Showing 1–50 of 433 results for author: Bai, J

dMLLM-TTS: Self-Verified and Efficient Test-Time Scaling for Diffusion Multi-Modal Large Language Models

Authors: Yi Xin, Siqi Luo, Qi Qin, Haoxing Chen, Kaiwen Zhu, Zhiwei Zhang, Yangfan He, Rongchao Zhang, Jinbin Bai, Shuo Cao, Bin Fu, Junjun He, Yihao Liu, Yuewen Cao, Xiaohong Liu

Abstract: Diffusion Multi-modal Large Language Models (dMLLMs) have recently emerged as a novel architecture unifying image generation and understanding. However, developing effective and efficient Test-Time Scaling (TTS) methods to unlock their full generative potential remains an underexplored challenge. To address this, we propose dMLLM-TTS, a novel framework operating on two complementary scaling axes:… ▽ More Diffusion Multi-modal Large Language Models (dMLLMs) have recently emerged as a novel architecture unifying image generation and understanding. However, developing effective and efficient Test-Time Scaling (TTS) methods to unlock their full generative potential remains an underexplored challenge. To address this, we propose dMLLM-TTS, a novel framework operating on two complementary scaling axes: (1) trajectory exploration scaling to enhance the diversity of generated hypotheses, and (2) iterative refinement scaling for stable generation. Conventional TTS approaches typically perform linear search across these two dimensions, incurring substantial computational costs of O(NT) and requiring an external verifier for best-of-N selection. To overcome these limitations, we propose two innovations. First, we design an efficient hierarchical search algorithm with O(N+T) complexity that adaptively expands and prunes sampling trajectories. Second, we introduce a self-verified feedback mechanism that leverages the dMLLMs' intrinsic image understanding capabilities to assess text-image alignment, eliminating the need for external verifier. Extensive experiments on the GenEval benchmark across three representative dMLLMs (e.g., Lumina-DiMOO, MMaDA, Muddit) show that our framework substantially improves generation quality while achieving up to 6x greater efficiency than linear search. Project page: https://github.com/Alpha-VLLM/Lumina-DiMOO. △ Less

Submitted 22 December, 2025; originally announced December 2025.

Comments: Project page: https://github.com/Alpha-VLLM/Lumina-DiMOO

IC-Effect: Precise and Efficient Video Effects Editing via In-Context Learning

Authors: Yuanhang Li, Yiren Song, Junzhe Bai, Xinran Liang, Hu Yang, Libiao Jin, Qi Mao

Abstract: We propose \textbf{IC-Effect}, an instruction-guided, DiT-based framework for few-shot video VFX editing that synthesizes complex effects (\eg flames, particles and cartoon characters) while strictly preserving spatial and temporal consistency. Video VFX editing is highly challenging because injected effects must blend seamlessly with the background, the background must remain entirely unchanged,… ▽ More We propose \textbf{IC-Effect}, an instruction-guided, DiT-based framework for few-shot video VFX editing that synthesizes complex effects (\eg flames, particles and cartoon characters) while strictly preserving spatial and temporal consistency. Video VFX editing is highly challenging because injected effects must blend seamlessly with the background, the background must remain entirely unchanged, and effect patterns must be learned efficiently from limited paired data. However, existing video editing models fail to satisfy these requirements. IC-Effect leverages the source video as clean contextual conditions, exploiting the contextual learning capability of DiT models to achieve precise background preservation and natural effect injection. A two-stage training strategy, consisting of general editing adaptation followed by effect-specific learning via Effect-LoRA, ensures strong instruction following and robust effect modeling. To further improve efficiency, we introduce spatiotemporal sparse tokenization, enabling high fidelity with substantially reduced computation. We also release a paired VFX editing dataset spanning $15$ high-quality visual styles. Extensive experiments show that IC-Effect delivers high-quality, controllable, and temporally consistent VFX editing, opening new possibilities for video creation. △ Less

Submitted 17 December, 2025; originally announced December 2025.

RecTok: Reconstruction Distillation along Rectified Flow

Authors: Qingyu Shi, Size Wu, Jinbin Bai, Kaidong Yu, Yujing Wang, Yunhai Tong, Xiangtai Li, Xuelong Li

Abstract: Visual tokenizers play a crucial role in diffusion models. The dimensionality of latent space governs both reconstruction fidelity and the semantic expressiveness of the latent feature. However, a fundamental trade-off is inherent between dimensionality and generation quality, constraining existing methods to low-dimensional latent spaces. Although recent works have leveraged vision foundation mod… ▽ More Visual tokenizers play a crucial role in diffusion models. The dimensionality of latent space governs both reconstruction fidelity and the semantic expressiveness of the latent feature. However, a fundamental trade-off is inherent between dimensionality and generation quality, constraining existing methods to low-dimensional latent spaces. Although recent works have leveraged vision foundation models to enrich the semantics of visual tokenizers and accelerate convergence, high-dimensional tokenizers still underperform their low-dimensional counterparts. In this work, we propose RecTok, which overcomes the limitations of high-dimensional visual tokenizers through two key innovations: flow semantic distillation and reconstruction--alignment distillation. Our key insight is to make the forward flow in flow matching semantically rich, which serves as the training space of diffusion transformers, rather than focusing on the latent space as in previous works. Specifically, our method distills the semantic information in VFMs into the forward flow trajectories in flow matching. And we further enhance the semantics by introducing a masked feature reconstruction loss. Our RecTok achieves superior image reconstruction, generation quality, and discriminative performance. It achieves state-of-the-art results on the gFID-50K under both with and without classifier-free guidance settings, while maintaining a semantically rich latent space structure. Furthermore, as the latent dimensionality increases, we observe consistent improvements. Code and model are available at https://shi-qingyu.github.io/rectok.github.io. △ Less

Submitted 17 December, 2025; v1 submitted 15 December, 2025; originally announced December 2025.

EditMGT: Unleashing Potentials of Masked Generative Transformers in Image Editing

Authors: Wei Chow, Linfeng Li, Lingdong Kong, Zefeng Li, Qi Xu, Hang Song, Tian Ye, Xian Wang, Jinbin Bai, Shilin Xu, Xiangtai Li, Junting Pan, Shaoteng Liu, Ran Zhou, Tianshu Yang, Songhua Liu

Abstract: Recent advances in diffusion models (DMs) have achieved exceptional visual quality in image editing tasks. However, the global denoising dynamics of DMs inherently conflate local editing targets with the full-image context, leading to unintended modifications in non-target regions. In this paper, we shift our attention beyond DMs and turn to Masked Generative Transformers (MGTs) as an alternative… ▽ More Recent advances in diffusion models (DMs) have achieved exceptional visual quality in image editing tasks. However, the global denoising dynamics of DMs inherently conflate local editing targets with the full-image context, leading to unintended modifications in non-target regions. In this paper, we shift our attention beyond DMs and turn to Masked Generative Transformers (MGTs) as an alternative approach to tackle this challenge. By predicting multiple masked tokens rather than holistic refinement, MGTs exhibit a localized decoding paradigm that endows them with the inherent capacity to explicitly preserve non-relevant regions during the editing process. Building upon this insight, we introduce the first MGT-based image editing framework, termed EditMGT. We first demonstrate that MGT's cross-attention maps provide informative localization signals for localizing edit-relevant regions and devise a multi-layer attention consolidation scheme that refines these maps to achieve fine-grained and precise localization. On top of these adaptive localization results, we introduce region-hold sampling, which restricts token flipping within low-attention areas to suppress spurious edits, thereby confining modifications to the intended target regions and preserving the integrity of surrounding non-target areas. To train EditMGT, we construct CrispEdit-2M, a high-resolution dataset spanning seven diverse editing categories. Without introducing additional parameters, we adapt a pre-trained text-to-image MGT into an image editing model through attention injection. Extensive experiments across four standard benchmarks demonstrate that, with fewer than 1B parameters, our model achieves similarity performance while enabling 6 times faster editing. Moreover, it delivers comparable or superior editing quality, with improvements of 3.6% and 17.6% on style change and style transfer tasks, respectively. △ Less

Submitted 12 December, 2025; originally announced December 2025.

Openpi Comet: Competition Solution For 2025 BEHAVIOR Challenge

Authors: Junjie Bai, Yu-Wei Chao, Qizhi Chen, Jinwei Gu, Moo Jin Kim, Zhaoshuo Li, Xuan Li, Tsung-Yi Lin, Ming-Yu Liu, Nic Ma, Kaichun Mo, Delin Qu, Shangkun Sun, Hongchi Xia, Fangyin Wei, Xiaohui Zeng

Abstract: The 2025 BEHAVIOR Challenge is designed to rigorously track progress toward solving long-horizon tasks by physical agents in simulated environments. BEHAVIOR-1K focuses on everyday household tasks that people most want robots to assist with and these tasks introduce long-horizon mobile manipulation challenges in realistic settings, bridging the gap between current research and real-world, human-ce… ▽ More The 2025 BEHAVIOR Challenge is designed to rigorously track progress toward solving long-horizon tasks by physical agents in simulated environments. BEHAVIOR-1K focuses on everyday household tasks that people most want robots to assist with and these tasks introduce long-horizon mobile manipulation challenges in realistic settings, bridging the gap between current research and real-world, human-centric applications. This report presents our solution to the 2025 BEHAVIOR Challenge in a very close 2nd place and substantially outperforms the rest of the submissions. Building on $π_{0.5}$, we focus on systematically building our solution by studying the effects of training techniques and data. Through careful ablations, we show the scaling power in pre-training and post-training phases for competitive performance. We summarize our practical lessons and design recommendations that we hope will provide actionable insights for the broader embodied AI community when adapting powerful foundation models to complex embodied scenarios. △ Less

Submitted 12 December, 2025; v1 submitted 10 December, 2025; originally announced December 2025.

Comments: preprint

Native Parallel Reasoner: Reasoning in Parallelism via Self-Distilled Reinforcement Learning

Authors: Tong Wu, Yang Liu, Jun Bai, Zixia Jia, Shuyi Zhang, Ziyong Lin, Yanting Wang, Song-Chun Zhu, Zilong Zheng

Abstract: We introduce Native Parallel Reasoner (NPR), a teacher-free framework that enables Large Language Models (LLMs) to self-evolve genuine parallel reasoning capabilities. NPR transforms the model from sequential emulation to native parallel cognition through three key innovations: 1) a self-distilled progressive training paradigm that transitions from ``cold-start'' format discovery to strict topolog… ▽ More We introduce Native Parallel Reasoner (NPR), a teacher-free framework that enables Large Language Models (LLMs) to self-evolve genuine parallel reasoning capabilities. NPR transforms the model from sequential emulation to native parallel cognition through three key innovations: 1) a self-distilled progressive training paradigm that transitions from ``cold-start'' format discovery to strict topological constraints without external supervision; 2) a novel Parallel-Aware Policy Optimization (PAPO) algorithm that optimizes branching policies directly within the execution graph, allowing the model to learn adaptive decomposition via trial and error; and 3) a robust NPR Engine that refactors memory management and flow control of SGLang to enable stable, large-scale parallel RL training. Across eight reasoning benchmarks, NPR trained on Qwen3-4B achieves performance gains of up to 24.5% and inference speedups up to 4.6x. Unlike prior baselines that often fall back to autoregressive decoding, NPR demonstrates 100% genuine parallel execution, establishing a new standard for self-evolving, efficient, and scalable agentic reasoning. △ Less

Submitted 18 December, 2025; v1 submitted 8 December, 2025; originally announced December 2025.

