# Research Papers — 2025-10 · bundle page 159 of 217

> Full-text markdown of papers #7901–7950 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/159
[← Previous bundle](https://aipod.dev/papers/archive/2025-10/page/158.md)
[Next bundle →](https://aipod.dev/papers/archive/2025-10/page/160.md)

### FedBook: A Unified Federated Graph Foundation Codebook with Intra-domain and Inter-domain Knowledge Modeling

Authors: Zhengyu Wu, Yinlin Zhu, Xunkai Li, Ziang Qiu, Rong-Hua Li, Guoren Wang, Chenghu Zhou
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07755v1

Foundation models have shown remarkable cross-domain generalization in language and vision, inspiring the development of graph foundation models (GFMs). However, existing GFMs typically assume centralized access to multi-domain graphs, which is often infeasible due to privacy and institutional constraints. Federated Graph Foundation Models (FedGFMs) address this limitation, but their effectiveness fundamentally hinges on constructing a robust global codebook that achieves intra-domain coherence by consolidating mutually reinforcing semantics within each domain, while also maintaining inter-domain diversity by retaining heterogeneous knowledge across domains. To this end, we propose FedBook, a unified federated graph foundation codebook that systematically aggregates clients' local codebooks during server-side federated pre-training. FedBook follows a two-phase process: (1) Intra-domain Collaboration, where low-frequency tokens are refined by referencing more semantically reliable high-frequency tokens across clients to enhance domain-specific coherence; and (2) Inter-domain Integration, where client contributions are weighted by the semantic distinctiveness of their codebooks during the aggregation of the global GFM, thereby preserving cross-domain diversity. Extensive experiments on 8 benchmarks across multiple domains and tasks demonstrate that FedBook consistently outperforms 21 baselines, including isolated supervised learning, FL/FGL, federated adaptations of centralized GFMs, and FedGFM techniques.

---

### DEGS: Deformable Event-based 3D Gaussian Splatting from RGB and Event Stream

Authors: Junhao He, Jiaxu Wang, Jia Li, Mingyuan Sun, Qiang Zhang, Jiahang Cao, Ziyi Zhang, Yi Gu, Jingkai Sun, Renjing Xu
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07752v2

Reconstructing Dynamic 3D Gaussian Splatting (3DGS) from low-framerate RGB videos is challenging. This is because large inter-frame motions will increase the uncertainty of the solution space. For example, one pixel in the first frame might have more choices to reach the corresponding pixel in the second frame. Event cameras can asynchronously capture rapid visual changes and are robust to motion blur, but they do not provide color information. Intuitively, the event stream can provide deterministic constraints for the inter-frame large motion by the event trajectories. Hence, combining low-temporal-resolution images with high-framerate event streams can address this challenge. However, it is challenging to jointly optimize Dynamic 3DGS using both RGB and event modalities due to the significant discrepancy between these two data modalities. This paper introduces a novel framework that jointly optimizes dynamic 3DGS from the two modalities. The key idea is to adopt event motion priors to guide the optimization of the deformation fields. First, we extract the motion priors encoded in event streams by using the proposed LoCM unsupervised fine-tuning framework to adapt an event flow estimator to a certain unseen scene. Then, we present the geometry-aware data association method to build the event-Gaussian motion correspondence, which is the primary foundation of the pipeline, accompanied by two useful strategies, namely motion decomposition and inter-frame pseudo-label. Extensive experiments show that our method outperforms existing image and event-based approaches across synthetic and real scenes and prove that our method can effectively optimize dynamic 3DGS with the help of event data.

---

### Calibrating Decision Robustness via Inverse Conformal Risk Control

Authors: Wenbin Zhou, Shixiang Zhu
Published: 2025-10-09
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2510.07750v3

Robust optimization safeguards decisions against uncertainty by optimizing against worst-case scenarios, yet their effectiveness hinges on a prespecified robustness level that is often chosen ad hoc, leading to either insufficient protection or overly conservative and costly solutions. Recent approaches using conformal prediction construct data-driven uncertainty sets with finite-sample coverage guarantees, but they still fix coverage targets a priori and offer little guidance for selecting robustness levels. We propose a new framework that provides distribution-free, finite-sample guarantees on both miscoverage and regret for any family of robust predict-then-optimize policies. Our method constructs valid estimators that trace out the miscoverage--regret Pareto frontier, enabling decision-makers to reliably evaluate and calibrate robustness levels according to their cost--risk preferences. The framework is simple to implement, broadly applicable across classical optimization formulations, and achieves sharper finite-sample performance. This paper offers a principled data-driven methodology for guiding robustness selection and empowers practitioners to balance robustness and conservativeness in high-stakes decision-making.

---

### Haibu Mathematical-Medical Intelligent Agent:Enhancing Large Language Model Reliability in Medical Tasks via Verifiable Reasoning Chains

Authors: Yilun Zhang, Dexing Kong
Published: 2025-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.07748v1

Large Language Models (LLMs) show promise in medicine but are prone to factual and logical errors, which is unacceptable in this high-stakes field. To address this, we introduce the "Haibu Mathematical-Medical Intelligent Agent" (MMIA), an LLM-driven architecture that ensures reliability through a formally verifiable reasoning process. MMIA recursively breaks down complex medical tasks into atomic, evidence-based steps. This entire reasoning chain is then automatically audited for logical coherence and evidence traceability, similar to theorem proving. A key innovation is MMIA's "bootstrapping" mode, which stores validated reasoning chains as "theorems." Subsequent tasks can then be efficiently solved using Retrieval-Augmented Generation (RAG), shifting from costly first-principles reasoning to a low-cost verification model. We validated MMIA across four healthcare administration domains, including DRG/DIP audits and medical insurance adjudication, using expert-validated benchmarks. Results showed MMIA achieved an error detection rate exceeding 98% with a false positive rate below 1%, significantly outperforming baseline LLMs. Furthermore, the RAG matching mode is projected to reduce average processing costs by approximately 85% as the knowledge base matures. In conclusion, MMIA's verifiable reasoning framework is a significant step toward creating trustworthy, transparent, and cost-effective AI systems, making LLM technology viable for critical applications in medicine.

---

### t-SNE Exaggerates Clusters, Provably

Authors: Noah Bergam, Szymon Snoeck, Nakul Verma
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07746v2

Central to the widespread use of t-distributed stochastic neighbor embedding (t-SNE) is the conviction that it produces visualizations whose structure roughly matches that of the input. To the contrary, we prove that (1) the strength of the input clustering, and (2) the extremity of outlier points, cannot be reliably inferred from the t-SNE output. We demonstrate the prevalence of these failure modes in practice as well.

---

### Parallel Test-Time Scaling for Latent Reasoning Models

Authors: Runyang You, Yongqi Li, Meng Liu, Wenjie Wang, Liqiang Nie, Wenjie Li
Published: 2025-10-09
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2510.07745v4

Parallel test-time scaling (TTS) is a pivotal approach for enhancing large language models (LLMs), typically by sampling multiple token-based chains-of-thought in parallel and aggregating outcomes through voting or search. Recent advances in latent reasoning, where intermediate reasoning unfolds in continuous vector spaces, offer a more efficient alternative to explicit Chain-of-Thought, yet whether such latent models can similarly benefit from parallel TTS remains open, mainly due to the absence of sampling mechanisms in continuous space, and the lack of probabilistic signals for advanced trajectory aggregation. This work enables parallel TTS for latent reasoning models by addressing the above issues. For sampling, we introduce two uncertainty-inspired stochastic strategies: Monte Carlo Dropout and Additive Gaussian Noise. For aggregation, we design a Latent Reward Model (LatentRM) trained with step-wise contrastive objective to score and guide latent reasoning. Extensive experiments and visualization analyses show that both sampling strategies scale effectively with compute and exhibit distinct exploration dynamics, while LatentRM enables effective trajectory selection. Together, our explorations open a new direction for scalable inference in continuous spaces. Code and checkpoints released at https://github.com/ModalityDance/LatentTTS

---

### OpenRubrics: Towards Scalable Synthetic Rubric Generation for Reward Modeling and LLM Alignment

Authors: Tianci Liu, Ran Xu, Tony Yu, Ilgee Hong, Carl Yang, Tuo Zhao, Haoyu Wang
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07743v3

Reward modeling lies at the core of reinforcement learning from human feedback (RLHF), yet most existing reward models rely on scalar or pairwise judgments that fail to capture the multifaceted nature of human preferences. Recent studies have explored rubrics-as-rewards (RaR) that uses structured criteria to capture multiple dimensions of response quality. However, producing rubrics that are both reliable and scalable remains a key challenge. In this work, we introduce OpenRubrics, a diverse, large-scale collection of (prompt, rubric) pairs for training rubric-generation and rubric-based reward models. To elicit discriminative and comprehensive evaluation signals, we introduce Contrastive Rubric Generation (CRG), which derives both hard rules (explicit constraints) and principles (implicit qualities) by contrasting preferred and rejected responses. We further remove noisy rubrics via preserving preference-label consistency. Across multiple reward-modeling benchmarks, our rubric-based reward model, Rubric-RM, surpasses strong size-matched baselines by 8.4%. These gains transfer to policy models on instruction-following and biomedical benchmarks.

