NeurIPS

SORTeD Rashomon Sets of Sparse Decision Trees: Anytime Enumeration

Elif Arslan, Jacobus G. M. van der Linden, Serge Hoogendoorn, Marco Rinaldi, Emir Demirović

机构 * Delft University of Technology（代尔夫特理工大学） 纠错

Comments 32 pages, 10 figures, to be published in the proceedings of The Thirty-Ninth Annual Conference on Neural Information Processing Systems

A Probabilistic Approach to Pose Synchronization for Multi-Reference Alignment with Applications to MIMO Wireless Communication Systems

Rob Romijnders, Gabriele Cesa, Christos Louizos, Kumar Pratik, Arash Behboodi

机构 * University of Amsterdam（阿姆斯特丹大学） ； QUvA-Lab（QUvA实验室） ； Qualcomm AI Research（高通人工智能研究） 纠错

Comments To appear in NeurIPS workshop: AI and ML for Next-Generation Wireless Communications (AI4NextG)

FAPEX: Fractional Amplitude-Phase Expressor for Robust Cross-Subject Seizure Prediction

Ruizhe Zheng, Lingyan Mao, Dingding Han, Tian Luo, Yi Wang, Jing Ding, Yuguo Yu

Comments 39th Conference on Neural Information Processing Systems (NeurIPS 2025) Spotlight Poster

Decoupled Entropy Minimization

Jing Ma, Hanlin Li, Xiang Xiang

机构 * HUST AI and Visual Learning Lab (HAIV Lab)（华中科技大学人工智能与视觉学习实验室） ； Huazhong University of Science and Technology (HUST)（华中科技大学） 纠错

Comments To appear at NeurIPS 2025 (main conference), San Diego, CA, USA. Codes available at https://github.com/HAIV-Lab/DEM/

Scalable Evaluation and Neural Models for Compositional Generalization

Giacomo Camposampiero, Pietro Barbiero, Michael Hersche, Roger Wattenhofer, Abbas Rahimi

机构 * IBM Research – Zurich（IBM瑞士研究院） ； ETH Zurich（苏黎世联邦理工学院） 纠错

Comments Accepted at the Thirty-ninth Annual Conference on Neural Information Processing Systems (NeurIPS), 2025

Proximal Regret and Proximal Correlated Equilibria: A New Tractable Solution Concept for Online Learning and Games

Yang Cai, Constantinos Daskalakis, Haipeng Luo, Chen-Yu Wei, Weiqiang Zheng

机构 * Yale University（耶鲁大学） ； MIT CSAIL（麻省理工学院CSAIL） ； University of Southern California（南加州大学） ； University of Virginia（弗吉尼亚大学） 纠错

Comments This paper presents proximal regret and proximal correlated equilibria results that do not appear in the NeurIPS version of arXiv:2403.08171

Shift Before You Learn: Enabling Low-Rank Representations in Reinforcement Learning

Bastien Dubail, Stefan Stojanovic, Alexandre Proutière

Comments 63 pages, 11 figures. Accepted to NeurIPS 2025 (Spotlight)

OrdShap: Feature Position Importance for Sequential Black-Box Models

Davin Hill, Brian L. Hill, Aria Masoomi, Vijay S. Nori, Robert E. Tillman, Jennifer Dy

机构 * Northeastern University（东北大学） ； Optum AI 纠错

Comments Advances in Neural Information Processing Systems, Volume 38 (NeurIPS 2025)

ThinkSound: Chain-of-Thought Reasoning in Multimodal Large Language Models for Audio Generation and Editing

Huadai Liu, Kaicheng Luo, Jialei Wang, Wen Wang, Qian Chen, Zhou Zhao, Wei Xue

机构 * Hong Kong University of Science and Technology (HKUST)（香港理工大学） ； Tongyi Fun Team, Alibaba Group（阿里云团队） ； Zhejiang University（浙江大学） 纠错

Comments Accepted by NeurIPS 2025 Main

Dense SAE Latents Are Features, Not Bugs

Xiaoqing Sun, Alessandro Stolfo, Joshua Engels, Ben Wu, Senthooran Rajamanoharan, Mrinmaya Sachan, Max Tegmark

机构 * MIT（麻省理工学院） ； ETH Zürich（苏黎世联邦理工学院） ； University of Sheffield（谢菲尔德大学） 纠错

Comments NeurIPS 2025 poster

Token Perturbation Guidance for Diffusion Models

Javad Rajabi, Soroush Mehraban, Seyedmorteza Sadat, Babak Taati

机构 * University of Toronto（多伦多大学） ； Vector Institute for Artificial Intelligence（人工智能向量研究所） ； KITE Research Institute（KITE研究机构） ； ETH Zürich（苏黎世联邦理工学院） 纠错