Data-regularized Reinforcement Learning for Diffusion Models at Scale

Authors: Haotian Ye, Kaiwen Zheng, Jiashu Xu, Puheng Li, Huayu Chen, Jiaqi Han, Sheng Liu, Qinsheng Zhang, Hanzi Mao, Zekun Hao, Prithvijit Chattopadhyay, Dinghao Yang, Liang Feng, Maosheng Liao, Junjie Bai, Ming-Yu Liu, James Zou, Stefano Ermon

Abstract: Aligning generative diffusion models with human preferences via reinforcement learning (RL) is critical yet challenging. Most existing algorithms are often vulnerable to reward hacking, such as quality degradation, over-stylization, or reduced diversity. Our analysis demonstrates that this can be attributed to the inherent limitations of their regularization, which provides unreliable penalties. W… ▽ More Aligning generative diffusion models with human preferences via reinforcement learning (RL) is critical yet challenging. Most existing algorithms are often vulnerable to reward hacking, such as quality degradation, over-stylization, or reduced diversity. Our analysis demonstrates that this can be attributed to the inherent limitations of their regularization, which provides unreliable penalties. We introduce Data-regularized Diffusion Reinforcement Learning (DDRL), a novel framework that uses the forward KL divergence to anchor the policy to an off-policy data distribution. Theoretically, DDRL enables robust, unbiased integration of RL with standard diffusion training. Empirically, this translates into a simple yet effective algorithm that combines reward maximization with diffusion loss minimization. With over a million GPU hours of experiments and ten thousand double-blind human evaluations, we demonstrate on high-resolution video generation tasks that DDRL significantly improves rewards while alleviating the reward hacking seen in baselines, achieving the highest human preference and establishing a robust and scalable paradigm for diffusion post-training. △ Less

Submitted 6 December, 2025; v1 submitted 3 December, 2025; originally announced December 2025.

A transfer learning approach for automatic conflicts detection in software requirement sentence pairs based on dual encoders

Authors: Yizheng Wang, Tao Jiang, Jinyan Bai, Zhengbin Zou, Tiancheng Xue, Nan Zhang, Jie Luan

Abstract: Software Requirement Document (RD) typically contain tens of thousands of individual requirements, and ensuring consistency among these requirements is critical for the success of software engineering projects. Automated detection methods can significantly enhance efficiency and reduce costs; however, existing approaches still face several challenges, including low detection accuracy on imbalanced… ▽ More Software Requirement Document (RD) typically contain tens of thousands of individual requirements, and ensuring consistency among these requirements is critical for the success of software engineering projects. Automated detection methods can significantly enhance efficiency and reduce costs; however, existing approaches still face several challenges, including low detection accuracy on imbalanced data, limited semantic extraction due to the use of a single encoder, and suboptimal performance in cross-domain transfer learning. To address these issues, this paper proposes a Transferable Software Requirement Conflict Detection Framework based on SBERT and SimCSE, termed TSRCDF-SS. First, the framework employs two independent encoders, Sentence-BERT (SBERT) and Simple Contrastive Sentence Embedding (SimCSE), to generate sentence embeddings for requirement pairs, followed by a six-element concatenation strategy. Furthermore, the classifier is enhanced by a two-layer fully connected feedforward neural network (FFNN) with a hybrid loss optimization strategy that integrates a variant of Focal Loss, domain-specific constraints, and a confidence-based penalty term. Finally, the framework synergistically integrates sequential and cross-domain transfer learning. Experimental results demonstrate that the proposed framework achieves a 10.4% improvement in both macro-F1 and weighted-F1 scores in in-domain settings, and an 11.4% increase in macro-F1 in cross-domain scenarios. △ Less

Submitted 28 November, 2025; originally announced November 2025.

Comments: 22 pages, 7 figures, 3 tables

MSC Class: 68T01(Primary)68T50; 68N19(Secondary) ACM Class: I.2.7

"Power of Words": Stealthy and Adaptive Private Information Elicitation via LLM Communication Strategies

Authors: Shuning Zhang, Jiaqi Bai, Linzhi Wang, Shixuan Li, Xin Yi, Hewu Li

Abstract: While communication strategies of Large Language Models (LLMs) are crucial for human-LLM interactions, they can also be weaponized to elicit private information, yet such stealthy attacks remain under-explored. This paper introduces the first adaptive attack framework for stealthy and targeted private information elicitation via communication strategies. Our framework operates in a dynamic closed-… ▽ More While communication strategies of Large Language Models (LLMs) are crucial for human-LLM interactions, they can also be weaponized to elicit private information, yet such stealthy attacks remain under-explored. This paper introduces the first adaptive attack framework for stealthy and targeted private information elicitation via communication strategies. Our framework operates in a dynamic closed-loop: it first performs real-time psychological profiling of the users' state, then adaptively selects an optimized communication strategy, and finally maintains stealthiness through prompt-based rewriting. We validated this framework through a user study (N=84), demonstrating its generalizability across 3 distinct LLMs and 3 scenarios. The targeted attacks achieved a 205.4% increase in eliciting specific targeted information compared to stealthy interactions without strategies. Even stealthy interactions without specific strategies successfully elicited private information in 54.8% cases. Notably, users not only failed to detect the manipulation but paradoxically rated the attacking chatbot as more empathetic and trustworthy. Finally, we advocate for mitigations, encouraging developers to integrate adaptive, just-in-time alerts, users to build literacy against specific manipulative tactics, and regulators to define clear ethical boundaries distinguishing benign persuasion from coercion. △ Less

Submitted 14 November, 2025; originally announced November 2025.

Cross-Modal Unlearning via Influential Neuron Path Editing in Multimodal Large Language Models

Authors: Kunhao Li, Wenhao Li, Di Wu, Lei Yang, Jun Bai, Ju Jia, Jason Xue

Abstract: Multimodal Large Language Models (MLLMs) extend foundation models to real-world applications by integrating inputs such as text and vision. However, their broad knowledge capacity raises growing concerns about privacy leakage, toxicity mitigation, and intellectual property violations. Machine Unlearning (MU) offers a practical solution by selectively forgetting targeted knowledge while preserving… ▽ More Multimodal Large Language Models (MLLMs) extend foundation models to real-world applications by integrating inputs such as text and vision. However, their broad knowledge capacity raises growing concerns about privacy leakage, toxicity mitigation, and intellectual property violations. Machine Unlearning (MU) offers a practical solution by selectively forgetting targeted knowledge while preserving overall model utility. When applied to MLLMs, existing neuron-editing-based MU approaches face two fundamental challenges: (1) forgetting becomes inconsistent across modalities because existing point-wise attribution methods fail to capture the structured, layer-by-layer information flow that connects different modalities; and (2) general knowledge performance declines when sensitive neurons that also support important reasoning paths are pruned, as this disrupts the model's ability to generalize. To alleviate these limitations, we propose a multimodal influential neuron path editor (MIP-Editor) for MU. Our approach introduces modality-specific attribution scores to identify influential neuron paths responsible for encoding forget-set knowledge and applies influential-path-aware neuron-editing via representation misdirection. This strategy also enables effective and coordinated forgetting across modalities while preserving the model's general capabilities. Experimental results demonstrate that MIP-Editor achieves a superior unlearning performance on multimodal tasks, with a maximum forgetting rate of 87.75% and up to 54.26% improvement in general knowledge retention. On textual tasks, MIP-Editor achieves up to 80.65% forgetting and preserves 77.9% of general performance. Codes are available at https://github.com/PreckLi/MIP-Editor. △ Less

Submitted 10 November, 2025; originally announced November 2025.

Comments: Accepted at AAAI 2026 as a Conference Paper (Oral Presentation)

RelightMaster: Precise Video Relighting with Multi-plane Light Images

Authors: Weikang Bian, Xiaoyu Shi, Zhaoyang Huang, Jianhong Bai, Qinghe Wang, Xintao Wang, Pengfei Wan, Kun Gai, Hongsheng Li

Abstract: Recent advances in diffusion models enable high-quality video generation and editing, but precise relighting with consistent video contents, which is critical for shaping scene atmosphere and viewer attention, remains unexplored. Mainstream text-to-video (T2V) models lack fine-grained lighting control due to text's inherent limitation in describing lighting details and insufficient pre-training on… ▽ More Recent advances in diffusion models enable high-quality video generation and editing, but precise relighting with consistent video contents, which is critical for shaping scene atmosphere and viewer attention, remains unexplored. Mainstream text-to-video (T2V) models lack fine-grained lighting control due to text's inherent limitation in describing lighting details and insufficient pre-training on lighting-related prompts. Additionally, constructing high-quality relighting training data is challenging, as real-world controllable lighting data is scarce. To address these issues, we propose RelightMaster, a novel framework for accurate and controllable video relighting. First, we build RelightVideo, the first dataset with identical dynamic content under varying precise lighting conditions based on the Unreal Engine. Then, we introduce Multi-plane Light Image (MPLI), a novel visual prompt inspired by Multi-Plane Image (MPI). MPLI models lighting via K depth-aligned planes, representing 3D light source positions, intensities, and colors while supporting multi-source scenarios and generalizing to unseen light setups. Third, we design a Light Image Adapter that seamlessly injects MPLI into pre-trained Video Diffusion Transformers (DiT): it compresses MPLI via a pre-trained Video VAE and injects latent light features into DiT blocks, leveraging the base model's generative prior without catastrophic forgetting. Experiments show that RelightMaster generates physically plausible lighting and shadows and preserves original scene content. Demos are available at https://wkbian.github.io/Projects/RelightMaster/. △ Less

Submitted 9 November, 2025; originally announced November 2025.

Comments: Project Page: https://wkbian.github.io/Projects/RelightMaster/

Alpamayo-R1: Bridging Reasoning and Action Prediction for Generalizable Autonomous Driving in the Long Tail

Authors: NVIDIA, :, Yan Wang, Wenjie Luo, Junjie Bai, Yulong Cao, Tong Che, Ke Chen, Yuxiao Chen, Jenna Diamond, Yifan Ding, Wenhao Ding, Liang Feng, Greg Heinrich, Jack Huang, Peter Karkus, Boyi Li, Pinyi Li, Tsung-Yi Lin, Dongran Liu, Ming-Yu Liu, Langechuan Liu, Zhijian Liu, Jason Lu, Yunxiang Mao , et al. (19 additional authors not shown)

Abstract: End-to-end architectures trained via imitation learning have advanced autonomous driving by scaling model size and data, yet performance remains brittle in safety-critical long-tail scenarios where supervision is sparse and causal understanding is limited. To address this, we introduce Alpamayo-R1 (AR1), a vision-language-action model (VLA) that integrates Chain of Causation reasoning with traject… ▽ More End-to-end architectures trained via imitation learning have advanced autonomous driving by scaling model size and data, yet performance remains brittle in safety-critical long-tail scenarios where supervision is sparse and causal understanding is limited. To address this, we introduce Alpamayo-R1 (AR1), a vision-language-action model (VLA) that integrates Chain of Causation reasoning with trajectory planning to enhance decision-making in complex driving scenarios. Our approach features three key innovations: (1) the Chain of Causation (CoC) dataset, built through a hybrid auto-labeling and human-in-the-loop pipeline producing decision-grounded, causally linked reasoning traces aligned with driving behaviors; (2) a modular VLA architecture combining Cosmos-Reason, a Vision-Language Model pre-trained for Physical AI applications, with a diffusion-based trajectory decoder that generates dynamically feasible plans in real time; (3) a multi-stage training strategy using supervised fine-tuning to elicit reasoning and reinforcement learning (RL) to optimize reasoning quality via large reasoning model feedback and enforce reasoning-action consistency. Evaluation shows AR1 achieves up to a 12% improvement in planning accuracy on challenging cases compared to a trajectory-only baseline, with a 35% reduction in off-road rate and 25% reduction in close encounter rate in closed-loop simulation. RL post-training improves reasoning quality by 45% as measured by a large reasoning model critic and reasoning-action consistency by 37%. Model scaling from 0.5B to 7B parameters shows consistent improvements. On-vehicle road tests confirm real-time performance (99 ms latency) and successful urban deployment. By bridging interpretable reasoning with precise control, AR1 demonstrates a practical path towards Level 4 autonomous driving. We plan to release AR1 models and a subset of the CoC in a future update. △ Less

Submitted 29 October, 2025; originally announced November 2025.