---

### UltraLED: Learning to See Everything in Ultra-High Dynamic Range Scenes

Authors: Yuang Meng, Xin Jin, Lina Lei, Chun-Le Guo, Chongyi Li
Published: 2025-10-09
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2510.07741v1

Ultra-high dynamic range (UHDR) scenes exhibit significant exposure disparities between bright and dark regions. Such conditions are commonly encountered in nighttime scenes with light sources. Even with standard exposure settings, a bimodal intensity distribution with boundary peaks often emerges, making it difficult to preserve both highlight and shadow details simultaneously. RGB-based bracketing methods can capture details at both ends using short-long exposure pairs, but are susceptible to misalignment and ghosting artifacts. We found that a short-exposure image already retains sufficient highlight detail. The main challenge of UHDR reconstruction lies in denoising and recovering information in dark regions. In comparison to the RGB images, RAW images, thanks to their higher bit depth and more predictable noise characteristics, offer greater potential for addressing this challenge. This raises a key question: can we learn to see everything in UHDR scenes using only a single short-exposure RAW image? In this study, we rely solely on a single short-exposure frame, which inherently avoids ghosting and motion blur, making it particularly robust in dynamic scenes. To achieve that, we introduce UltraLED, a two-stage framework that performs exposure correction via a ratio map to balance dynamic range, followed by a brightness-aware RAW denoiser to enhance detail recovery in dark regions. To support this setting, we design a 9-stop bracketing pipeline to synthesize realistic UHDR images and contribute a corresponding dataset based on diverse scenes, using only the shortest exposure as input for reconstruction. Extensive experiments show that UltraLED significantly outperforms existing single-frame approaches. Our code and dataset are made publicly available at https://srameo.github.io/projects/ultraled.

---

### AppForge: From Assistant to Independent Developer -- Are GPTs Ready for Software Development?

Authors: Dezhi Ran, Yuan Cao, Mengzhou Wu, Simin Chen, Yuzhe Guo, Jun Ren, Zihe Song, Hao Yu, Jialei Wei, Linyi Li, Wei Yang, Baishakhi Ray et al.
Published: 2025-10-09
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2510.07740v1

Large language models (LLMs) have demonstrated remarkable capability in function-level code generation tasks. Unlike isolated functions, real-world applications demand reasoning over the entire software system: developers must orchestrate how different components interact, maintain consistency across states over time, and ensure the application behaves correctly within the lifecycle and framework constraints. Yet, no existing benchmark adequately evaluates whether LLMs can bridge this gap and construct entire software systems from scratch. To address this gap, we propose APPFORGE, a benchmark consisting of 101 software development problems drawn from real-world Android apps. Given a natural language specification detailing the app functionality, a language model is tasked with implementing the functionality into an Android app from scratch. Developing an Android app from scratch requires understanding and coordinating app states, lifecycle management, and asynchronous operations, calling for LLMs to generate context-aware, robust, and maintainable code. To construct APPFORGE, we design a multi-agent system to automatically summarize the main functionalities from app documents and navigate the app to synthesize test cases validating the functional correctness of app implementation. Following rigorous manual verification by Android development experts, APPFORGE incorporates the test cases within an automated evaluation framework that enables reproducible assessment without human intervention, making it easily adoptable for future research. Our evaluation on 12 flagship LLMs show that all evaluated models achieve low effectiveness, with the best-performing model (GPT-5) developing only 18.8% functionally correct applications, highlighting fundamental limitations in current models' ability to handle complex, multi-component software engineering challenges.

---

### MeSH: Memory-as-State-Highways for Recursive Transformers

Authors: Chengting Yu, Xiaobo Shu, Yadao Wang, Yizhen Zhang, Haoyi Wu, Jiaang Li, Rujiao Long, Ziheng Chen, Yuchi Xu, Wenbo Su, Bo Zheng
Published: 2025-10-09
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.07739v2

Recursive transformers reuse parameters and iterate over hidden states multiple times, decoupling compute depth from parameter depth. However, under matched compute, recursive models with fewer parameters often lag behind non-recursive counterparts. By probing hidden states, we trace this performance gap to two primary bottlenecks: undifferentiated computation, where the core is forced to adopt a similar computational pattern at every iteration, and information overload, where long-lived and transient information must coexist in a single hidden state. To address the issues, we introduce a Memory-as-State-Highways (MeSH) scheme, which externalizes state management into an explicit memory buffer and employs lightweight routers to dynamically diversify computation across iterations. Probing visualizations confirm that MeSH successfully resolves the pathologies by inducing functional specialization across iterations. On the Pythia suite (160M-6.9B), MeSH-enhanced recursive transformers consistently improve over recursive baselines and outperforms its larger non-recursive counterpart at the 1.4B scale, improving average downstream accuracy by +1.06% with 33% fewer non-embedding parameters. Our analysis establishes MeSH as a scalable and principled architecture for building stronger recursive models. Our code is available at https://github.com/LivingFutureLab/MeSH/ .

---

### ToolExpander: Extending the Frontiers of Tool-Using Reinforcement Learning to Weak LLMs

Authors: Fu Chen, Peng Wang, Xiyin Li, Wen Li, Shichi Lei, Dongdong Xiang
Published: 2025-10-09
Categories: cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2510.07737v1

Training Large Language Models (LLMs) with Group Relative Policy Optimization (GRPO) encounters a significant challenge: models often fail to produce accurate responses, particularly in small-scale architectures. This limitation not only diminishes performance improvements and undermines the potential of GRPO but also frequently leads to mid-training collapse, adversely affecting stability and final efficacy. To address these issues, we propose ToolExpander, a novel framework that advances tool-oriented reinforcement learning for resource-constrained LLMs through two key innovations:(1) Dynamic Multi-Round Hard Sampling, which dynamically substitutes challenging samples(those without correct outputs over 10 rollouts) with high-quality few-shot demonstrations during training, coupled with an exponential learning rate decay strategy to mitigate oscillations;(2) Self-Exemplifying Thinking, an enhanced GRPO framework that eliminates KL divergence and incorporates adjusted clipping coefficients, encouraging models to autonomously generate and analyze few-shot examples via a minimal additional reward (0.01).Experimental results demonstrate that ToolExpander significantly enhances tool-using capabilities in LLMs, especially in weaker small-scale models, improving both training stability and overall performance.

---

### Multilingual Knowledge Graph Completion via Efficient Multilingual Knowledge Sharing

Authors: Cunli Mao, Xiaofei Gao, Ran Song, Shizhu He, Shengxiang Gao, Kang Liu, Zhengtao Yu
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07736v1

Large language models (LLMs) based Multilingual Knowledge Graph Completion (MKGC) aim to predict missing facts by leveraging LLMs' multilingual understanding capabilities, improving the completeness of multilingual knowledge graphs (KGs). However, existing MKGC research underutilizes the multilingual capabilities of LLMs and ignores the shareability of cross-lingual knowledge. In this paper, we propose a novel MKGC framework that leverages multilingual shared knowledge to significantly enhance performance through two components: Knowledge-level Grouped Mixture of Experts (KL-GMoE) and Iterative Entity Reranking (IER). KL-GMoE efficiently models shared knowledge, while IER significantly enhances its utilization. To evaluate our framework, we constructed a mKG dataset containing 5 languages and conducted comprehensive comparative experiments with existing state-of-the-art (SOTA) MKGC method. The experimental results demonstrate that our framework achieves improvements of 5.47%, 3.27%, and 1.01% in the Hits@1, Hits@3, and Hits@10 metrics, respectively, compared with SOTA MKGC method. Further experimental analysis revealed the properties of knowledge sharing in settings of unseen and unbalanced languages. We have released the dataset and code for our work on https://github.com/gaoxiaofei07/KL-GMoE.