Comments Accepted at NeurIPS 2025. Project page: https://github.com/TaatiTeam/Token-Perturbation-Guidance

Inv-Entropy: A Fully Probabilistic Framework for Uncertainty Quantification in Language Models

Haoyi Song, Ruihan Ji, Naichen Shi, Fan Lai, Raed Al Kontar

机构 * University of Michigan（密歇根大学） ； University of Minnesota（明尼苏达大学） ； Northwestern University（西北大学） ； University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校） 纠错

Journal ref NeurIPS, 2025

MagCache: Fast Video Generation with Magnitude-Aware Cache

Zehong Ma, Longhui Wei, Feng Wang, Shiliang Zhang, Qi Tian

机构 * State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University（多媒体信息处理国家重点实验室，计算机学院，北京大学） ； Huawei Inc.（华为公司） 纠错

Comments Project Page: https://zehong-ma.github.io/MagCache Accepted by NeurIPS 2025

Journal ref In Proceedings of NeurIPS 2025

Struct2D: A Perception-Guided Framework for Spatial Reasoning in MLLMs

Fangrui Zhu, Hanhui Wang, Yiming Xie, Jing Gu, Tianye Ding, Jianwei Yang, Huaizu Jiang

机构 * Northeastern University（东北大学） ； Microsoft Research（微软研究院） ； University of Southern California（南加州大学） ； University of California, Santa Cruz（加州大学圣克鲁兹分校） 纠错

Comments NeurIPS 2025, code link: https://github.com/neu-vi/struct2d

A Theoretical Framework for Grokking: Interpolation followed by Riemannian Norm Minimisation

Etienne Boursier, Scott Pesme, Radu-Alexandru Dragomir

机构 * INRIA LMO, Université Paris-Saclay Orsay, France（INRIA LMO，巴黎-萨克雷大学奥赛分校，法国） ； INRIA Grenoble, France（INRIA 格勒诺布尔，法国） ； Télécom Paris Institut Polytechnique de Paris Palaiseau, France（巴黎电信研究院巴黎理工学院帕莱索分校，法国） 纠错

Comments NeurIPS 2025 camera ready version

Distilling LLM Agent into Small Models with Retrieval and Code Tools

Minki Kang, Jongwon Jeong, Seanie Lee, Jaewoong Cho, Sung Ju Hwang

机构 * KAIST（韩国科学技术院） ； University of Wisconsin-Madison（威斯康星大学麦迪逊分校） ； KRAFTON（KRAFTON公司） 纠错

Comments NeurIPS 2025 Spotlight

Traversal Verification for Speculative Tree Decoding

Yepeng Weng, Qiao Hu, Xujie Chen, Li Liu, Dianwen Mei, Huishi Qiu, Jiang Tian, Zhongchao Shi

机构 * Lenovo AI Technology Center, Lenovo（联想人工智能技术中心，联想） ； National Center for Mathematics and Interdisciplinary Sciences (NCMIS), AMSS, CAS（国家数学与交叉科学中心（NCMIS），中国科学院数学研究所） 纠错

Comments NeurIPS 2025 poster

Unifying Symbolic Music Arrangement: Track-Aware Reconstruction and Structured Tokenization

Longshen Ou, Jingwei Zhao, Ziyu Wang, Gus Xia, Qihao Liang, Torin Hopkins Ye Wang

机构 * Sound and Music Computing Lab, School of Computing, NUS（新加坡国立大学计算机学院声音与音乐计算实验室） ； Courant Institute of Mathematical Sciences, New York University（纽约大学应用数学科学学院） ； Music X Lab, MBZUAI（MBZUAI音乐X实验室） 纠错

Comments NeurIPS 2025 camera ready version

A Probabilistic U-Net Approach to Downscaling Climate Simulations

Maryam Alipourhajiagha, Pierre-Louis Lemaire, Youssef Diouane, Julie Carreau

机构 * Polytechnique Montréal（蒙特利尔理工学院） 纠错

Comments NeurIPS 2025 AI4Science

UnCLe: Towards Scalable Dynamic Causal Discovery in Non-linear Temporal Systems

Tingzhu Bi, Yicheng Pan, Xinrui Jiang, Huize Sun, Meng Ma, Ping Wang

机构 * Peking University（北京大学） 纠错

Comments 12 pages main content, 18 pages appendix, NeurIPS 2025. Code: https://github.com/etigerstudio/uncle-causal-discovery