World Simulation with Video Foundation Models for Physical AI

Authors: NVIDIA, :, Arslan Ali, Junjie Bai, Maciej Bala, Yogesh Balaji, Aaron Blakeman, Tiffany Cai, Jiaxin Cao, Tianshi Cao, Elizabeth Cha, Yu-Wei Chao, Prithvijit Chattopadhyay, Mike Chen, Yongxin Chen, Yu Chen, Shuai Cheng, Yin Cui, Jenna Diamond, Yifan Ding, Jiaojiao Fan, Linxi Fan, Liang Feng, Francesco Ferroni, Sanja Fidler , et al. (65 additional authors not shown)

Abstract: We introduce [Cosmos-Predict2.5], the latest generation of the Cosmos World Foundation Models for Physical AI. Built on a flow-based architecture, [Cosmos-Predict2.5] unifies Text2World, Image2World, and Video2World generation in a single model and leverages [Cosmos-Reason1], a Physical AI vision-language model, to provide richer text grounding and finer control of world simulation. Trained on 200… ▽ More We introduce [Cosmos-Predict2.5], the latest generation of the Cosmos World Foundation Models for Physical AI. Built on a flow-based architecture, [Cosmos-Predict2.5] unifies Text2World, Image2World, and Video2World generation in a single model and leverages [Cosmos-Reason1], a Physical AI vision-language model, to provide richer text grounding and finer control of world simulation. Trained on 200M curated video clips and refined with reinforcement learning-based post-training, [Cosmos-Predict2.5] achieves substantial improvements over [Cosmos-Predict1] in video quality and instruction alignment, with models released at 2B and 14B scales. These capabilities enable more reliable synthetic data generation, policy evaluation, and closed-loop simulation for robotics and autonomous systems. We further extend the family with [Cosmos-Transfer2.5], a control-net style framework for Sim2Real and Real2Real world translation. Despite being 3.5$\times$ smaller than [Cosmos-Transfer1], it delivers higher fidelity and robust long-horizon video generation. Together, these advances establish [Cosmos-Predict2.5] and [Cosmos-Transfer2.5] as versatile tools for scaling embodied intelligence. To accelerate research and deployment in Physical AI, we release source code, pretrained checkpoints, and curated benchmarks under the NVIDIA Open Model License at https://github.com/nvidia-cosmos/cosmos-predict2.5 and https://github.com/nvidia-cosmos/cosmos-transfer2.5. We hope these open resources lower the barrier to adoption and foster innovation in building the next generation of embodied intelligence. △ Less

Submitted 28 October, 2025; originally announced November 2025.

InteractComp: Evaluating Search Agents With Ambiguous Queries

Authors: Mingyi Deng, Lijun Huang, Yani Fan, Jiayi Zhang, Fashen Ren, Jinyi Bai, Fuzhen Yang, Dayi Miao, Zhaoyang Yu, Yifan Wu, Yanfei Zhang, Fengwei Teng, Yingjia Wan, Song Hu, Yude Li, Xin Jin, Conghao Hu, Haoyu Li, Qirui Fu, Tai Zhong, Xinyu Wang, Xiangru Tang, Nan Tang, Chenglin Wu, Yuyu Luo

Abstract: Language agents have demonstrated remarkable potential in web search and information retrieval. However, these search agents assume user queries are complete and unambiguous, an assumption that diverges from reality where users begin with incomplete queries requiring clarification through interaction. Yet most agents lack interactive mechanisms during the search process, and existing benchmarks ca… ▽ More Language agents have demonstrated remarkable potential in web search and information retrieval. However, these search agents assume user queries are complete and unambiguous, an assumption that diverges from reality where users begin with incomplete queries requiring clarification through interaction. Yet most agents lack interactive mechanisms during the search process, and existing benchmarks cannot assess this capability. To address this gap, we introduce InteractComp, a benchmark designed to evaluate whether search agents can recognize query ambiguity and actively interact to resolve it during search. Following the principle of easy to verify, interact to disambiguate, we construct 210 expert-curated questions across 9 domains through a target-distractor methodology that creates genuine ambiguity resolvable only through interaction. Evaluation of 17 models reveals striking failure: the best model achieves only 13.73% accuracy despite 71.50% with complete context, exposing systematic overconfidence rather than reasoning deficits. Forced interaction produces dramatic gains, demonstrating latent capability current strategies fail to engage. Longitudinal analysis shows interaction capabilities stagnated over 15 months while search performance improved seven-fold, revealing a critical blind spot. This stagnation, coupled with the immediate feedback inherent to search tasks, makes InteractComp a valuable resource for both evaluating and training interaction capabilities in search agents. The code is available at https://github.com/FoundationAgents/InteractComp. △ Less

Submitted 28 October, 2025; originally announced October 2025.

Language-Conditioned Representations and Mixture-of-Experts Policy for Robust Multi-Task Robotic Manipulation

Authors: Xiucheng Zhang, Yang Jiang, Hongwei Qing, Jiashuo Bai

Abstract: Perceptual ambiguity and task conflict limit multitask robotic manipulation via imitation learning. We propose a framework combining a Language-Conditioned Visual Representation (LCVR) module and a Language-conditioned Mixture-ofExperts Density Policy (LMoE-DP). LCVR resolves perceptual ambiguities by grounding visual features with language instructions, enabling differentiation between visually s… ▽ More Perceptual ambiguity and task conflict limit multitask robotic manipulation via imitation learning. We propose a framework combining a Language-Conditioned Visual Representation (LCVR) module and a Language-conditioned Mixture-ofExperts Density Policy (LMoE-DP). LCVR resolves perceptual ambiguities by grounding visual features with language instructions, enabling differentiation between visually similar tasks. To mitigate task conflict, LMoE-DP uses a sparse expert architecture to specialize in distinct, multimodal action distributions, stabilized by gradient modulation. On real-robot benchmarks, LCVR boosts Action Chunking with Transformers (ACT) and Diffusion Policy (DP) success rates by 33.75% and 25%, respectively. The full framework achieves a 79% average success, outperforming the advanced baseline by 21%. Our work shows that combining semantic grounding and expert specialization enables robust, efficient multi-task manipulation △ Less

Submitted 28 October, 2025; originally announced October 2025.

Comments: 8 pages

Your Dense Retriever is Secretly an Expeditious Reasoner

Authors: Yichi Zhang, Jun Bai, Zhixin Cai, Shuhan Qin, Zhuofan Chen, Jinghua Guan, Wenge Rong

Abstract: Dense retrievers enhance retrieval by encoding queries and documents into continuous vectors, but they often struggle with reasoning-intensive queries. Although Large Language Models (LLMs) can reformulate queries to capture complex reasoning, applying them universally incurs significant computational cost. In this work, we propose Adaptive Query Reasoning (AdaQR), a hybrid query rewriting framewo… ▽ More Dense retrievers enhance retrieval by encoding queries and documents into continuous vectors, but they often struggle with reasoning-intensive queries. Although Large Language Models (LLMs) can reformulate queries to capture complex reasoning, applying them universally incurs significant computational cost. In this work, we propose Adaptive Query Reasoning (AdaQR), a hybrid query rewriting framework. Within this framework, a Reasoner Router dynamically directs each query to either fast dense reasoning or deep LLM reasoning. The dense reasoning is achieved by the Dense Reasoner, which performs LLM-style reasoning directly in the embedding space, enabling a controllable trade-off between efficiency and accuracy. Experiments on large-scale retrieval benchmarks BRIGHT show that AdaQR reduces reasoning cost by 28% while preserving-or even improving-retrieval performance by 7%. △ Less

Submitted 27 October, 2025; v1 submitted 27 September, 2025; originally announced October 2025.

Comments: 16 pages, 11 figures

From Masks to Worlds: A Hitchhiker's Guide to World Models

Authors: Jinbin Bai, Yu Lei, Hecong Wu, Yuchen Zhu, Shufan Li, Yi Xin, Xiangtai Li, Molei Tao, Aditya Grover, Ming-Hsuan Yang

Abstract: This is not a typical survey of world models; it is a guide for those who want to build worlds. We do not aim to catalog every paper that has ever mentioned a ``world model". Instead, we follow one clear road: from early masked models that unified representation learning across modalities, to unified architectures that share a single paradigm, then to interactive generative models that close the a… ▽ More This is not a typical survey of world models; it is a guide for those who want to build worlds. We do not aim to catalog every paper that has ever mentioned a ``world model". Instead, we follow one clear road: from early masked models that unified representation learning across modalities, to unified architectures that share a single paradigm, then to interactive generative models that close the action-perception loop, and finally to memory-augmented systems that sustain consistent worlds over time. We bypass loosely related branches to focus on the core: the generative heart, the interactive loop, and the memory system. We show that this is the most promising path towards true world models. △ Less

Submitted 23 October, 2025; originally announced October 2025.

Comments: Github: https://github.com/M-E-AGI-Lab/Awesome-World-Models

AtlasKV: Augmenting LLMs with Billion-Scale Knowledge Graphs in 20GB VRAM

Authors: Haoyu Huang, Hong Ting Tsang, Jiaxin Bai, Xi Peng, Gong Zhang, Yangqiu Song

Abstract: Retrieval-augmented generation (RAG) has shown some success in augmenting large language models (LLMs) with external knowledge. However, as a non-parametric knowledge integration paradigm for LLMs, RAG methods heavily rely on external retrieval modules and the retrieved textual context prior. Especially for very large scale knowledge augmentation, they would introduce substantial inference latency… ▽ More Retrieval-augmented generation (RAG) has shown some success in augmenting large language models (LLMs) with external knowledge. However, as a non-parametric knowledge integration paradigm for LLMs, RAG methods heavily rely on external retrieval modules and the retrieved textual context prior. Especially for very large scale knowledge augmentation, they would introduce substantial inference latency due to expensive searches and much longer relevant context. In this paper, we propose a parametric knowledge integration method, called \textbf{AtlasKV}, a scalable, effective, and general way to augment LLMs with billion-scale knowledge graphs (KGs) (e.g. 1B triples) using very little GPU memory cost (e.g. less than 20GB VRAM). In AtlasKV, we introduce KG2KV and HiKVP to integrate KG triples into LLMs at scale with sub-linear time and memory complexity. It maintains strong knowledge grounding and generalization performance using the LLMs' inherent attention mechanism, and requires no external retrievers, long context priors, or retraining when adapting to new knowledge. △ Less

Submitted 20 October, 2025; originally announced October 2025.