---

### GeoGen: A Two-stage Coarse-to-Fine Framework for Fine-grained Synthetic Location-based Social Network Trajectory Generation

Authors: Rongchao Xu, Kunlin Cai, Lin Jiang, Zhiqing Hong, Yuan Tian, Guang Wang
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07735v3

Location-Based Social Network (LBSN) check-in trajectory data are important for many practical applications, like POI recommendation, advertising, and pandemic intervention. However, the high collection costs and ever-increasing privacy concerns prevent us from accessing large-scale LBSN trajectory data. The recent advances in synthetic data generation provide us with a new opportunity to achieve this, which utilizes generative AI to generate synthetic data that preserves the characteristics of real data while ensuring privacy protection. However, generating synthetic LBSN check-in trajectories remains challenging due to their spatially discrete, temporally irregular nature and the complex spatio-temporal patterns caused by sparse activities and uncertain human mobility. To address this challenge, we propose GeoGen, a two-stage coarse-to-fine framework for large-scale LBSN check-in trajectory generation. In the first stage, we reconstruct spatially continuous, temporally regular latent movement sequences from the original LBSN check-in trajectories and then design a Sparsity-aware Spatio-temporal Diffusion model (S$^2$TDiff) with an efficient denosing network to learn their underlying behavioral patterns. In the second stage, we design Coarse2FineNet, a Transformer-based Seq2Seq architecture equipped with a dynamic context fusion mechanism in the encoder and a multi-task hybrid-head decoder, which generates fine-grained LBSN trajectories based on coarse-grained latent movement sequences by modeling semantic relevance and behavioral uncertainty. Extensive experiments on four real-world datasets show that GeoGen excels state-of-the-art models for both fidelity and utility evaluation, e.g., it increases over 69% and 55% in distance and radius metrics on the FS-TKY dataset.

---

### SurveyG: A Multi-Agent LLM Framework with Hierarchical Citation Graph for Automated Survey Generation

Authors: Minh-Anh Nguye, Minh-Duc Nguyen, Ha Lan N. T., Kieu Hai Dang, Nguyen Tien Dong, Dung D. Le
Published: 2025-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.07733v3

Large language models (LLMs) are increasingly adopted for automating survey paper generation \cite{wang2406autosurvey, liang2025surveyx, yan2025surveyforge,su2025benchmarking,wen2025interactivesurvey}. Existing approaches typically extract content from a large collection of related papers and prompt LLMs to summarize them directly. However, such methods often overlook the structural relationships among papers, resulting in generated surveys that lack a coherent taxonomy and a deeper contextual understanding of research progress. To address these shortcomings, we propose \textbf{SurveyG}, an LLM-based agent framework that integrates \textit{hierarchical citation graph}, where nodes denote research papers and edges capture both citation dependencies and semantic relatedness between their contents, thereby embedding structural and contextual knowledge into the survey generation process. The graph is organized into three layers: \textbf{Foundation}, \textbf{Development}, and \textbf{Frontier}, to capture the evolution of research from seminal works to incremental advances and emerging directions. By combining horizontal search within layers and vertical depth traversal across layers, the agent produces multi-level summaries, which are consolidated into a structured survey outline. A multi-agent validation stage then ensures consistency, coverage, and factual accuracy in generating the final survey. Experiments, including evaluations by human experts and LLM-as-a-judge, demonstrate that SurveyG outperforms state-of-the-art frameworks, producing surveys that are more comprehensive and better structured to the underlying knowledge taxonomy of a field.

---

### Rotated Mean-Field Variational Inference and Iterative Gaussianization

Authors: Yifan Chen, Sifan Liu
Published: 2025-10-09
Categories: stat.CO, stat.ML
arXiv: https://arxiv.org/abs/2510.07732v2

We propose an iterative Gaussianization method for sampling from unnormalized densities by repeatedly applying mean-field variational inference (MFVI) in rotated coordinate systems. At each iteration, the method selects a rotation, solves an MFVI subproblem in the rotated coordinates, and applies the inverse coordinatewise map to transform the current target closer to the standard Gaussian. The resulting algorithm provides a computationally efficient way to construct flow-like transport maps: it requires only MFVI subproblems, avoids large-scale optimization, and produces transformations that are easy to invert and evaluate. The effectiveness of the procedure depends on selecting informative rotations. We develop an efficient PCA-type method that chooses rotations from the leading eigenvectors of a cross-covariance matrix involving the target's score function. Experiments on Bayesian posterior sampling tasks show that performing MFVI in the proposed PCA-rotated coordinate systems substantially improves over standard MFVI, and that the resulting iterative Gaussianization procedure provides accurate flow-like approximations at lower computational cost than conventional normalizing-flow variational approximations.

---

### oMeBench: Towards Robust Benchmarking of LLMs in Organic Mechanism Elucidation and Reasoning

Authors: Ruiling Xu, Yifan Zhang
Published: 2025-10-09
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2510.07731v4

Organic reaction mechanisms describe the step-wise elementary processes by which reactants transform into intermediates and products, and are fundamental to understanding chemical reactivity and guiding molecular and reaction de-sign. While large language models (LLMs) have shown promise on chemical tasks such as synthesis design, it remains unclear to what extent this reflects genuine chemical reasoning capabilities: the ability to generate chemically valid intermediates, maintain consistency across reaction steps, and follow logically coherent multi-step pathways. To investigate this, we introduce oMeBench, the first large-scale, expert-curated benchmark for organic mechanism reasoning, comprising over 10,000 annotated mechanistic steps with reaction type labels, intermediate structures, and difficulty ratings. To enable fine-grained evaluation, we further propose oMeS, a dynamic scoring framework that jointly assesses step-level logical consistency and chemical structural similarity. Systematic evaluation of state-of-the-art LLMs reveals that while current models exhibit promising chemical intuition, they struggle to produce correct and consistent reasoning across multi-step mechanisms. Notably, combining prompting strategies with fine-tuning enables smaller-scale models to achieve performance comparable to closed-source frontier models. We hope oMeBench will serve as a rigorous foundation for advancing AI systems toward genuine chemical reasoning.

---

### DEAS: DEtached value learning with Action Sequence for Scalable Offline RL

Authors: Changyeon Kim, Haeone Lee, Younggyo Seo, Kimin Lee, Yuke Zhu
Published: 2025-10-09
Categories: cs.LG, cs.AI, cs.RO
arXiv: https://arxiv.org/abs/2510.07730v1

Offline reinforcement learning (RL) presents an attractive paradigm for training intelligent agents without expensive online interactions. However, current approaches still struggle with complex, long-horizon sequential decision making. In this work, we introduce DEtached value learning with Action Sequence (DEAS), a simple yet effective offline RL framework that leverages action sequences for value learning. These temporally extended actions provide richer information than single-step actions and can be interpreted through the options framework via semi-Markov decision process Q-learning, enabling reduction of the effective planning horizon by considering longer sequences at once. However, directly adopting such sequences in actor-critic algorithms introduces excessive value overestimation, which we address through detached value learning that steers value estimates toward in-distribution actions that achieve high return in the offline dataset. We demonstrate that DEAS consistently outperforms baselines on complex, long-horizon tasks from OGBench and can be applied to enhance the performance of large-scale Vision-Language-Action models that predict action sequences, significantly boosting performance in both RoboCasa Kitchen simulation tasks and real-world manipulation tasks.

---

### ComGS: Efficient 3D Object-Scene Composition via Surface Octahedral Probes

Authors: Jian Gao, Mengqi Yuan, Yifei Zeng, Chang Zeng, Zhihao Li, Zhenyu Chen, Weichao Qiu, Xiao-Xiao Long, Hao Zhu, Xun Cao, Yao Yao
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07729v1

Gaussian Splatting (GS) enables immersive rendering, but realistic 3D object-scene composition remains challenging. Baked appearance and shadow information in GS radiance fields cause inconsistencies when combining objects and scenes. Addressing this requires relightable object reconstruction and scene lighting estimation. For relightable object reconstruction, existing Gaussian-based inverse rendering methods often rely on ray tracing, leading to low efficiency. We introduce Surface Octahedral Probes (SOPs), which store lighting and occlusion information and allow efficient 3D querying via interpolation, avoiding expensive ray tracing. SOPs provide at least a 2x speedup in reconstruction and enable real-time shadow computation in Gaussian scenes. For lighting estimation, existing Gaussian-based inverse rendering methods struggle to model intricate light transport and often fail in complex scenes, while learning-based methods predict lighting from a single image and are viewpoint-sensitive. We observe that 3D object-scene composition primarily concerns the object's appearance and nearby shadows. Thus, we simplify the challenging task of full scene lighting estimation by focusing on the environment lighting at the object's placement. Specifically, we capture a 360 degrees reconstructed radiance field of the scene at the location and fine-tune a diffusion model to complete the lighting. Building on these advances, we propose ComGS, a novel 3D object-scene composition framework. Our method achieves high-quality, real-time rendering at around 28 FPS, produces visually harmonious results with vivid shadows, and requires only 36 seconds for editing. Code and dataset are available at https://nju-3dv.github.io/projects/ComGS/.

