (GRPO) have significantly enhanced Large Language Models, adapting them to diffusion models remains challenging. In particular, GRPO demands a stochastic policy, yet the most cost-effective diffusion samplers are based on deterministic ODEs. Recent work addresses this issue by using inefficient SDE-based samplers to induce stochasticity, but this reliance on model-agnostic Gaussian noise leads to slow convergence. To resolve this conflict, we propose Direct Group Preference Optimization (DGPO), a new online RL algorithm that dispenses with the policy-gradient framework entirely. DGPO learns directly from group-level preferences, which utilize relative information of samples within groups. This design eliminates the need for inefficient stochastic policies, unlocking the use of efficient deterministic ODE samplers and faster training. Extensive results show that DGPO trains around 20 times faster than existing state-of-the-art methods and achieves superior performance on both in-domain and out-of-domain reward metrics. Code is available at https://github.com/Luo-Yihong/DGPO.

---

### Prompts Generalize with Low Data: Non-vacuous Generalization Bounds for Optimizing Prompts with More Informative Priors

Authors: David Madras, Joshua Safyan, Qiuyi, Zhang
Published: 2025-10-09
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.08413v1

Many prompt engineering techniques have been successful in practice, even when optimizing over a large prompt space with with a small amount of task-specific data. Recent work has partially explained this success by showing generalization bounds which apply PAC-Bayes theory to the discrete prompt space, but they are non-vacuous only in data-rich scenarios. We argue that such widespread success can be more fully explained through more carefully considering data- or distribution-dependent perplexity, which acts as an effective prior and steers the optimization towards prompts that are more ``natural'' for the task at hand. We derive novel generalization bounds that are non-vacuous for data-scarce prompt optimization via more useful priors, formally analyzing how perplexity regularization tightens these bounds by limiting exploration. Empirically, we explore both the bounds' effectiveness and the practical benefits of perplexity regularization in improving prompt generalization.

---

### Optimal Stopping in Latent Diffusion Models

Authors: Yu-Han Wu, Quentin Berthet, Gérard Biau, Claire Boyer, Romuald Elie, Pierre Marion
Published: 2025-10-09
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2510.08409v2

We identify and analyze a surprising phenomenon of Latent Diffusion Models (LDMs) where the final steps of the diffusion can degrade sample quality. In contrast to conventional arguments that justify early stopping for numerical stability, this phenomenon is intrinsic to the dimensionality reduction in LDMs. We provide a principled explanation by analyzing the interaction between latent dimension and stopping time. Under a Gaussian framework with linear autoencoders, we characterize the conditions under which early stopping is needed to minimize the distance between generated and target distributions. More precisely, we show that lower-dimensional representations benefit from earlier termination, whereas higher-dimensional latent spaces require later stopping time. We further establish that the latent dimension interplays with other hyperparameters of the problem such as constraints in the parameters of score matching. Experiments on synthetic and real datasets illustrate these properties, underlining that early stopping can improve generative quality. Together, our results offer a theoretical foundation for understanding how the latent dimension influences the sample quality, and highlight stopping time as a key hyperparameter in LDMs.

---

### Biology-driven assessment of deep learning super-resolution imaging of the porosity network in dentin

Authors: Lauren Anderson, Lucas Chatelain, Nicolas Tremblay, Kathryn Grandfield, David Rousseau, Aurélien Gourrier
Published: 2025-10-09
Categories: cs.LG, cs.CV, q-bio.TO
arXiv: https://arxiv.org/abs/2510.08407v1

The mechanosensory system of teeth is currently believed to partly rely on Odontoblast cells stimulation by fluid flow through a porosity network extending through dentin. Visualizing the smallest sub-microscopic porosity vessels therefore requires the highest achievable resolution from confocal fluorescence microscopy, the current gold standard. This considerably limits the extent of the field of view to very small sample regions. To overcome this limitation, we tested different deep learning (DL) super-resolution (SR) models to allow faster experimental acquisitions of lower resolution images and restore optimal image quality by post-processing. Three supervised 2D SR models (RCAN, pix2pix, FSRCNN) and one unsupervised (CycleGAN) were applied to a unique set of experimentally paired high- and low-resolution confocal images acquired with different sampling schemes, resulting in a pixel size increase of x2, x4, x8. Model performance was quantified using a broad set of similarity and distribution-based image quality assessment (IQA) metrics, which yielded inconsistent results that mostly contradicted our visual perception. This raises the question of the relevance of such generic metrics to efficiently target the specific structure of dental porosity. To resolve this conflicting information, the generated SR images were segmented taking into account the specific scales and morphology of the porosity network and analysed by comparing connected components. Additionally, the capacity of the SR models to preserve 3D porosity connectivity throughout the confocal image stacks was evaluated using graph analysis. This biology-driven assessment allowed a far better mechanistic interpretation of SR performance, highlighting differences in model sensitivity to weak intensity features and the impact of non-linearity in image generation, which explains the failure of standard IQA metrics.

---

### Single layer tiny Co$^4$ outpaces GPT-2 and GPT-BERT

Authors: Noor Ul Zain, Mohsin Raza, Ahsan Adeel
Published: 2025-10-09
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2510.08404v1

We show that a tiny Co$^4$ machine(Adeel,2025) with a single layer, two heads, and 8M parameters, operating at an approximate cost of $O(N)$ (where $N$ is the number of input tokens), outpaces the BabyLM Challenge baselines GPT-2 (124M, 12 layers, $O(N^2))$ and GPT-BERT (30M, 12 layers, $O(N^2))$ in just two epochs, while both are trained for ten. Co$^4$ achieves orders-of-magnitude greater training efficiency on 10M tokens, demonstrating highly sample efficient pretraining. Using the BabyLM challenge evaluation pipeline across complex benchmarks, Co$^4$ exhibits strong zero-shot and fine-tuning performance on SuperGLUE tasks. Specifically, Co$^4$ outperforms GPT-2 on 5 out of 7 zero-shot metrics and 6 out of 7 fine-tuning tasks, and GPT-BERT on 4 out of 7 metrics in both cases. These results suggest the need to rethink prevailing deep learning paradigms and associated scaling laws.

---

### VideoVerse: Does Your T2V Generator Have World Model Capability to Synthesize Videos?

Authors: Zeqing Wang, Xinyu Wei, Bairui Li, Zhen Guo, Jinrui Zhang, Hongyang Wei, Keze Wang, Lei Zhang
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.08398v4

The recent rapid advancement of Text-to-Video (T2V) generation technologies are engaging the trained models with more world model ability, making the existing benchmarks increasingly insufficient to evaluate state-of-the-art T2V models. First, current evaluation dimensions, such as per-frame aesthetic quality and temporal consistency, are no longer able to differentiate state-of-the-art T2V models. Second, event-level temporal causality-an essential property that differentiates videos from other modalities-remains largely unexplored. Third, existing benchmarks lack a systematic assessment of world knowledge, which are essential capabilities for building world models. To address these issues, we introduce VideoVerse, a comprehensive benchmark focusing on evaluating whether the current T2V model could understand complex temporal causality and world knowledge to synthesize videos. We collect representative videos across diverse domains and extract their event-level descriptions with inherent temporal causality, which are then rewritten into text-to-video prompts by independent annotators. For each prompt, we design ten evaluation dimensions covering dynamic and static properties, resulting in 300 prompts, 815 events, and 793 evaluation questions. Consequently, a human preference-aligned QA-based evaluation pipeline is developed by using modern vision-language models to systematically benchmark leading open- and closed-source T2V systems, revealing the current gap between T2V models and desired world modeling abilities.

---