Adaptive Discretization for Consistency Models

Authors: Jiayu Bai, Zhanbo Feng, Zhijie Deng, Tianqi Hou, Robert C. Qiu, Zenan Ling

Abstract: Consistency Models (CMs) have shown promise for efficient one-step generation. However, most existing CMs rely on manually designed discretization schemes, which can cause repeated adjustments for different noise schedules and datasets. To address this, we propose a unified framework for the automatic and adaptive discretization of CMs, formulating it as an optimization problem with respect to the… ▽ More Consistency Models (CMs) have shown promise for efficient one-step generation. However, most existing CMs rely on manually designed discretization schemes, which can cause repeated adjustments for different noise schedules and datasets. To address this, we propose a unified framework for the automatic and adaptive discretization of CMs, formulating it as an optimization problem with respect to the discretization step. Concretely, during the consistency training process, we propose using local consistency as the optimization objective to ensure trainability by avoiding excessive discretization, and taking global consistency as a constraint to ensure stability by controlling the denoising error in the training target. We establish the trade-off between local and global consistency with a Lagrange multiplier. Building on this framework, we achieve adaptive discretization for CMs using the Gauss-Newton method. We refer to our approach as ADCMs. Experiments demonstrate that ADCMs significantly improve the training efficiency of CMs, achieving superior generative performance with minimal training overhead on both CIFAR-10 and ImageNet. Moreover, ADCMs exhibit strong adaptability to more advanced DM variants. Code is available at https://github.com/rainstonee/ADCM. △ Less

Submitted 20 October, 2025; originally announced October 2025.

Comments: Accepted by NeurIPS 2025

JudgeSQL: Reasoning over SQL Candidates with Weighted Consensus Tournament

Authors: Jiayuan Bai, Xuan-guang Pan, Chongyang Tao, Shuai Ma

Abstract: Text-to-SQL is a pivotal task that bridges natural language understanding and structured data access, yet it remains fundamentally challenging due to semantic ambiguity and complex compositional reasoning. While large language models (LLMs) have greatly advanced SQL generation though prompting, supervised finetuning and reinforced tuning, the shift toward test-time scaling exposes a new bottleneck… ▽ More Text-to-SQL is a pivotal task that bridges natural language understanding and structured data access, yet it remains fundamentally challenging due to semantic ambiguity and complex compositional reasoning. While large language models (LLMs) have greatly advanced SQL generation though prompting, supervised finetuning and reinforced tuning, the shift toward test-time scaling exposes a new bottleneck: selecting the correct query from a diverse candidate pool. Existing selection approaches, such as self-consistency or best-of-$N$ decoding, provide only shallow signals, making them prone to inconsistent scoring, fragile reasoning chains, and a failure to capture fine-grained semantic distinctions between closely related SQL candidates. To this end, we introduce JudgeSQL, a principled framework that redefines SQL candidate selection through structured reasoning and weighted consensus tournament mechanism. JudgeSQL develops a reasoning-based SQL judge model that distills reasoning traces with reinforcement learning guided by verifiable rewards, enabling accurate and interpretable judgments. Building on this, a weighted consensus tournament integrates explicit reasoning preferences with implicit generator confidence, yielding selections that are both more reliable and more efficient. Extensive experiments on the BIRD benchmark demonstrate that JudgeSQL exhibits superior SQL judgment capabilities and good cross-scale generalization and robustness to generator capacity. △ Less

Submitted 17 October, 2025; originally announced October 2025.

Comments: 13 pages

AutoGraph-R1: End-to-End Reinforcement Learning for Knowledge Graph Construction

Authors: Hong Ting Tsang, Jiaxin Bai, Haoyu Huang, Qiao Xiao, Tianshi Zheng, Baixuan Xu, Shujie Liu, Yangqiu Song

Abstract: Building effective knowledge graphs (KGs) for Retrieval-Augmented Generation (RAG) is pivotal for advancing question answering (QA) systems. However, its effectiveness is hindered by a fundamental disconnect: the knowledge graph (KG) construction process is decoupled from its downstream application, yielding suboptimal graph structures. To bridge this gap, we introduce AutoGraph-R1, the first fram… ▽ More Building effective knowledge graphs (KGs) for Retrieval-Augmented Generation (RAG) is pivotal for advancing question answering (QA) systems. However, its effectiveness is hindered by a fundamental disconnect: the knowledge graph (KG) construction process is decoupled from its downstream application, yielding suboptimal graph structures. To bridge this gap, we introduce AutoGraph-R1, the first framework to directly optimize KG construction for task performance using Reinforcement Learning (RL). AutoGraph-R1 trains an LLM constructor by framing graph generation as a policy learning problem, where the reward is derived from the graph's functional utility in a RAG pipeline. We design two novel, task-aware reward functions, one for graphs as knowledge carriers and another as knowledge indices. Across multiple QA benchmarks, AutoGraph-R1 consistently enables graph RAG methods to achieve significant performance gains over using task-agnostic baseline graphs. Our work shows it is possible to close the loop between construction and application, shifting the paradigm from building intrinsically ``good'' graphs to building demonstrably ``useful'' ones. △ Less

Submitted 19 October, 2025; v1 submitted 17 October, 2025; originally announced October 2025.

Unifying Deductive and Abductive Reasoning in Knowledge Graphs with Masked Diffusion Model

Authors: Yisen Gao, Jiaxin Bai, Yi Huang, Xingcheng Fu, Qingyun Sun, Yangqiu Song

Abstract: Deductive and abductive reasoning are two critical paradigms for analyzing knowledge graphs, enabling applications from financial query answering to scientific discovery. Deductive reasoning on knowledge graphs usually involves retrieving entities that satisfy a complex logical query, while abductive reasoning generates plausible logical hypotheses from observations. Despite their clear synergisti… ▽ More Deductive and abductive reasoning are two critical paradigms for analyzing knowledge graphs, enabling applications from financial query answering to scientific discovery. Deductive reasoning on knowledge graphs usually involves retrieving entities that satisfy a complex logical query, while abductive reasoning generates plausible logical hypotheses from observations. Despite their clear synergistic potential, where deduction can validate hypotheses and abduction can uncover deeper logical patterns, existing methods address them in isolation. To bridge this gap, we propose DARK, a unified framework for Deductive and Abductive Reasoning in Knowledge graphs. As a masked diffusion model capable of capturing the bidirectional relationship between queries and conclusions, DARK has two key innovations. First, to better leverage deduction for hypothesis refinement during abductive reasoning, we introduce a self-reflective denoising process that iteratively generates and validates candidate hypotheses against the observed conclusion. Second, to discover richer logical associations, we propose a logic-exploration reinforcement learning approach that simultaneously masks queries and conclusions, enabling the model to explore novel reasoning compositions. Extensive experiments on multiple benchmark knowledge graphs show that DARK achieves state-of-the-art performance on both deductive and abductive reasoning tasks, demonstrating the significant benefits of our unified approach. △ Less

Submitted 13 October, 2025; originally announced October 2025.

Comments: Under Review

AdaViewPlanner: Adapting Video Diffusion Models for Viewpoint Planning in 4D Scenes

Authors: Yu Li, Menghan Xia, Gongye Liu, Jianhong Bai, Xintao Wang, Conglang Zhang, Yuxuan Lin, Ruihang Chu, Pengfei Wan, Yujiu Yang

Abstract: Recent Text-to-Video (T2V) models have demonstrated powerful capability in visual simulation of real-world geometry and physical laws, indicating its potential as implicit world models. Inspired by this, we explore the feasibility of leveraging the video generation prior for viewpoint planning from given 4D scenes, since videos internally accompany dynamic scenes with natural viewpoints. To this e… ▽ More Recent Text-to-Video (T2V) models have demonstrated powerful capability in visual simulation of real-world geometry and physical laws, indicating its potential as implicit world models. Inspired by this, we explore the feasibility of leveraging the video generation prior for viewpoint planning from given 4D scenes, since videos internally accompany dynamic scenes with natural viewpoints. To this end, we propose a two-stage paradigm to adapt pre-trained T2V models for viewpoint prediction, in a compatible manner. First, we inject the 4D scene representation into the pre-trained T2V model via an adaptive learning branch, where the 4D scene is viewpoint-agnostic and the conditional generated video embeds the viewpoints visually. Then, we formulate viewpoint extraction as a hybrid-condition guided camera extrinsic denoising process. Specifically, a camera extrinsic diffusion branch is further introduced onto the pre-trained T2V model, by taking the generated video and 4D scene as input. Experimental results show the superiority of our proposed method over existing competitors, and ablation studies validate the effectiveness of our key technical designs. To some extent, this work proves the potential of video generation models toward 4D interaction in real world. △ Less

Submitted 12 October, 2025; originally announced October 2025.

DixitWorld: Evaluating Multimodal Abductive Reasoning in Vision-Language Models with Multi-Agent Dixit Gameplay

Authors: Yunxiang Mo, Tianshi Zheng, Qing Zong, Jiayu Liu, Baixuan Xu, Yauwai Yim, Chunkit Chan, Jiaxin Bai, Yangqiu Song

Abstract: Multimodal abductive reasoning--the generation and selection of explanatory hypotheses from partial observations--is a cornerstone of intelligence. Current evaluations of this ability in vision-language models (VLMs) are largely confined to static, single-agent tasks. Inspired by Dixit, we introduce DixitWorld, a comprehensive evaluation suite designed to deconstruct this challenge. DIXITWORLD fea… ▽ More Multimodal abductive reasoning--the generation and selection of explanatory hypotheses from partial observations--is a cornerstone of intelligence. Current evaluations of this ability in vision-language models (VLMs) are largely confined to static, single-agent tasks. Inspired by Dixit, we introduce DixitWorld, a comprehensive evaluation suite designed to deconstruct this challenge. DIXITWORLD features two core components: DixitArena, a dynamic, multi-agent environment that evaluates both hypothesis generation (a "storyteller" crafting cryptic clues) and hypothesis selection ("listeners" choosing the target image from decoys) under imperfect information; and DixitBench, a static QA benchmark that isolates the listener's task for efficient, controlled evaluation. Results from DixitArena reveal distinct, role-dependent behaviors: smaller open-source models often excel as creative storytellers, producing imaginative yet less discriminative clues, whereas larger proprietary models demonstrate superior overall performance, particularly as listeners. Performance on DixitBench strongly correlates with listener results in DixitArena, validating it as a reliable proxy for hypothesis selection. Our findings reveal a key trade-off between generative creativity and discriminative understanding in multimodal abductive reasoning, a central challenge for developing more balanced and capable vision-language agents. △ Less

Submitted 11 October, 2025; originally announced October 2025.