---

### Who Stole Your Data? A Method for Detecting Unauthorized RAG Theft

Authors: Peiyang Liu, Ziqiang Cui, Di Liang, Wei Ye
Published: 2025-10-09
Categories: cs.IR, cs.CL
arXiv: https://arxiv.org/abs/2510.07728v1

Retrieval-augmented generation (RAG) enhances Large Language Models (LLMs) by mitigating hallucinations and outdated information issues, yet simultaneously facilitates unauthorized data appropriation at scale. This paper addresses this challenge through two key contributions. First, we introduce RPD, a novel dataset specifically designed for RAG plagiarism detection that encompasses diverse professional domains and writing styles, overcoming limitations in existing resources. Second, we develop a dual-layered watermarking system that embeds protection at both semantic and lexical levels, complemented by an interrogator-detective framework that employs statistical hypothesis testing on accumulated evidence. Extensive experimentation demonstrates our approach's effectiveness across varying query volumes, defense prompts, and retrieval parameters, while maintaining resilience against adversarial evasion techniques. This work establishes a foundational framework for intellectual property protection in retrieval-augmented AI systems.

---

### SyncHuman: Synchronizing 2D and 3D Generative Models for Single-view Human Reconstruction

Authors: Wenyue Chen, Peng Li, Wangguandong Zheng, Chengfeng Zhao, Mengfei Li, Yaolong Zhu, Zhiyang Dou, Ronggang Wang, Yuan Liu
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07723v2

Photorealistic 3D full-body human reconstruction from a single image is a critical yet challenging task for applications in films and video games due to inherent ambiguities and severe self-occlusions. While recent approaches leverage SMPL estimation and SMPL-conditioned image generative models to hallucinate novel views, they suffer from inaccurate 3D priors estimated from SMPL meshes and have difficulty in handling difficult human poses and reconstructing fine details. In this paper, we propose SyncHuman, a novel framework that combines 2D multiview generative model and 3D native generative model for the first time, enabling high-quality clothed human mesh reconstruction from single-view images even under challenging human poses. Multiview generative model excels at capturing fine 2D details but struggles with structural consistency, whereas 3D native generative model generates coarse yet structurally consistent 3D shapes. By integrating the complementary strengths of these two approaches, we develop a more effective generation framework. Specifically, we first jointly fine-tune the multiview generative model and the 3D native generative model with proposed pixel-aligned 2D-3D synchronization attention to produce geometrically aligned 3D shapes and 2D multiview images. To further improve details, we introduce a feature injection mechanism that lifts fine details from 2D multiview images onto the aligned 3D shapes, enabling accurate and high-fidelity reconstruction. Extensive experiments demonstrate that SyncHuman achieves robust and photo-realistic 3D human reconstruction, even for images with challenging poses. Our method outperforms baseline methods in geometric accuracy and visual fidelity, demonstrating a promising direction for future 3D generation models.

---

### RePainter: Empowering E-commerce Object Removal via Spatial-matting Reinforcement Learning

Authors: Zipeng Guo, Lichen Ma, Xiaolong Fu, Gaojing Zhou, Lan Yang, Yuchen Zhou, Linkai Liu, Yu He, Ximan Liu, Shiping Dong, Jingling Fu, Zhen Chen et al.
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07721v1

In web data, product images are central to boosting user engagement and advertising efficacy on e-commerce platforms, yet the intrusive elements such as watermarks and promotional text remain major obstacles to delivering clear and appealing product visuals. Although diffusion-based inpainting methods have advanced, they still face challenges in commercial settings due to unreliable object removal and limited domain-specific adaptation. To tackle these challenges, we propose Repainter, a reinforcement learning framework that integrates spatial-matting trajectory refinement with Group Relative Policy Optimization (GRPO). Our approach modulates attention mechanisms to emphasize background context, generating higher-reward samples and reducing unwanted object insertion. We also introduce a composite reward mechanism that balances global, local, and semantic constraints, effectively reducing visual artifacts and reward hacking. Additionally, we contribute EcomPaint-100K, a high-quality, large-scale e-commerce inpainting dataset, and a standardized benchmark EcomPaint-Bench for fair evaluation. Extensive experiments demonstrate that Repainter significantly outperforms state-of-the-art methods, especially in challenging scenes with intricate compositions. We will release our code and weights upon acceptance.

---

### SUBQRAG: Sub-Question Driven Dynamic Graph RAG

Authors: Jiaoyang Li, Junhao Ruan, Shengwei Tang, Saihan Chen, Kaiyan Chang, Yuan Ge, Tong Xiao, Jingbo Zhu
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07718v2

Graph Retrieval-Augmented Generation (Graph RAG) effectively builds a knowledge graph (KG) to connect disparate facts across a large document corpus. However, this broad-view approach often lacks the deep structured reasoning needed for complex multi-hop question answering (QA), leading to incomplete evidence and error accumulation. To address these limitations, we propose SubQRAG, a sub-question-driven framework that enhances reasoning depth. SubQRAG decomposes a complex question into an ordered chain of verifiable sub-questions. For each sub-question, it retrieves relevant triples from the graph. When the existing graph is insufficient, the system dynamically expands it by extracting new triples from source documents in real time. All triples used in the reasoning process are aggregated into a "graph memory," forming a structured and traceable evidence path for final answer generation. Experiments on three multi-hop QA benchmarks demonstrate that SubQRAG achieves consistent and significant improvements, especially in Exact Match scores.

---

### Computationally-efficient Graph Modeling with Refined Graph Random Features

Authors: Krzysztof Choromanski, Avinava Dubey, Arijit Sehanobish, Isaac Reid
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07716v2

We propose refined GRFs (GRFs++), a new class of Graph Random Features (GRFs) for efficient and accurate computations involving kernels defined on the nodes of a graph. GRFs++ resolve some of the long-standing limitations of regular GRFs, including difficulty modeling relationships between more distant nodes. They reduce dependence on sampling long graph random walks via a novel walk-stitching technique, concatenating several shorter walks without breaking unbiasedness. By applying these techniques, GRFs++ inherit the approximation quality provided by longer walks but with greater efficiency, trading sequential, inefficient sampling of a long walk for parallel computation of short walks and matrix-matrix multiplication. Furthermore, GRFs++ extend the simplistic GRFs walk termination mechanism (Bernoulli schemes with fixed halting probabilities) to a broader class of strategies, applying general distributions on the walks' lengths. This improves the approximation accuracy of graph kernels, without incurring extra computational cost. We provide empirical evaluations to showcase all our claims and complement our results with theoretical analysis.

---

### Control Synthesis of Cyber-Physical Systems for Real-Time Specifications through Causation-Guided Reinforcement Learning

Authors: Xiaochen Tang, Zhenya Zhang, Miaomiao Zhang, Jie An
Published: 2025-10-09
Categories: cs.AI
arXiv: https://arxiv.org/abs/2510.07715v1

In real-time and safety-critical cyber-physical systems (CPSs), control synthesis must guarantee that generated policies meet stringent timing and correctness requirements under uncertain and dynamic conditions. Signal temporal logic (STL) has emerged as a powerful formalism of expressing real-time constraints, with its semantics enabling quantitative assessment of system behavior. Meanwhile, reinforcement learning (RL) has become an important method for solving control synthesis problems in unknown environments. Recent studies incorporate STL-based reward functions into RL to automatically synthesize control policies. However, the automatically inferred rewards obtained by these methods represent the global assessment of a whole or partial path but do not accumulate the rewards of local changes accurately, so the sparse global rewards may lead to non-convergence and unstable training performances. In this paper, we propose an online reward generation method guided by the online causation monitoring of STL. Our approach continuously monitors system behavior against an STL specification at each control step, computing the quantitative distance toward satisfaction or violation and thereby producing rewards that reflect instantaneous state dynamics. Additionally, we provide a smooth approximation of the causation semantics to overcome the discontinuity of the causation semantics and make it differentiable for using deep-RL methods. We have implemented a prototype tool and evaluated it in the Gym environment on a variety of continuously controlled benchmarks. Experimental results show that our proposed STL-guided RL method with online causation semantics outperforms existing relevant STL-guided RL methods, providing a more robust and efficient reward generation framework for deep-RL.