No-Human in the Loop: Agentic Evaluation at Scale for Recommendation

Tao Zhang, Kehui Yao, Luyi Ma, Jiao Chen, Reza Yousefi Maragheh, Kai Zhao, Jianpeng Xu, Evren Korpeoglu, Sushant Kumar, Kannan Achan

机构 * Walmart Global Tech（沃尔玛全球技术） 纠错

Comments 4 page, NeurIPS 2025 Workshop: Evaluating the Evolving LLM Lifecycle

Data-Efficient Realized Volatility Forecasting with Vision Transformers

Emi Soroka, Artem Arzyn

机构 * Stanford University（斯坦福大学） 纠错

Comments NeurIPS Generative AI in Finance

Inference-Time Personalized Alignment with a Few User Preference Queries

Victor-Alexandru Pădurean, Parameswaran Kamalaruban, Nachiket Kotalwar, Alkis Gotovos, Adish Singla

机构 * MPI-SWS 纠错

Comments NeurIPS'25 paper

Domain-Adaptive Transformer for Data-Efficient Glioma Segmentation in Sub-Saharan MRI

Ilerioluwakiiye Abolade, Aniekan Udo, Augustine Ojo, Abdulbasit Oyetunji, Hammed Ajigbotosho, Aondana Iorumbur, Confidence Raymond, Maruf Adewole

机构 * Federal University of Agriculture Abeokuta（非洲联邦农业大学阿贝奥克塔） 纠错

Comments 4 pages, 2 figures. Accepted as an abstract at the Women in Machine Learning (WiML) Workshop at NeurIPS 2025

EEGReXferNet: A Lightweight Gen-AI Framework for EEG Subspace Reconstruction via Cross-Subject Transfer Learning and Channel-Aware Embedding

Shantanu Sarkar, Piotr Nabrzyski, Saurabh Prasad, Jose Luis Contreras-Vidal

机构 * IUCRC BRAIN Center, Cullen College of Engineering, University of Houston（IUCRC BRAIN中心、工程学院、休斯顿大学） ； Department of Computer Engineering, Purdue University（计算机工程系、普渡大学） 纠错

Comments Accepted for presentation at the NeurIPS 2025 Workshop on Foundation Models for the Brain and Body

PlanU: Large Language Model Reasoning through Planning under Uncertainty

Ziwei Deng, Mian Deng, Chenjing Liang, Zeming Gao, Chennan Ma, Chenxing Lin, Haipeng Zhang, Songzhu Mei, Siqi Shen, Cheng Wang

机构 * Fujian Key Laboratory of Sensing and Computing for Smart Cities, School of Informatics, Xiamen University（福建智慧城市感知与计算重点实验室，信息学院，厦门大学） ； Key Laboratory of Multimedia Trusted Perception and Efficient Computing, XMU（多媒体可信感知与高效计算重点实验室，XMU） ； School of Computer, National University of Defense Technology（计算机学院，国防科技大学） 纠错

Comments 38 pages, 19 figures, NeurIPS 2025 Accepted

SmartWilds: Multimodal Wildlife Monitoring Dataset

Jenna Kline, Anirudh Potlapally, Bharath Pillai, Tanishka Wani, Rugved Katole, Vedant Patil, Penelope Covey, Hari Subramoni, Tanya Berger-Wolf, Christopher Stewart

机构 * Department of Computer Science and Engineering（计算机科学与工程系） 纠错

Comments Accepted to Imageomics Workshop at Neurips 2025

Inference-Time Reward Hacking in Large Language Models

Hadi Khalaf, Claudio Mayrink Verdun, Alex Oesterling, Himabindu Lakkaraju, Flavio du Pin Calmon

机构 * Harvard University（哈佛大学） 纠错

Comments Accepted to NeurIPS 2025 (Spotlight Paper)

Recurrent Self-Attention Dynamics: An Energy-Agnostic Perspective from Jacobians

Akiyoshi Tomihari, Ryo Karakida

机构 * Artificial Intelligence Research Center, AIST, Japan（日本AIST人工智能研究中心） ； Department of Computer Science, The University of Tokyo, Japan（东京大学计算机科学系） ； RIKEN Center for Advanced Intelligence Project（日本理化学研究所高级智能项目中心） 纠错

Comments NeurIPS 2025 (poster). Some typos fixed

Revisiting semi-supervised learning in the era of foundation models

Ping Zhang, Zheda Mai, Quang-Huy Nguyen, Wei-Lun Chao

机构 * The Ohio State University（俄亥俄州立大学） 纠错

Comments The paper has been accepted to NeurIPS 2025. Ping Zhang and Zheda Mai contributed equally to this work