Comments: EMNLP 2025 Wordplay (Spotlight)

Arbitrary Entropy Policy Optimization Breaks The Exploration Bottleneck of Reinforcement Learning

Authors: Chen Wang, Zhaochun Li, Jionghao Bai, Yuzhi Zhang, Shisheng Cui, Zhou Zhao, Yue Wang

Abstract: Reinforcement Learning (RL) is essential for enhancing the reasoning capabilities of large language models (LLMs), yet the widely adopted Group Relative Policy Optimization (GRPO) suffers from entropy collapse, causing exploration to vanish and policies to converge prematurely. As a result, RL is widely believed to be incapable of expanding the reasoning frontier of LLMs. Existing entropy-regulari… ▽ More Reinforcement Learning (RL) is essential for enhancing the reasoning capabilities of large language models (LLMs), yet the widely adopted Group Relative Policy Optimization (GRPO) suffers from entropy collapse, causing exploration to vanish and policies to converge prematurely. As a result, RL is widely believed to be incapable of expanding the reasoning frontier of LLMs. Existing entropy-regularized methods introduce an inevitable trade-off between reward and entropy, leading to exploration accompanied by non-negligible optimization bias. In this work, we prove that temperature-guided REINFORCE can modulate policy entropy, and propose Arbitrary Entropy Policy Optimization (AEPO), which reformulates entropy regularization as a policy-gradient optimization problem. Rather than manipulating entropy directly, AEPO implicitly regulates it by applying a REINFORCE regularization term on temperature-adjusted samples, ensuring that entropy is controlled but never dominates optimization, thereby enabling arbitrary and principled entropy regulation. Experiments show that AEPO outperforms RL baselines on both pass@1 and pass@$k$, and even surpasses the base model on pass@1024. By modulating entropy precisely, AEPO achieves more effective optimization dynamics and provides direct empirical evidence that entropy, exploration, and performance are intrinsically linked. △ Less

Submitted 16 December, 2025; v1 submitted 9 October, 2025; originally announced October 2025.

NewtonBench: Benchmarking Generalizable Scientific Law Discovery in LLM Agents

Authors: Tianshi Zheng, Kelvin Kiu-Wai Tam, Newt Hue-Nam K. Nguyen, Baixuan Xu, Zhaowei Wang, Jiayang Cheng, Hong Ting Tsang, Weiqi Wang, Jiaxin Bai, Tianqing Fang, Yangqiu Song, Ginny Y. Wong, Simon See

Abstract: Large language models are emerging as powerful tools for scientific law discovery, a foundational challenge in AI-driven science. However, existing benchmarks for this task suffer from a fundamental methodological trilemma, forcing a trade-off between scientific relevance, scalability, and resistance to memorization. Furthermore, they oversimplify discovery as static function fitting, failing to c… ▽ More Large language models are emerging as powerful tools for scientific law discovery, a foundational challenge in AI-driven science. However, existing benchmarks for this task suffer from a fundamental methodological trilemma, forcing a trade-off between scientific relevance, scalability, and resistance to memorization. Furthermore, they oversimplify discovery as static function fitting, failing to capture the authentic scientific process of uncovering embedded laws through the interactive exploration of complex model systems. To address these critical gaps, we introduce NewtonBench, a benchmark comprising 324 scientific law discovery tasks across 12 physics domains. Our design mitigates the evaluation trilemma by using counterfactual law shifts - systematic alterations of canonical laws - to generate a vast suite of problems that are scalable, scientifically relevant, and memorization-resistant. Moreover, we elevate the evaluation from static function fitting to interactive model discovery, requiring agents to experimentally probe simulated complex systems to uncover hidden principles. Our extensive experiment reveals a clear but fragile capability for discovery in frontier LLMs: this ability degrades precipitously with increasing system complexity and exhibits extreme sensitivity to observational noise. Notably, we uncover a paradoxical effect of tool assistance: providing a code interpreter can hinder more capable models by inducing a premature shift from exploration to exploitation, causing them to satisfice on suboptimal solutions. These results demonstrate that robust, generalizable discovery in complex, interactive environments remains the core challenge. By providing a scalable, robust, and scientifically authentic testbed, NewtonBench offers a crucial tool for measuring true progress and guiding the development of next-generation AI agents capable of genuine scientific discovery. △ Less

Submitted 9 December, 2025; v1 submitted 8 October, 2025; originally announced October 2025.

Comments: 71 pages, 21 figures, 21 tables

Federated Unlearning in the Wild: Rethinking Fairness and Data Discrepancy

Authors: ZiHeng Huang, Di Wu, Jun Bai, Jiale Zhang, Sicong Cao, Ji Zhang, Yingjie Hu

Abstract: Machine unlearning is critical for enforcing data deletion rights like the "right to be forgotten." As a decentralized paradigm, Federated Learning (FL) also requires unlearning, but realistic implementations face two major challenges. First, fairness in Federated Unlearning (FU) is often overlooked. Exact unlearning methods typically force all clients into costly retraining, even those uninvolved… ▽ More Machine unlearning is critical for enforcing data deletion rights like the "right to be forgotten." As a decentralized paradigm, Federated Learning (FL) also requires unlearning, but realistic implementations face two major challenges. First, fairness in Federated Unlearning (FU) is often overlooked. Exact unlearning methods typically force all clients into costly retraining, even those uninvolved. Approximate approaches, using gradient ascent or distillation, make coarse interventions that can unfairly degrade performance for clients with only retained data. Second, most FU evaluations rely on synthetic data assumptions (IID/non-IID) that ignore real-world heterogeneity. These unrealistic benchmarks obscure the true impact of unlearning and limit the applicability of current methods. We first conduct a comprehensive benchmark of existing FU methods under realistic data heterogeneity and fairness conditions. We then propose a novel, fairness-aware FU approach, Federated Cross-Client-Constrains Unlearning (FedCCCU), to explicitly address both challenges. FedCCCU offers a practical and scalable solution for real-world FU. Experimental results show that existing methods perform poorly in realistic settings, while our approach consistently outperforms them. △ Less

Submitted 8 October, 2025; originally announced October 2025.

Lumina-DiMOO: An Omni Diffusion Large Language Model for Multi-Modal Generation and Understanding

Authors: Yi Xin, Qi Qin, Siqi Luo, Kaiwen Zhu, Juncheng Yan, Yan Tai, Jiayi Lei, Yuewen Cao, Keqi Wang, Yibin Wang, Jinbin Bai, Qian Yu, Dengyang Jiang, Yuandong Pu, Haoxing Chen, Le Zhuo, Junjun He, Gen Luo, Tianbin Li, Ming Hu, Jin Ye, Shenglong Ye, Bo Zhang, Chang Xu, Wenhai Wang , et al. (7 additional authors not shown)

Abstract: We introduce Lumina-DiMOO, an open-source foundational model for seamless multi-modal generation and understanding. Lumina-DiMOO sets itself apart from prior unified models by utilizing a fully discrete diffusion modeling to handle inputs and outputs across various modalities. This innovative approach allows Lumina-DiMOO to achieve higher sampling efficiency compared to previous autoregressive (AR… ▽ More We introduce Lumina-DiMOO, an open-source foundational model for seamless multi-modal generation and understanding. Lumina-DiMOO sets itself apart from prior unified models by utilizing a fully discrete diffusion modeling to handle inputs and outputs across various modalities. This innovative approach allows Lumina-DiMOO to achieve higher sampling efficiency compared to previous autoregressive (AR) or hybrid AR-Diffusion paradigms and adeptly support a broad spectrum of multi-modal tasks, including text-to-image generation, image-to-image generation (e.g., image editing, subject-driven generation, and image inpainting, etc.), as well as image understanding. Lumina-DiMOO achieves state-of-the-art performance on multiple benchmarks, surpassing existing open-source unified multi-modal models. To foster further advancements in multi-modal and discrete diffusion model research, we release our code and checkpoints to the community. Project Page: https://synbol.github.io/Lumina-DiMOO. △ Less

Submitted 7 October, 2025; originally announced October 2025.

Comments: 33 pages, 13 figures, 10 tables

EEPO: Exploration-Enhanced Policy Optimization via Sample-Then-Forget

Authors: Liang Chen, Xueting Han, Qizhou Wang, Bo Han, Jing Bai, Hinrich Schutze, Kam-Fai Wong

Abstract: Balancing exploration and exploitation remains a central challenge in reinforcement learning with verifiable rewards (RLVR) for large language models (LLMs). Current RLVR methods often overemphasize exploitation, leading to entropy collapse, diminished exploratory capacity, and ultimately limited performance gains. Although techniques that increase policy stochasticity can promote exploration, the… ▽ More Balancing exploration and exploitation remains a central challenge in reinforcement learning with verifiable rewards (RLVR) for large language models (LLMs). Current RLVR methods often overemphasize exploitation, leading to entropy collapse, diminished exploratory capacity, and ultimately limited performance gains. Although techniques that increase policy stochasticity can promote exploration, they frequently fail to escape dominant behavioral modes. This creates a self-reinforcing loop-repeatedly sampling and rewarding dominant modes-that further erodes exploration. We introduce Exploration-Enhanced Policy Optimization (EEPO), a framework that promotes exploration via two-stage rollouts with adaptive unlearning. In the first stage, the model generates half of the trajectories; it then undergoes a lightweight unlearning step to temporarily suppress these sampled responses, forcing the second stage to explore different regions of the output space. This sample-then-forget mechanism disrupts the self-reinforcing loop and promotes wider exploration during rollouts. Across five reasoning benchmarks, EEPO outperforms GRPO, achieving average relative gains of 24.3% on Qwen2.5-3B, 33.0% on Llama3.2-3B-Instruct, and 10.4% on Qwen3-8B-Base. △ Less

Submitted 7 October, 2025; originally announced October 2025.

Correlating Cross-Iteration Noise for DP-SGD using Model Curvature

Authors: Xin Gu, Yingtai Xiao, Guanlin He, Jiamu Bai, Daniel Kifer, Kiwan Maeng

Abstract: Differentially private stochastic gradient descent (DP-SGD) offers the promise of training deep learning models while mitigating many privacy risks. However, there is currently a large accuracy gap between DP-SGD and normal SGD training. This has resulted in different lines of research investigating orthogonal ways of improving privacy-preserving training. One such line of work, known as DP-MF, co… ▽ More Differentially private stochastic gradient descent (DP-SGD) offers the promise of training deep learning models while mitigating many privacy risks. However, there is currently a large accuracy gap between DP-SGD and normal SGD training. This has resulted in different lines of research investigating orthogonal ways of improving privacy-preserving training. One such line of work, known as DP-MF, correlates the privacy noise across different iterations of stochastic gradient descent -- allowing later iterations to cancel out some of the noise added to earlier iterations. In this paper, we study how to improve this noise correlation. We propose a technique called NoiseCurve that uses model curvature, estimated from public unlabeled data, to improve the quality of this cross-iteration noise correlation. Our experiments on various datasets, models, and privacy parameters show that the noise correlations computed by NoiseCurve offer consistent and significant improvements in accuracy over the correlation scheme used by DP-MF. △ Less

Submitted 6 October, 2025; originally announced October 2025.