---

### MemWeaver: A Hierarchical Memory from Textual Interactive Behaviors for Personalized Generation

Authors: Shuo Yu, Mingyue Cheng, Daoyu Wang, Qi Liu, Zirui Liu, Ze Guo, Xiaoyu Tao
Published: 2025-10-09
Categories: cs.CL
arXiv: https://arxiv.org/abs/2510.07713v2

The primary form of user-internet engagement is shifting from leveraging implicit feedback signals, such as browsing and clicks, to harnessing the rich explicit feedback provided by textual interactive behaviors. This shift unlocks a rich source of user textual history, presenting a profound opportunity for a deeper form of personalization. However, prevailing approaches offer only a shallow form of personalization, as they treat user history as a flat list of texts for retrieval and fail to model the rich temporal and semantic structures reflecting dynamic nature of user interests. In this work, we propose \textbf{MemWeaver}, a framework that weaves the user's entire textual history into a hierarchical memory to power deeply personalized generation. The core innovation of our memory lies in its ability to capture both the temporal evolution of interests and the semantic relationships between different activities. To achieve this, MemWeaver builds two complementary memory components that both integrate temporal and semantic information, but at different levels of abstraction: behavioral memory, which captures specific user actions, and cognitive memory, which represents long-term preferences. This dual-component memory serves as a comprehensive representation of the user, allowing large language models (LLMs) to reason over both concrete behaviors and abstracted cognitive traits. This leads to content generation that is deeply aligned with their latent preferences. Experiments on the six datasets of the Language Model Personalization (LaMP) benchmark validate the efficacy of MemWeaver. Our code is available.

---

### Multimodal Safety Evaluation in Generative Agent Social Simulations

Authors: Alhim Vera, Karen Sanchez, Carlos Hinojosa, Haidar Bin Hamid, Donghoon Kim, Bernard Ghanem
Published: 2025-10-09
Categories: cs.AI, cs.CL, cs.CY, cs.MA
arXiv: https://arxiv.org/abs/2510.07709v1

Can generative agents be trusted in multimodal environments? Despite advances in large language and vision-language models that enable agents to act autonomously and pursue goals in rich settings, their ability to reason about safety, coherence, and trust across modalities remains limited. We introduce a reproducible simulation framework for evaluating agents along three dimensions: (1) safety improvement over time, including iterative plan revisions in text-visual scenarios; (2) detection of unsafe activities across multiple categories of social situations; and (3) social dynamics, measured as interaction counts and acceptance ratios of social exchanges. Agents are equipped with layered memory, dynamic planning, multimodal perception, and are instrumented with SocialMetrics, a suite of behavioral and structural metrics that quantifies plan revisions, unsafe-to-safe conversions, and information diffusion across networks. Experiments show that while agents can detect direct multimodal contradictions, they often fail to align local revisions with global safety, reaching only a 55 percent success rate in correcting unsafe plans. Across eight simulation runs with three models - Claude, GPT-4o mini, and Qwen-VL - five agents achieved average unsafe-to-safe conversion rates of 75, 55, and 58 percent, respectively. Overall performance ranged from 20 percent in multi-risk scenarios with GPT-4o mini to 98 percent in localized contexts such as fire/heat with Claude. Notably, 45 percent of unsafe actions were accepted when paired with misleading visuals, showing a strong tendency to overtrust images. These findings expose critical limitations in current architectures and provide a reproducible platform for studying multimodal safety, coherence, and social dynamics.

---

### Causality Guided Representation Learning for Cross-Style Hate Speech Detection

Authors: Chengshuai Zhao, Shu Wan, Paras Sheth, Karan Patwa, K. Selçuk Candan, Huan Liu
Published: 2025-10-09
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2510.07707v2

The proliferation of online hate speech poses a significant threat to the harmony of the web. While explicit hate is easily recognized through overt slurs, implicit hate speech is often conveyed through sarcasm, irony, stereotypes, or coded language -- making it harder to detect. Existing hate speech detection models, which predominantly rely on surface-level linguistic cues, fail to generalize effectively across diverse stylistic variations. Moreover, hate speech spread on different platforms often targets distinct groups and adopts unique styles, potentially inducing spurious correlations between them and labels, further challenging current detection approaches. Motivated by these observations, we hypothesize that the generation of hate speech can be modeled as a causal graph involving key factors: contextual environment, creator motivation, target, and style. Guided by this graph, we propose CADET, a causal representation learning framework that disentangles hate speech into interpretable latent factors and then controls confounders, thereby isolating genuine hate intent from superficial linguistic cues. Furthermore, CADET allows counterfactual reasoning by intervening on style within the latent space, naturally guiding the model to robustly identify hate speech in varying forms. CADET demonstrates superior performance in comprehensive experiments, highlighting the potential of causal priors in advancing generalizable hate speech detection.

---

### Large Language Models Meet Virtual Cell: A Survey

Authors: Krinos Li, Xianglu Xiao, Shenglong Deng, Lucas He, Zijun Zhong, Yuanjie Zou, Zhonghao Zhan, Zheng Hui, Weiye Bao, Guang Yang
Published: 2025-10-09
Categories: cs.CL, cs.CE, cs.LG, q-bio.CB
arXiv: https://arxiv.org/abs/2510.07706v1

Large language models (LLMs) are transforming cellular biology by enabling the development of "virtual cells"--computational systems that represent, predict, and reason about cellular states and behaviors. This work provides a comprehensive review of LLMs for virtual cell modeling. We propose a unified taxonomy that organizes existing methods into two paradigms: LLMs as Oracles, for direct cellular modeling, and LLMs as Agents, for orchestrating complex scientific tasks. We identify three core tasks--cellular representation, perturbation prediction, and gene regulation inference--and review their associated models, datasets, evaluation benchmarks, as well as the critical challenges in scalability, generalizability, and interpretability.

---

### Mutual Learning for Hashing: Unlocking Strong Hash Functions from Weak Supervision

Authors: Xiaoxu Ma, Runhao Li, Zhenyu Weng
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07703v1

Deep hashing has been widely adopted for large-scale image retrieval, with numerous strategies proposed to optimize hash function learning. Pairwise-based methods are effective in learning hash functions that preserve local similarity relationships, whereas center-based methods typically achieve superior performance by more effectively capturing global data distributions. However, the strength of center-based methods in modeling global structures often comes at the expense of underutilizing important local similarity information. To address this limitation, we propose Mutual Learning for Hashing (MLH), a novel weak-to-strong framework that enhances a center-based hashing branch by transferring knowledge from a weaker pairwise-based branch. MLH consists of two branches: a strong center-based branch and a weaker pairwise-based branch. Through an iterative mutual learning process, the center-based branch leverages local similarity cues learned by the pairwise-based branch. Furthermore, inspired by the mixture-of-experts paradigm, we introduce a novel mixture-of-hash-experts module that enables effective cross-branch interaction, further enhancing the performance of both branches. Extensive experiments demonstrate that MLH consistently outperforms state-of-the-art hashing methods across multiple benchmark datasets.

---

### Rethinking Reasoning: A Survey on Reasoning-based Backdoors in LLMs

Authors: Man Hu, Xinyi Wu, Zuofeng Suo, Jinbo Feng, Linghui Meng, Yanhao Jia, Anh Tuan Luu, Shuai Zhao
Published: 2025-10-09
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2510.07697v1

With the rise of advanced reasoning capabilities, large language models (LLMs) are receiving increasing attention. However, although reasoning improves LLMs' performance on downstream tasks, it also introduces new security risks, as adversaries can exploit these capabilities to conduct backdoor attacks. Existing surveys on backdoor attacks and reasoning security offer comprehensive overviews but lack in-depth analysis of backdoor attacks and defenses targeting LLMs' reasoning abilities. In this paper, we take the first step toward providing a comprehensive review of reasoning-based backdoor attacks in LLMs by analyzing their underlying mechanisms, methodological frameworks, and unresolved challenges. Specifically, we introduce a new taxonomy that offers a unified perspective for summarizing existing approaches, categorizing reasoning-based backdoor attacks into associative, passive, and active. We also present defense strategies against such attacks and discuss current challenges alongside potential directions for future research. This work offers a novel perspective, paving the way for further exploration of secure and trustworthy LLM communities.