Truncated Kernel Stochastic Gradient Descent with General Losses and Spherical Radial Basis Functions

Authors: Jinhui Bai, Andreas Christmann, Lei Shi

Abstract: In this paper, we propose a novel kernel stochastic gradient descent (SGD) algorithm for large-scale supervised learning with general losses. Compared to traditional kernel SGD, our algorithm improves efficiency and scalability through an innovative regularization strategy. By leveraging the infinite series expansion of spherical radial basis functions, this strategy projects the stochastic gradie… ▽ More In this paper, we propose a novel kernel stochastic gradient descent (SGD) algorithm for large-scale supervised learning with general losses. Compared to traditional kernel SGD, our algorithm improves efficiency and scalability through an innovative regularization strategy. By leveraging the infinite series expansion of spherical radial basis functions, this strategy projects the stochastic gradient onto a finite-dimensional hypothesis space, which is adaptively scaled according to the bias-variance trade-off, thereby enhancing generalization performance. Based on a new estimation of the spectral structure of the kernel-induced covariance operator, we develop an analytical framework that unifies optimization and generalization analyses. We prove that both the last iterate and the suffix average converge at minimax-optimal rates, and we further establish optimal strong convergence in the reproducing kernel Hilbert space. Our framework accommodates a broad class of classical loss functions, including least-squares, Huber, and logistic losses. Moreover, the proposed algorithm significantly reduces computational complexity and achieves optimal storage complexity by incorporating coordinate-wise updates from linear SGD, thereby avoiding the costly pairwise operations typical of kernel SGD and enabling efficient processing of streaming data. Finally, extensive numerical experiments demonstrate the efficiency of our approach. △ Less

Submitted 30 October, 2025; v1 submitted 5 October, 2025; originally announced October 2025.

Comments: 54 pages, 20 figures

MSC Class: 68T05; 68Q32; 62L20

Towards Better Optimization For Listwise Preference in Diffusion Models

Authors: Jiamu Bai, Xin Yu, Meilong Xu, Weitao Lu, Xin Pan, Kiwan Maeng, Daniel Kifer, Jian Wang, Yu Wang

Abstract: Reinforcement learning from human feedback (RLHF) has proven effectiveness for aligning text-to-image (T2I) diffusion models with human preferences. Although Direct Preference Optimization (DPO) is widely adopted for its computational efficiency and avoidance of explicit reward modeling, its applications to diffusion models have primarily relied on pairwise preferences. The precise optimization of… ▽ More Reinforcement learning from human feedback (RLHF) has proven effectiveness for aligning text-to-image (T2I) diffusion models with human preferences. Although Direct Preference Optimization (DPO) is widely adopted for its computational efficiency and avoidance of explicit reward modeling, its applications to diffusion models have primarily relied on pairwise preferences. The precise optimization of listwise preferences remains largely unaddressed. In practice, human feedback on image preferences often contains implicit ranked information, which conveys more precise human preferences than pairwise comparisons. In this work, we propose Diffusion-LPO, a simple and effective framework for Listwise Preference Optimization in diffusion models with listwise data. Given a caption, we aggregate user feedback into a ranked list of images and derive a listwise extension of the DPO objective under the Plackett-Luce model. Diffusion-LPO enforces consistency across the entire ranking by encouraging each sample to be preferred over all of its lower-ranked alternatives. We empirically demonstrate the effectiveness of Diffusion-LPO across various tasks, including text-to-image generation, image editing, and personalized preference alignment. Diffusion-LPO consistently outperforms pairwise DPO baselines on visual quality and preference alignment. △ Less

Submitted 1 October, 2025; originally announced October 2025.

SAGE-Music: Low-Latency Symbolic Music Generation via Attribute-Specialized Key-Value Head Sharing

Authors: Jiaye Tan, Haonan Luo, Linfeng Song, Shuaiqi Chen, Yishan Lyu, Zian Zhong, Roujia Wang, Daniel Jiang, Haoran Zhang, Jiaming Bai, Haoran Cheng, Q. Vera Liao, Hao-Wen Dong

Abstract: Low-latency symbolic music generation is essential for real-time improvisation and human-AI co-creation. Existing transformer-based models, however, face a trade-off between inference speed and musical quality. Traditional acceleration techniques such as embedding pooling significantly degrade quality, while recently proposed Byte Pair Encoding (BPE) methods - though effective on single-track pian… ▽ More Low-latency symbolic music generation is essential for real-time improvisation and human-AI co-creation. Existing transformer-based models, however, face a trade-off between inference speed and musical quality. Traditional acceleration techniques such as embedding pooling significantly degrade quality, while recently proposed Byte Pair Encoding (BPE) methods - though effective on single-track piano data - suffer large performance drops in multi-track settings, as revealed by our analysis. We propose Attribute-Specialized Key-Value Head Sharing (AS-KVHS), adapted to music's structured symbolic representation, achieving about 30% inference speedup with only a negligible (about 0.4%) quality drop in objective evaluations and slight improvements in subjective listening tests. Our main contributions are (1) the first systematic study of BPE's generalizability in multi-track symbolic music, and (2) the introduction of AS-KVHS for low-latency symbolic music generation. Beyond these, we also release SAGE-Music, an open-source benchmark that matches or surpasses state-of-the-art models in generation quality. △ Less

Submitted 14 October, 2025; v1 submitted 30 September, 2025; originally announced October 2025.

Comments: Withdrawn after identifying that results in Section 5 require additional re-analysis before public dissemination

Pre-training CLIP against Data Poisoning with Optimal Transport-based Matching and Alignment

Authors: Tong Zhang, Kuofeng Gao, Jiawang Bai, Leo Yu Zhang, Xin Yin, Zonghui Wang, Shouling Ji, Wenzhi Chen

Abstract: Recent studies have shown that Contrastive Language-Image Pre-training (CLIP) models are threatened by targeted data poisoning and backdoor attacks due to massive training image-caption pairs crawled from the Internet. Previous defense methods correct poisoned image-caption pairs by matching a new caption for each image. However, the matching process relies solely on the global representations of… ▽ More Recent studies have shown that Contrastive Language-Image Pre-training (CLIP) models are threatened by targeted data poisoning and backdoor attacks due to massive training image-caption pairs crawled from the Internet. Previous defense methods correct poisoned image-caption pairs by matching a new caption for each image. However, the matching process relies solely on the global representations of images and captions, overlooking fine-grained features of visual and textual features. It may introduce incorrect image-caption pairs and harm the CLIP pre-training. To address their limitations, we propose an Optimal Transport-based framework to reconstruct image-caption pairs, named OTCCLIP. We propose a new optimal transport-based distance measure between fine-grained visual and textual feature sets and re-assign new captions based on the proposed optimal transport distance. Additionally, to further reduce the negative impact of mismatched pairs, we encourage the inter- and intra-modality fine-grained alignment by employing optimal transport-based objective functions. Our experiments demonstrate that OTCCLIP can successfully decrease the attack success rates of poisoning attacks. Also, compared to previous methods, OTCCLIP significantly improves CLIP's zero-shot and linear probing performance trained on poisoned datasets. △ Less

Submitted 23 September, 2025; originally announced September 2025.

TERAG: Token-Efficient Graph-Based Retrieval-Augmented Generation

Authors: Qiao Xiao, Hong Ting Tsang, Jiaxin Bai

Abstract: Graph-based Retrieval-augmented generation (RAG) has become a widely studied approach for improving the reasoning, accuracy, and factuality of Large Language Models (LLMs). However, many existing graph-based RAG systems overlook the high cost associated with LLM token usage during graph construction, hindering large-scale adoption. To address this, we propose TERAG, a simple yet effective framewor… ▽ More Graph-based Retrieval-augmented generation (RAG) has become a widely studied approach for improving the reasoning, accuracy, and factuality of Large Language Models (LLMs). However, many existing graph-based RAG systems overlook the high cost associated with LLM token usage during graph construction, hindering large-scale adoption. To address this, we propose TERAG, a simple yet effective framework designed to build informative graphs at a significantly lower cost. Inspired by HippoRAG, we incorporate Personalized PageRank (PPR) during the retrieval phase, and we achieve at least 80% of the accuracy of widely used graph-based RAG methods while consuming only 3%-11% of the output tokens. With its low token footprint and efficient construction pipeline, TERAG is well-suited for large-scale and cost-sensitive deployment scenarios. △ Less

Submitted 10 November, 2025; v1 submitted 23 September, 2025; originally announced September 2025.

Comments: 16 pages, 3 figures, 4 tables. Code available at https://github.com/wocqcm2/TERAG

CogAtom: From Cognitive Atoms to Olympiad-level Mathematical Reasoning in Large Language Models

Authors: Zhuofan Chen, Jiyuan He, Yichi Zhang, Xing Hu, Haoxing Wen, Jun Bai, Wenge Rong

Abstract: Mathematical reasoning poses significant challenges for Large Language Models (LLMs) due to its demand for multi-step reasoning and abstract conceptual integration. While recent test-time scaling techniques rely heavily on high-quality, challenging problems, the scarcity of Olympiad-level math problems remains a bottleneck. We introduce CogAtom, a novel cognitive atom-based framework for synthesiz… ▽ More Mathematical reasoning poses significant challenges for Large Language Models (LLMs) due to its demand for multi-step reasoning and abstract conceptual integration. While recent test-time scaling techniques rely heavily on high-quality, challenging problems, the scarcity of Olympiad-level math problems remains a bottleneck. We introduce CogAtom, a novel cognitive atom-based framework for synthesizing mathematically rigorous and cognitively diverse problems. Unlike prior approaches, CogAtom models problem construction as a process of selecting and recombining fundamental reasoning units, cognitive atoms, extracted from human-authored solutions. A diversity-promoting random walk algorithm enables exploration of the cognitive atom space, while a constraint-based recombination mechanism ensures logical soundness and structural validity. The combinatorial nature of the graph structure provides a near-infinite space of reasoning paths, and the walk algorithm systematically explores this space to achieve large-scale synthesis of high-quality problems; meanwhile, by controlling the number of cognitive atoms, we can precisely adjust problem difficulty, ensuring diversity, scalability, and controllability of the generated problems. Experimental results demonstrate that CogAtom outperforms existing methods in accuracy, reasoning depth, and diversity, generating problems that closely match the difficulty of AIME while exceeding it in structural variation. Our work offers a cognitively grounded pathway toward scalable, high-quality math problem generation.Our code is publicly available at https://github.com/Icarus-1111/CogAtom. △ Less

Submitted 24 September, 2025; v1 submitted 21 September, 2025; originally announced September 2025.

SFN-YOLO: Towards Free-Range Poultry Detection via Scale-aware Fusion Networks

Authors: Jie Chen, Yuhong Feng, Tao Dai, Mingzhe Liu, Hongtao Chen, Zhaoxi He, Jiancong Bai

Abstract: Detecting and localizing poultry is essential for advancing smart poultry farming. Despite the progress of detection-centric methods, challenges persist in free-range settings due to multiscale targets, obstructions, and complex or dynamic backgrounds. To tackle these challenges, we introduce an innovative poultry detection approach named SFN-YOLO that utilizes scale-aware fusion. This approach co… ▽ More Detecting and localizing poultry is essential for advancing smart poultry farming. Despite the progress of detection-centric methods, challenges persist in free-range settings due to multiscale targets, obstructions, and complex or dynamic backgrounds. To tackle these challenges, we introduce an innovative poultry detection approach named SFN-YOLO that utilizes scale-aware fusion. This approach combines detailed local features with broader global context to improve detection in intricate environments. Furthermore, we have developed a new expansive dataset (M-SCOPE) tailored for varied free-range conditions. Comprehensive experiments demonstrate our model achieves an mAP of 80.7% with just 7.2M parameters, which is 35.1% fewer than the benchmark, while retaining strong generalization capability across different domains. The efficient and real-time detection capabilities of SFN-YOLO support automated smart poultry farming. The code and dataset can be accessed at https://github.com/chenjessiee/SFN-YOLO. △ Less

Submitted 21 September, 2025; originally announced September 2025.

Comments: Submitted to ICASSP 2026

AgentCTG: Harnessing Multi-Agent Collaboration for Fine-Grained Precise Control in Text Generation

Authors: Xinxu Zhou, Jiaqi Bai, Zhenqi Sun, Fanxiang Zeng, Yue Liu

Abstract: Although significant progress has been made in many tasks within the field of Natural Language Processing (NLP), Controlled Text Generation (CTG) continues to face numerous challenges, particularly in achieving fine-grained conditional control over generation. Additionally, in real scenario and online applications, cost considerations, scalability, domain knowledge learning and more precise contro… ▽ More Although significant progress has been made in many tasks within the field of Natural Language Processing (NLP), Controlled Text Generation (CTG) continues to face numerous challenges, particularly in achieving fine-grained conditional control over generation. Additionally, in real scenario and online applications, cost considerations, scalability, domain knowledge learning and more precise control are required, presenting more challenge for CTG. This paper introduces a novel and scalable framework, AgentCTG, which aims to enhance precise and complex control over the text generation by simulating the control and regulation mechanisms in multi-agent workflows. We explore various collaboration methods among different agents and introduce an auto-prompt module to further enhance the generation effectiveness. AgentCTG achieves state-of-the-art results on multiple public datasets. To validate its effectiveness in practical applications, we propose a new challenging Character-Driven Rewriting task, which aims to convert the original text into new text that conform to specific character profiles and simultaneously preserve the domain knowledge. When applied to online navigation with role-playing, our approach significantly enhances the driving experience through improved content delivery. By optimizing the generation of contextually relevant text, we enable a more immersive interaction within online communities, fostering greater personalization and user engagement. △ Less

Submitted 17 September, 2025; originally announced September 2025.

Beyond Two-Stage Training: Cooperative SFT and RL for LLM Reasoning

Authors: Liang Chen, Xueting Han, Li Shen, Jing Bai, Kam-Fai Wong

Abstract: Reinforcement learning (RL) has proven effective in incentivizing the reasoning abilities of large language models (LLMs), but suffers from severe efficiency challenges due to its trial-and-error nature. While the common practice employs supervised fine-tuning (SFT) as a warm-up stage for RL, this decoupled two-stage approach suffers from catastrophic forgetting: second-stage RL gradually loses SF… ▽ More Reinforcement learning (RL) has proven effective in incentivizing the reasoning abilities of large language models (LLMs), but suffers from severe efficiency challenges due to its trial-and-error nature. While the common practice employs supervised fine-tuning (SFT) as a warm-up stage for RL, this decoupled two-stage approach suffers from catastrophic forgetting: second-stage RL gradually loses SFT-acquired behaviors and inefficiently explores new patterns. This study introduces a novel method for learning reasoning models that employs bilevel optimization to facilitate better cooperation between these training paradigms. By conditioning the SFT objective on the optimal RL policy, our approach enables SFT to meta-learn how to guide RL's optimization process. During training, the lower level performs RL updates while simultaneously receiving SFT supervision, and the upper level explicitly maximizes the cooperative gain-the performance advantage of joint SFT-RL training over RL alone. Empirical evaluations on five reasoning benchmarks demonstrate that our method consistently outperforms baselines and achieves a better balance between effectiveness and efficiency. △ Less

Submitted 16 October, 2025; v1 submitted 8 September, 2025; originally announced September 2025.

Loong: Synthesize Long Chain-of-Thoughts at Scale through Verifiers

Authors: Xingyue Huang, Rishabh, Gregor Franke, Ziyi Yang, Jiamu Bai, Weijie Bai, Jinhe Bi, Zifeng Ding, Yiqun Duan, Chengyu Fan, Wendong Fan, Xin Gao, Ruohao Guo, Yuan He, Zhuangzhuang He, Xianglong Hu, Neil Johnson, Bowen Li, Fangru Lin, Siyu Lin, Tong Liu, Yunpu Ma, Hao Shen, Hao Sun, Beibei Wang , et al. (21 additional authors not shown)

Abstract: Recent advances in Large Language Models (LLMs) have shown that their reasoning capabilities can be significantly improved through Reinforcement Learning with Verifiable Reward (RLVR), particularly in domains like mathematics and programming, where ground-truth correctness can be automatically evaluated. However, extending this success to other reasoning-intensive domains remains challenging due t… ▽ More Recent advances in Large Language Models (LLMs) have shown that their reasoning capabilities can be significantly improved through Reinforcement Learning with Verifiable Reward (RLVR), particularly in domains like mathematics and programming, where ground-truth correctness can be automatically evaluated. However, extending this success to other reasoning-intensive domains remains challenging due to the scarcity of high-quality, verifiable datasets and the high cost of human supervision. In this work, we introduce the Loong Project: an open-source framework for scalable synthetic data generation and verification across a diverse range of reasoning-intensive domains. The framework consists of two key components: (1) LoongBench, a curated seed dataset containing 8,729 human-vetted examples across 12 domains (e.g., Advanced Mathematics, Chemistry, Logic), each paired with executable code and rich metadata; and (2) LoongEnv, a modular synthetic data generation environment that supports multiple prompting strategies to produce new question-answer-code triples. Together, these components form an agent-environment loop that enables reinforcement learning, where an LLM-based agent is rewarded for generating Chain-of-Thought (CoT) solutions that align with code-executed answers. Empirically, we benchmark LoongBench on a broad suite of both open-source and proprietary LLMs to evaluate domain coverage and reveal performance bottlenecks. In addition, we conduct a comprehensive analysis of synthetic data generated by LoongEnv, examining correctness, difficulty, and diversity. Code and documentation are available at https://github.com/camel-ai/loong. △ Less

Submitted 3 September, 2025; originally announced September 2025.

Understanding and Leveraging the Expert Specialization of Context Faithfulness in Mixture-of-Experts LLMs

Authors: Jun Bai, Minghao Tong, Yang Liu, Zixia Jia, Zilong Zheng

Abstract: Context faithfulness is essential for reliable reasoning in context-dependent scenarios. However, large language models often struggle to ground their outputs in the provided context, resulting in irrelevant responses. Inspired by the emergent expert specialization observed in mixture-of-experts architectures, this work investigates whether certain experts exhibit specialization in context utiliza… ▽ More Context faithfulness is essential for reliable reasoning in context-dependent scenarios. However, large language models often struggle to ground their outputs in the provided context, resulting in irrelevant responses. Inspired by the emergent expert specialization observed in mixture-of-experts architectures, this work investigates whether certain experts exhibit specialization in context utilization, offering a potential pathway toward targeted optimization for improved context faithfulness. To explore this, we propose Router Lens, a method that accurately identifies context-faithful experts. Our analysis reveals that these experts progressively amplify attention to relevant contextual information, thereby enhancing context grounding. Building on this insight, we introduce Context-faithful Expert Fine-Tuning (CEFT), a lightweight optimization approach that selectively fine-tunes context-faithful experts. Experiments across a wide range of benchmarks and models demonstrate that CEFT matches or surpasses the performance of full fine-tuning while being significantly more efficient. △ Less

Submitted 12 November, 2025; v1 submitted 27 August, 2025; originally announced August 2025.

Comments: EMNLP 2025 Main

Dynamic Fusion Multimodal Network for SpeechWellness Detection

Authors: Wenqiang Sun, Han Yin, Jisheng Bai, Jianfeng Chen

Abstract: Suicide is one of the leading causes of death among adolescents. Previous suicide risk prediction studies have primarily focused on either textual or acoustic information in isolation, the integration of multimodal signals, such as speech and text, offers a more comprehensive understanding of an individual's mental state. Motivated by this, and in the context of the 1st SpeechWellness detection ch… ▽ More Suicide is one of the leading causes of death among adolescents. Previous suicide risk prediction studies have primarily focused on either textual or acoustic information in isolation, the integration of multimodal signals, such as speech and text, offers a more comprehensive understanding of an individual's mental state. Motivated by this, and in the context of the 1st SpeechWellness detection challenge, we explore a lightweight multi-branch multimodal system based on a dynamic fusion mechanism for speechwellness detection. To address the limitation of prior approaches that rely on time-domain waveforms for acoustic analysis, our system incorporates both time-domain and time-frequency (TF) domain acoustic features, as well as semantic representations. In addition, we introduce a dynamic fusion block to adaptively integrate information from different modalities. Specifically, it applies learnable weights to each modality during the fusion process, enabling the model to adjust the contribution of each modality. To enhance computational efficiency, we design a lightweight structure by simplifying the original baseline model. Experimental results demonstrate that the proposed system exhibits superior performance compared to the challenge baseline, achieving a 78% reduction in model parameters and a 5% improvement in accuracy. △ Less

Submitted 1 September, 2025; v1 submitted 25 August, 2025; originally announced August 2025.

Comments: 6 pages, 5figures

Report number: Paper #534

No Pixel Left Behind: A Detail-Preserving Architecture for Robust High-Resolution AI-Generated Image Detection

Authors: Lianrui Mu, Zou Xingze, Jianhong Bai, Jiaqi Hu, Wenjie Zheng, Jiangnan Ye, Jiedong Zhuang, Mudassar Ali, Jing Wang, Haoji Hu

Abstract: The rapid growth of high-resolution, meticulously crafted AI-generated images poses a significant challenge to existing detection methods, which are often trained and evaluated on low-resolution, automatically generated datasets that do not align with the complexities of high-resolution scenarios. A common practice is to resize or center-crop high-resolution images to fit standard network inputs.… ▽ More The rapid growth of high-resolution, meticulously crafted AI-generated images poses a significant challenge to existing detection methods, which are often trained and evaluated on low-resolution, automatically generated datasets that do not align with the complexities of high-resolution scenarios. A common practice is to resize or center-crop high-resolution images to fit standard network inputs. However, without full coverage of all pixels, such strategies risk either obscuring subtle, high-frequency artifacts or discarding information from uncovered regions, leading to input information loss. In this paper, we introduce the High-Resolution Detail-Aggregation Network (HiDA-Net), a novel framework that ensures no pixel is left behind. We use the Feature Aggregation Module (FAM), which fuses features from multiple full-resolution local tiles with a down-sampled global view of the image. These local features are aggregated and fused with global representations for final prediction, ensuring that native-resolution details are preserved and utilized for detection. To enhance robustness against challenges such as localized AI manipulations and compression, we introduce Token-wise Forgery Localization (TFL) module for fine-grained spatial sensitivity and JPEG Quality Factor Estimation (QFE) module to disentangle generative artifacts from compression noise explicitly. Furthermore, to facilitate future research, we introduce HiRes-50K, a new challenging benchmark consisting of 50,568 images with up to 64 megapixels. Extensive experiments show that HiDA-Net achieves state-of-the-art, increasing accuracy by over 13% on the challenging Chameleon dataset and 10% on our HiRes-50K. △ Less

Submitted 24 August, 2025; originally announced August 2025.