---

### Hybrid CNN-BYOL Approach for Fault Detection in Induction Motors Using Thermal Images

Authors: Tangin Amir Smrity, MD Zahin Muntaqim Hasan Muhammad Kafi, Abu Saleh Musa Miah, Najmul Hassan, Yuichi Okuyama, Nobuyoshi Asai, Taro Suzuki, Jungpil Shin
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07692v1

Induction motors (IMs) are indispensable in industrial and daily life, but they are susceptible to various faults that can lead to overheating, wasted energy consumption, and service failure. Early detection of faults is essential to protect the motor and prolong its lifespan. This paper presents a hybrid method that integrates BYOL with CNNs for classifying thermal images of induction motors for fault detection. The thermal dataset used in this work includes different operating states of the motor, such as normal operation, overload, and faults. We employed multiple deep learning (DL) models for the BYOL technique, ranging from popular architectures such as ResNet-50, DenseNet-121, DenseNet-169, EfficientNetB0, VGG16, and MobileNetV2. Additionally, we introduced a new high-performance yet lightweight CNN model named BYOL-IMNet, which comprises four custom-designed blocks tailored for fault classification in thermal images. Our experimental results demonstrate that the proposed BYOL-IMNet achieves 99.89\% test accuracy and an inference time of 5.7 ms per image, outperforming state-of-the-art models. This study highlights the promising performance of the CNN-BYOL hybrid method in enhancing accuracy for detecting faults in induction motors, offering a robust methodology for online monitoring in industrial settings.

---

### Stress-Testing Model Specs Reveals Character Differences among Language Models

Authors: Jifan Zhang, Henry Sleight, Andi Peng, John Schulman, Esin Durmus
Published: 2025-10-09
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.07686v2

Large language models (LLMs) are increasingly trained from AI constitutions and model specifications that establish behavioral guidelines and ethical principles. However, these specifications face critical challenges, including internal conflicts between principles and insufficient coverage of nuanced scenarios. We present a systematic methodology for stress-testing model character specifications, automatically identifying numerous cases of principle contradictions and interpretive ambiguities in current model specs. We stress test current model specs by generating scenarios that force explicit tradeoffs between competing value-based principles. Using a comprehensive taxonomy we generate diverse value tradeoff scenarios where models must choose between pairs of legitimate principles that cannot be simultaneously satisfied. We evaluate responses from twelve frontier LLMs across major providers (Anthropic, OpenAI, Google, xAI) and measure behavioral disagreement through value classification scores. Among these scenarios, we identify over 70,000 cases exhibiting significant behavioral divergence. Empirically, we show this high divergence in model behavior strongly predicts underlying problems in model specifications. Through qualitative analysis, we provide numerous example issues in current model specs such as direct contradiction and interpretive ambiguities of several principles. Additionally, our generated dataset also reveals both clear misalignment cases and false-positive refusals across all of the frontier models we study. Lastly, we also provide value prioritization patterns and differences of these models.

---

### LiveThinking: Enabling Real-Time Efficient Reasoning for AI-Powered Livestreaming via Reinforcement Learning

Authors: Yuhan Sun, Zhiwei Huang, Wanqing Cui, Shaopan Xiong, Yazhi Guo, Meiguang Jin, Junfeng Ma
Published: 2025-10-09
Categories: cs.LG, cs.CL
arXiv: https://arxiv.org/abs/2510.07685v1

In AI-powered e-commerce livestreaming, digital avatars require real-time responses to drive engagement, a task for which high-latency Large Reasoning Models (LRMs) are ill-suited. We introduce LiveThinking, a practical two-stage optimization framework to bridge this gap. First, we address computational cost by distilling a 670B teacher LRM into a lightweight 30B Mixture-of-Experts (MoE) model (3B active) using Rejection Sampling Fine-Tuning (RFT). This reduces deployment overhead but preserves the teacher's verbose reasoning, causing latency. To solve this, our second stage employs reinforcement learning with Group Relative Policy Optimization (GRPO) to compress the model's reasoning path, guided by a multi-objective reward function balancing correctness, helpfulness, and brevity. LiveThinking achieves a 30-fold reduction in computational cost, enabling sub-second latency. In real-world application on Taobao Live, it improved response correctness by 3.3% and helpfulness by 21.8%. Tested by hundreds of thousands of viewers, our system led to a statistically significant increase in Gross Merchandise Volume (GMV), demonstrating its effectiveness in enhancing user experience and commercial performance in live, interactive settings.

---

### Curriculum Learning with Synthetic Data for Enhanced Pulmonary Nodule Detection in Chest Radiographs

Authors: Pranav Sambhu, Om Guin, Madhav Sambhu, Jinho Cha
Published: 2025-10-09
Categories: eess.IV, cs.AI, cs.CV
arXiv: https://arxiv.org/abs/2510.07681v2

This study evaluates whether integrating curriculum learning with diffusion-based synthetic augmentation can enhance the detection of difficult pulmonary nodules in chest radiographs, particularly those with low size, brightness, and contrast, which often challenge conventional AI models due to data imbalance and limited annotation. A Faster R-CNN with a Feature Pyramid Network (FPN) backbone was trained on a hybrid dataset comprising expert-labeled NODE21 (1,213 patients; 52.4 percent male; mean age 63.2 +/- 11.5 years), VinDr-CXR, CheXpert, and 11,206 DDPM-generated synthetic images. Difficulty scores based on size, brightness, and contrast guided curriculum learning. Performance was compared to a non-curriculum baseline using mean average precision (mAP), Dice score, and area under the curve (AUC). Statistical tests included bootstrapped confidence intervals, DeLong tests, and paired t-tests. The curriculum model achieved a mean AUC of 0.95 versus 0.89 for the baseline (p < 0.001), with improvements in sensitivity (70 percent vs. 48 percent) and accuracy (82 percent vs. 70 percent). Stratified analysis demonstrated consistent gains across all difficulty bins (Easy to Very Hard). Grad-CAM visualizations confirmed more anatomically focused attention under curriculum learning. These results suggest that curriculum-guided synthetic augmentation enhances model robustness and generalization for pulmonary nodule detection.

---

### Ctrl-VI: Controllable Video Synthesis via Variational Inference

Authors: Haoyi Duan, Yunzhi Zhang, Yilun Du, Jiajun Wu
Published: 2025-10-09
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2510.07670v2

Many video workflows benefit from a mixture of user controls with varying granularity, from exact 4D object trajectories and camera paths to coarse text prompts, while existing video generative models are typically trained for fixed input formats. We develop Ctrl-VI, a video synthesis method that addresses this need and generates samples with high controllability for specified elements while maintaining diversity for under-specified ones. We cast the task as variational inference to approximate a composed distribution, leveraging multiple video generation backbones to account for all task constraints collectively. To address the optimization challenge, we break down the problem into step-wise KL divergence minimization over an annealed sequence of distributions, and further propose a context-conditioned factorization technique that reduces modes in the solution space to circumvent local optima. Experiments suggest that our method produces samples with improved controllability, diversity, and 3D consistency compared to prior works.

---

### iPEAR: Iterative Pyramid Estimation with Attention and Residuals for Deformable Medical Image Registration

Authors: Heming Wu, Di Wang, Tai Ma, Peng Zhao, Yubin Xiao, Zhongke Wu, Xing-Ce Wang, Xuan Wu, You Zhou
Published: 2025-10-09
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2510.07666v3

Existing pyramid registration networks may accumulate anatomical misalignments and lack an effective mechanism to dynamically determine the number of optimization iterations under varying deformation requirements across images, leading to degraded performance. To solve these limitations, we propose iPEAR. Specifically, iPEAR adopts our proposed Fused Attention-Residual Module (FARM) for decoding, which comprises an attention pathway and a residual pathway to alleviate the accumulation of anatomical misalignment. We further propose a dual-stage Threshold-Controlled Iterative (TCI) strategy that adaptively determines the number of optimization iterations for varying images by evaluating registration stability and convergence. Extensive experiments on three public brain MRI datasets and one public abdomen CT dataset show that iPEAR outperforms state-of-the-art (SOTA) registration networks in terms of accuracy, while achieving on-par inference speed and model parameter size. Generalization and ablation studies further validate the effectiveness of the proposed FARM and TCI.

---

### Automatic Text Box Placement for Supporting Typographic Design

Authors: Jun Muraoka, Daichi Haraguchi, Naoto Inoue, Wataru Shimoda, Kota Yamaguchi, Seiichi Uchida
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07665v1

In layout design for advertisements and web pages, balancing visual appeal and communication efficiency is crucial. This study examines automated text box placement in incomplete layouts, comparing a standard Transformer-based method, a small Vision and Language Model (Phi3.5-vision), a large pretrained VLM (Gemini), and an extended Transformer that processes multiple images. Evaluations on the Crello dataset show the standard Transformer-based models generally outperform VLM-based approaches, particularly when incorporating richer appearance information. However, all methods face challenges with very small text or densely populated layouts. These findings highlight the benefits of task-specific architectures and suggest avenues for further improvement in automated layout design.

---

### FedQS: Optimizing Gradient and Model Aggregation for Semi-Asynchronous Federated Learning

Authors: Yunbo Li, Jiaping Gui, Zhihang Deng, Fanchao Meng, Yue Wu
Published: 2025-10-09
Categories: cs.LG, cs.DC
arXiv: https://arxiv.org/abs/2510.07664v2

Federated learning (FL) enables collaborative model training across multiple parties without sharing raw data, with semi-asynchronous FL (SAFL) emerging as a balanced approach between synchronous and asynchronous FL. However, SAFL faces significant challenges in optimizing both gradient-based (e.g., FedSGD) and model-based (e.g., FedAvg) aggregation strategies, which exhibit distinct trade-offs in accuracy, convergence speed, and stability. While gradient aggregation achieves faster convergence and higher accuracy, it suffers from pronounced fluctuations, whereas model aggregation offers greater stability but slower convergence and suboptimal accuracy. This paper presents FedQS, the first framework to theoretically analyze and address these disparities in SAFL. FedQS introduces a divide-and-conquer strategy to handle client heterogeneity by classifying clients into four distinct types and adaptively optimizing their local training based on data distribution characteristics and available computational resources. Extensive experiments on computer vision, natural language processing, and real-world tasks demonstrate that FedQS achieves the highest accuracy, attains the lowest loss, and ranks among the fastest in convergence speed, outperforming state-of-the-art baselines. Our work bridges the gap between aggregation strategies in SAFL, offering a unified solution for stable, accurate, and efficient federated learning. The code and datasets are available at https://github.com/bkjod/FedQS_.

---

### Incremental Hybrid Ensemble with Graph Attention and Frequency-Domain Features for Stable Long-Term Credit Risk Modeling

Authors: Jiajing Wang
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07663v1

Predicting long-term loan defaults is hard because borrower behavior often changes and data distributions shift over time. This paper presents HYDRA-EI, a hybrid ensemble incremental learning framework. It uses several stages of feature processing and combines multiple models. The framework builds relational, cross, and frequency-based features. It uses graph attention, automatic cross-feature creation, and transformations from the frequency domain. HYDRA-EI updates weekly using new data and adjusts the model weights with a simple performance-based method. It works without frequent manual changes or fixed retraining. HYDRA-EI improves model stability and generalization, which makes it useful for long-term credit risk tasks.

---

### Textual Entailment is not a Better Bias Metric than Token Probability

Authors: Virginia K. Felkner, Allison Lim, Jonathan May
Published: 2025-10-09
Categories: cs.CL, cs.CY
arXiv: https://arxiv.org/abs/2510.07662v2

Measurement of social bias in language models is typically by token probability (TP) metrics, which are broadly applicable but have been criticized for their distance from real-world language model use cases and harms. In this work, we test natural language inference (NLI) as an alternative bias metric. In extensive experiments across seven LM families, we show that NLI and TP bias evaluation behave substantially differently, with very low correlation among different NLI metrics and between NLI and TP metrics. NLI metrics are more brittle and unstable, slightly less sensitive to wording of counterstereotypical sentences, and slightly more sensitive to wording of tested stereotypes than TP approaches. Given this conflicting evidence, we conclude that neither token probability nor natural language inference is a ``better'' bias metric in all cases. We do not find sufficient evidence to justify NLI as a complete replacement for TP metrics in bias evaluation.

---

### IKNet: Interpretable Stock Price Prediction via Keyword-Guided Integration of News and Technical Indicators

Authors: Jinwoong Kim, Sangjin Park
Published: 2025-10-09
Categories: cs.CE, cs.AI
arXiv: https://arxiv.org/abs/2510.07661v2

The increasing influence of unstructured external information, such as news articles, on stock prices has attracted growing attention in financial markets. Despite recent advances, most existing newsbased forecasting models represent all articles using sentiment scores or average embeddings that capture the general tone but fail to provide quantitative, context-aware explanations of the impacts of public sentiment on predictions. To address this limitation, we propose an interpretable keyword-guided network (IKNet), which is an explainable forecasting framework that models the semantic association between individual news keywords and stock price movements. The IKNet identifies salient keywords via FinBERTbased contextual analysis, processes each embedding through a separate nonlinear projection layer, and integrates their representations with the time-series data of technical indicators to forecast next-day closing prices. By applying Shapley Additive Explanations the model generates quantifiable and interpretable attributions for the contribution of each keyword to predictions. Empirical evaluations of S&P 500 data from 2015 to 2024 demonstrate that IKNet outperforms baselines, including recurrent neural networks and transformer models, reducing RMSE by up to 32.9% and improving cumulative returns by 18.5%. Moreover, IKNet enhances transparency by offering contextualized explanations of volatility events driven by public sentiment.

---

### MONKEY: Masking ON KEY-Value Activation Adapter for Personalization

Authors: James Baker
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07656v3

Personalizing diffusion models allows users to generate new images that incorporate a given subject, allowing more control than a text prompt. These models often suffer somewhat when they end up just recreating the subject image and ignoring the text prompt. We observe that one popular method for personalization, IP-Adapter, automatically generates masks that segment the subject from the background during inference. We propose to use this automatically generated mask on a second pass to mask the image tokens, thus restricting them to the subject, not the background, allowing the text prompt to attend to the rest of the image. For text prompts describing locations and places, this produces images that accurately depict the subject while definitively matching the prompt. We compare our method to a few other test time personalization methods, and find our method displays high prompt and source image alignment. We also perform a user study to validate whether end users would appreciate our method. Code available at https://github.com/jamesBaker361/monkey

---

### Once Is Enough: Lightweight DiT-Based Video Virtual Try-On via One-Time Garment Appearance Injection

Authors: Yanjie Pan, Qingdong He, Lidong Wang, Bo Peng, Mingmin Chi
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07654v1

Video virtual try-on aims to replace the clothing of a person in a video with a target garment. Current dual-branch architectures have achieved significant success in diffusion models based on the U-Net; however, adapting them to diffusion models built upon the Diffusion Transformer remains challenging. Initially, introducing latent space features from the garment reference branch requires adding or modifying the backbone network, leading to a large number of trainable parameters. Subsequently, the latent space features of garments lack inherent temporal characteristics and thus require additional learning. To address these challenges, we propose a novel approach, OIE (Once is Enough), a virtual try-on strategy based on first-frame clothing replacement: specifically, we employ an image-based clothing transfer model to replace the clothing in the initial frame, and then, under the content control of the edited first frame, utilize pose and mask information to guide the temporal prior of the video generation model in synthesizing the remaining frames sequentially. Experiments show that our method achieves superior parameter efficiency and computational efficiency while still maintaining leading performance under these constraints.