MDIQA: Unified Image Quality Assessment for Multi-dimensional Evaluation and Restoration

Authors: Shunyu Yao, Ming Liu, Zhilu Zhang, Zhaolin Wan, Zhilong Ji, Jinfeng Bai, Wangmeng Zuo

Abstract: Recent advancements in image quality assessment (IQA), driven by sophisticated deep neural network designs, have significantly improved the ability to approach human perceptions. However, most existing methods are obsessed with fitting the overall score, neglecting the fact that humans typically evaluate image quality from different dimensions before arriving at an overall quality assessment. To o… ▽ More Recent advancements in image quality assessment (IQA), driven by sophisticated deep neural network designs, have significantly improved the ability to approach human perceptions. However, most existing methods are obsessed with fitting the overall score, neglecting the fact that humans typically evaluate image quality from different dimensions before arriving at an overall quality assessment. To overcome this problem, we propose a multi-dimensional image quality assessment (MDIQA) framework. Specifically, we model image quality across various perceptual dimensions, including five technical and four aesthetic dimensions, to capture the multifaceted nature of human visual perception within distinct branches. Each branch of our MDIQA is initially trained under the guidance of a separate dimension, and the respective features are then amalgamated to generate the final IQA score. Additionally, when the MDIQA model is ready, we can deploy it for a flexible training of image restoration (IR) models, enabling the restoration results to better align with varying user preferences through the adjustment of perceptual dimension weights. Extensive experiments demonstrate that our MDIQA achieves superior performance and can be effectively and flexibly applied to image restoration tasks. The code is available: https://github.com/YaoShunyu19/MDIQA. △ Less

Submitted 22 August, 2025; originally announced August 2025.

Repeater Swarm-Assisted Cellular Systems: Interaction Stability and Performance Analysis

Authors: Jianan Bai, Anubhab Chowdhury, Anders Hansson, Erik G. Larsson

Abstract: We consider a cellular massive MIMO system where swarms of wireless repeaters are deployed to improve coverage. These repeaters are full-duplex relays with small form factors that receive and instantaneously retransmit signals. They can be deployed in a plug-and-play manner at low cost, while being transparent to the network--conceptually they are active channel scatterers with amplification capab… ▽ More We consider a cellular massive MIMO system where swarms of wireless repeaters are deployed to improve coverage. These repeaters are full-duplex relays with small form factors that receive and instantaneously retransmit signals. They can be deployed in a plug-and-play manner at low cost, while being transparent to the network--conceptually they are active channel scatterers with amplification capabilities. Two fundamental questions need to be addressed in repeater deployments: (I) How can we prevent destructive effects of positive feedback caused by inter-repeater interaction (i.e., each repeater receives and amplifies signals from others)? (ii) How much performance improvement can be achieved given that repeaters also inject noise and may introduce more interference? To answer these questions, we first derive a generalized Nyquist stability criterion for the repeater swarm system, and provide an easy-to-check stability condition. Then, we study the uplink performance and develop an efficient iterative algorithm that jointly optimizes the repeater gains, user transmit powers, and receive combining weights to maximize the weighted sum rate while ensuring system stability. Numerical results corroborate our theoretical findings and show that the repeaters can significantly improve the system performance, both in sub-6 GHz and millimeter-wave bands. The results also warrant careful deployment to fully realize the benefits of repeaters, for example, by ensuring a high probability of line-of-sight links between repeaters and the base station. △ Less

Submitted 19 August, 2025; originally announced August 2025.

Comments: 16 pages, 13 figures. Submitted to IEEE Transactions on Wireless Communications

ESDD 2026: Environmental Sound Deepfake Detection Challenge Evaluation Plan

Authors: Han Yin, Yang Xiao, Rohan Kumar Das, Jisheng Bai, Ting Dang

Abstract: Recent advances in audio generation systems have enabled the creation of highly realistic and immersive soundscapes, which are increasingly used in film and virtual reality. However, these audio generators also raise concerns about potential misuse, such as generating deceptive audio content for fake videos and spreading misleading information. Existing datasets for environmental sound deepfake de… ▽ More Recent advances in audio generation systems have enabled the creation of highly realistic and immersive soundscapes, which are increasingly used in film and virtual reality. However, these audio generators also raise concerns about potential misuse, such as generating deceptive audio content for fake videos and spreading misleading information. Existing datasets for environmental sound deepfake detection (ESDD) are limited in scale and audio types. To address this gap, we have proposed EnvSDD, the first large-scale curated dataset designed for ESDD, consisting of 45.25 hours of real and 316.7 hours of fake sound. Based on EnvSDD, we are launching the Environmental Sound Deepfake Detection Challenge. Specifically, we present two different tracks: ESDD in Unseen Generators and Black-Box Low-Resource ESDD, covering various challenges encountered in real-life scenarios. The challenge will be held in conjunction with the 2026 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2026). △ Less

Submitted 6 August, 2025; originally announced August 2025.

On the Fast Adaptation of Delayed Clients in Decentralized Federated Learning: A Centroid-Aligned Distillation Approach

Authors: Jiahui Bai, Hai Dong, A. K. Qin

Abstract: Decentralized Federated Learning (DFL) struggles with the slow adaptation of late-joining delayed clients and high communication costs in asynchronous environments. These limitations significantly hinder overall performance. To address this, we propose DFedCAD, a novel framework for rapid adaptation via Centroid-Aligned Distillation. DFedCAD first employs Weighted Cluster Pruning (WCP) to compress… ▽ More Decentralized Federated Learning (DFL) struggles with the slow adaptation of late-joining delayed clients and high communication costs in asynchronous environments. These limitations significantly hinder overall performance. To address this, we propose DFedCAD, a novel framework for rapid adaptation via Centroid-Aligned Distillation. DFedCAD first employs Weighted Cluster Pruning (WCP) to compress models into representative centroids, drastically reducing communication overhead. It then enables delayed clients to intelligently weigh and align with peer knowledge using a novel structural distance metric and a differentiable k-means distillation module, facilitating efficient end-to-end knowledge transfer. Extensive experiments on CIFAR-10, CIFAR-100, and Tiny-ImageNet show that DFedCAD consistently achieves state-of-the-art performance, attaining the highest accuracy across all evaluated settings while reducing communication overhead by over 86%. Our framework provides a scalable and practical solution for efficient decentralized learning in dynamic, real-world scenarios. △ Less

Submitted 4 August, 2025; originally announced August 2025.

Comments: This paper is currently under peer review

Personalized Safety Alignment for Text-to-Image Diffusion Models

Authors: Yu Lei, Jinbin Bai, Qingyu Shi, Aosong Feng, Kaidong Yu

Abstract: Text-to-image diffusion models have revolutionized visual content generation, but current safety mechanisms apply uniform standards that often fail to account for individual user preferences. These models overlook the diverse safety boundaries shaped by factors like age, mental health, and personal beliefs. To address this, we propose Personalized Safety Alignment (PSA), a framework that allows us… ▽ More Text-to-image diffusion models have revolutionized visual content generation, but current safety mechanisms apply uniform standards that often fail to account for individual user preferences. These models overlook the diverse safety boundaries shaped by factors like age, mental health, and personal beliefs. To address this, we propose Personalized Safety Alignment (PSA), a framework that allows user-specific control over safety behaviors in generative models. PSA integrates personalized user profiles into the diffusion process, adjusting the model's behavior to match individual safety preferences while preserving image quality. We introduce a new dataset, Sage, which captures user-specific safety preferences and incorporates these profiles through a cross-attention mechanism. Experiments show that PSA outperforms existing methods in harmful content suppression and aligns generated content better with user constraints, achieving higher Win Rate and Pass Rate scores. Our code, data, and models are publicly available at https://m-e-agi-lab.github.io/PSAlign/. △ Less

Submitted 7 August, 2025; v1 submitted 1 August, 2025; originally announced August 2025.

Comments: metadata-only revision; corrected a typo in the abstract. No changes to the PDF content

The Manipulative Power of Voice Characteristics: Investigating Deceptive Patterns in Mandarin Chinese Female Synthetic Speech

Authors: Shuning Zhang, Han Chen, Yabo Wang, Yiqun Xu, Jiaqi Bai, Yuanyuan Wu, Shixuan Li, Xin Yi, Chunhui Wang, Hewu Li

Abstract: Pervasive voice interaction enables deceptive patterns through subtle voice characteristics, yet empirical investigation into this manipulation lags behind, especially within major non-English language contexts. Addressing this gap, our study presents the first systematic investigation into voice characteristic-based dark patterns employing female synthetic voices in Mandarin Chinese. This focus i… ▽ More Pervasive voice interaction enables deceptive patterns through subtle voice characteristics, yet empirical investigation into this manipulation lags behind, especially within major non-English language contexts. Addressing this gap, our study presents the first systematic investigation into voice characteristic-based dark patterns employing female synthetic voices in Mandarin Chinese. This focus is crucial given the prevalence of female personas in commercial assistants and the prosodic significance in the Chinese language. Guided by the conceptual framework identifying key influencing factors, we systematically evaluate effectiveness variations by manipulating voice characteristics (five characteristics, three intensities) across different scenarios (shopping vs. question-answering) with different commercial aims. A preliminary study (N=24) validated the experimental materials and the main study (N=36) revealed significant behavioral manipulation (up to +2027.6%). Crucially, the analysis showed that effectiveness varied significantly with voice characteristics and scenario, mediated by user perception (of tone, intonation, timbre) and user demographics (individual preferences, though limited demographic impact). These interconnected findings offer evidence-based insights for ethical design. △ Less

Submitted 1 August, 2025; originally announced August 2025.

DNN-based Methods of Jointly Sensing Number and Directions of Targets via a Green Massive H2AD MIMO Receiver

Authors: Bin Deng, Jiatong Bai, Feilong Zhao, Zuming Xie, Maolin Li, Yan Wang, Feng Shu

Abstract: As a green MIMO structure, the heterogeneous hybrid analog-digital H2AD MIMO architecture has been shown to own a great potential to replace the massive or extremely large-scale fully-digital MIMO in the future wireless networks to address the three challenging problems faced by the latter: high energy consumption, high circuit cost, and high complexity. However, how to intelligently sense the num… ▽ More As a green MIMO structure, the heterogeneous hybrid analog-digital H2AD MIMO architecture has been shown to own a great potential to replace the massive or extremely large-scale fully-digital MIMO in the future wireless networks to address the three challenging problems faced by the latter: high energy consumption, high circuit cost, and high complexity. However, how to intelligently sense the number and direction of multi-emitters via such a structure is still an open hard problem. To address this, we propose a two-stage sensing framework that jointly estimates the number and direction values of multiple targets. Specifically, three target number sensing methods are designed: an improved eigen-domain clustering (EDC) framework, an enhanced deep neural network (DNN) based on five key statistical features, and an improved one-dimensional convolutional neural network (1D-CNN) utilizing full eigenvalues. Subsequently, a low-complexity and high-accuracy DOA estimation is achieved via the introduced online micro-clustering (OMC-DOA) method. Furthermore, we derive the Cramér-Rao lower bound (CRLB) for the H2AD under multiple-source conditions as a theoretical performance benchmark. Simulation results show that the developed three methods achieve 100\% number of targets sensing at moderate-to-high SNRs, while the improved 1D-CNN exhibits superior under extremely-low SNR conditions. The introduced OMC-DOA outperforms existing clustering and fusion-based DOA methods in multi-source environments. △ Less

Submitted 15 July, 2025; originally announced July 2025.