---

### Dual-Stream Alignment for Action Segmentation

Authors: Harshala Gammulle, Clinton Fookes, Sridha Sridharan, Simon Denman
Published: 2025-10-09
Categories: cs.CV
arXiv: https://arxiv.org/abs/2510.07652v1

Action segmentation is a challenging yet active research area that involves identifying when and where specific actions occur in continuous video streams. Most existing work has focused on single-stream approaches that model the spatio-temporal aspects of frame sequences. However, recent research has shifted toward two-stream methods that learn action-wise features to enhance action segmentation performance. In this work, we propose the Dual-Stream Alignment Network (DSA Net) and investigate the impact of incorporating a second stream of learned action features to guide segmentation by capturing both action and action-transition cues. Communication between the two streams is facilitated by a Temporal Context (TC) block, which fuses complementary information using cross-attention and Quantum-based Action-Guided Modulation (Q-ActGM), enhancing the expressive power of the fused features. To the best of our knowledge, this is the first study to introduce a hybrid quantum-classical machine learning framework for action segmentation. Our primary objective is for the two streams (frame-wise and action-wise) to learn a shared feature space through feature alignment. This is encouraged by the proposed Dual-Stream Alignment Loss, which comprises three components: relational consistency, cross-level contrastive, and cycle-consistency reconstruction losses. Following prior work, we evaluate DSA Net on several diverse benchmark datasets: GTEA, Breakfast, 50Salads, and EgoProcel. We further demonstrate the effectiveness of each component through extensive ablation studies. Notably, DSA Net achieves state-of-the-art performance, significantly outperforming existing

---

### OBCache: Optimal Brain KV Cache Pruning for Efficient Long-Context LLM Inference

Authors: Yuzhe Gu, Xiyu Liang, Jiaojiao Zhao, Enmao Diao
Published: 2025-10-09
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.07651v4

Large language models (LLMs) with extended context windows enable powerful applications but impose significant memory overhead, as caching all key-value (KV) states scales linearly with sequence length and batch size. Existing cache eviction methods address this by exploiting attention sparsity, yet they typically rank tokens heuristically using accumulated attention weights without considering their true impact on attention outputs. We propose Optimal Brain Cache (OBCache), a principled framework that formulates cache eviction as a layer-wise structured pruning problem. Building upon the Optimal Brain Damage (OBD) theory, OBCache quantifies token saliency by measuring the perturbation in attention outputs induced by pruning tokens, with closed-form scores derived for isolated keys, isolated values, and joint key-value pairs. Our scores account not only for attention weights but also for information from value states and attention outputs, thereby enhancing existing eviction strategies with output-aware signals. Experiments on LLaMA and Qwen models demonstrate that replacing the heuristic scores in existing works, which estimate token saliency across different query positions, with OBCache's output-aware scores consistently improves long-context accuracy. Code is available at https://github.com/DreamSoul-AI/OBCache.

---

### Value Flows

Authors: Perry Dong, Chongyi Zheng, Chelsea Finn, Dorsa Sadigh, Benjamin Eysenbach
Published: 2025-10-09
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2510.07650v4

While most reinforcement learning methods today flatten the distribution of future returns to a single scalar value, distributional RL methods exploit the return distribution to provide stronger learning signals and to enable applications in exploration and safe RL. While the predominant method for estimating the return distribution is by modeling it as a categorical distribution over discrete bins or estimating a finite number of quantiles, such approaches leave unanswered questions about the fine-grained structure of the return distribution and about how to distinguish states with high return uncertainty for decision-making. The key idea in this paper is to use modern, flexible flow-based models to estimate the full future return distributions and identify those states with high return variance. We do so by formulating a new flow-matching objective that generates probability density paths satisfying the distributional Bellman equation. Building upon the learned flow models, we estimate the return uncertainty of distinct states using a new flow derivative ODE. We additionally use this uncertainty information to prioritize learning a more accurate return estimation on certain transitions. We compare our method (Value Flows) with prior methods in the offline and online-to-online settings. Experiments on $37$ state-based and $25$ image-based benchmark tasks demonstrate that Value Flows achieves a $1.3\times$ improvement on average in success rates. Website: https://pd-perry.github.io/value-flows Code: https://github.com/chongyi-zheng/value-flows

---

### A Honest Cross-Validation Estimator for Prediction Performance

Authors: Tianyu Pan, Vincent Z. Yu, Viswanath Devanarayan, Lu Tian
Published: 2025-10-09
Categories: stat.ML, cs.LG, stat.AP, stat.ME
arXiv: https://arxiv.org/abs/2510.07649v1

Cross-validation is a standard tool for obtaining a honest assessment of the performance of a prediction model. The commonly used version repeatedly splits data, trains the prediction model on the training set, evaluates the model performance on the test set, and averages the model performance across different data splits. A well-known criticism is that such cross-validation procedure does not directly estimate the performance of the particular model recommended for future use. In this paper, we propose a new method to estimate the performance of a model trained on a specific (random) training set. A naive estimator can be obtained by applying the model to a disjoint testing set. Surprisingly, cross-validation estimators computed from other random splits can be used to improve this naive estimator within a random-effects model framework. We develop two estimators -- a hierarchical Bayesian estimator and an empirical Bayes estimator -- that perform similarly to or better than both the conventional cross-validation estimator and the naive single-split estimator. Simulations and a real-data example demonstrate the superior performance of the proposed method.

---

### Failure Modes of Always-On Inter-Cluster Repulsion in Replay-Based Continual Learning

Authors: Md Hasibul Amin, Tamzid Tanvi Alam
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07648v3

Feature-space objectives are often added to replay-based continual learning systems with the expectation that better geometric separation will improve retention. We study a preliminary form of Cluster-Aware Replay (CAR) that combines a class-balanced replay memory with an always-active inter-cluster repulsion term (ICF). On five-task Split CIFAR-10 with a ResNet-18 backbone, the highest observed mean in a six-value sensitivity sweep reaches $22.5\pm1.4\%$ final average accuracy over three seeds, compared with $23.1\pm2.5\%$ for replay alone. ICF without replay reaches only $19.2\pm0.1\%$. All tested repulsion weights produce final accuracies between $20.1\%$ and $22.5\%$, and the detailed configuration exhibits $89.2\pm1.5$ percentage points of average forgetting. An instrumented rerun shows that the weighted repulsion contribution remains near $-0.13$ after cross-entropy has fallen close to zero, so the total objective becomes negative while old-task accuracy collapses. Importantly, the normalized distance objective is mathematically bounded; the failure is therefore better described as non-saturating, always-on repulsion rather than an unbounded loss. These negative results show that geometric separation is not automatically complementary to replay and motivate margin-gated objectives whose gradients deactivate once sufficient separation has been reached.

---

### Design-Based Bandits Under Network Interference: Trade-Off Between Regret and Statistical Inference

Authors: Zichen Wang, Haoyang Hong, Chuanhao Li, Haoxuan Li, Zhiheng Zhang, Huazheng Wang
Published: 2025-10-09
Categories: cs.LG
arXiv: https://arxiv.org/abs/2510.07646v2

In multi-armed bandits with network interference (MABNI), the action taken by one node can influence the rewards of others, creating complex interdependence. While existing research on MABNI largely concentrates on minimizing regret, it often overlooks the crucial concern that an excessive emphasis on the optimal arm can undermine the inference accuracy for sub-optimal arms. Although initial efforts have been made to address this trade-off in single-unit scenarios, these challenges have become more pronounced in the context of MABNI. In this paper, we establish, for the first time, a theoretical Pareto frontier characterizing the trade-off between regret minimization and inference accuracy in adversarial (design-based) MABNI. We further introduce an anytime-valid asymptotic confidence sequence along with a corresponding algorithm, $\texttt{EXP3-N-CS}$, specifically designed to balance the trade-off between regret minimization and inference accuracy in this setting.

---

### Banking Done Right: Redefining Retail Banking with Language-Centric AI

Authors: Xin Jie Chua, Jeraelyn Ming Li Tan, Jia Xuan Tan, Soon Chang Poh, Yi Xian Goh, Debbie Hui Tian Choong, Chee Mun Foong, Sze Jue Yang, Chee Seng Chan
Published: 2025-10-09
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2510.07645v1

This paper presents Ryt AI, an LLM-native agentic framework that powers Ryt Bank to enable customers to execute core financial transactions through natural language conversation. This represents the first global regulator-approved deployment worldwide where conversational AI functions as the primary banking interface, in contrast to prior assistants that have been limited to advisory or support roles. Built entirely in-house, Ryt AI is powered by ILMU, a closed-source LLM developed internally, and replaces rigid multi-screen workflows with a single dialogue orchestrated by four LLM-powered agents (Guardrails, Intent, Payment, and FAQ). Each agent attaches a task-specific LoRA adapter to ILMU, which is hosted within the bank's infrastructure to ensure consistent behavior with minimal overhead. Deterministic guardrails, human-in-the-loop confirmation, and a stateless audit architecture provide defense-in-depth for security and compliance. The result is Banking Done Right: demonstrating that regulator-approved natural-language interfaces can reliably support core financial operations under strict governance.

---