VQToken: Neural Discrete Token Representation Learning for Extreme Token Reduction in Video Large Language Models
机构 * Northeastern University（东北大学）
Comments Proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS 2025)
期刊&会议
NeurIPS
Conference on Neural Information Processing Systems · 会议 · Machine Learning
Recurrent Memory for Online Interdomain Gaussian Processes
机构 * Imperial College London（伦敦帝国学院） ； Canon Inc.（佳能公司） ； University of Copenhagen（哥本哈根大学）
Comments Published at the 39th Conference on Neural Information Processing Systems (NeurIPS 2025)
UniTraj: Learning a Universal Trajectory Foundation Model from Billion-Scale Worldwide Traces
机构 * Southern University of Science and Technology（南方科技大学） ； City University of Hong Kong（香港城市大学） ； The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州）） ； Harbin Institute of Technology, Shenzhen（哈尔滨工业大学（深圳））
Comments Accepted as a research paper by NeurIPS'25
Neuro-Symbolic Entity Alignment via Variational Inference
机构 * Department of Computing（计算系） ； The Hong Kong Polytechnic University（香港理工大学）
Comments This paper has been accepted by NeurIPS 2025
The Prevalence of Neural Collapse in Neural Multivariate Regression
机构 * New York University Abu Dhabi（纽约大学阿布扎克分校） ； SFSC of AI and DL, NYU Shanghai（人工智能与深度学习学院（SFSC），纽约大学上海） ； New York University Shanghai（纽约大学上海） ； New York University（纽约大学）
Comments Published at NeurIPS 2024
Journal ref Published in 38th Conference on Neural Information Processing Systems (NeurIPS 2024)
Position: Towards Bidirectional Human-AI Alignment
机构 * NYU Shanghai, New York University（纽约大学上海研究院，纽约大学） ； MBZUAI（穆桑大学人工智能研究所） ； Microsoft（微软公司） ； University of Michigan（密歇根大学） ； Apple（苹果公司） ； Google（谷歌公司） ； Carnegie Mellon University（卡内基梅隆大学） ； Stanford University（斯坦福大学） ； Google DeepMind（谷歌DeepMind）
Comments NeurIPS 2025 Position Paper
EWC-Guided Diffusion Replay for Exemplar-Free Continual Learning in Medical Imaging
机构 * University of Cambridge（剑桥大学） ； University of Kent（肯特大学）
Comments Accepted at AI That Keeps Up: NeurIPS 2025 Workshop on Continual and Compatible Foundation Model Updates
Taught Well Learned Ill: Towards Distillation-conditional Backdoor Attack
机构 * State Key Laboratory of Blockchain and Data Security（区块链与数据安全国家重点实验室） ； Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security（杭州高新技术区（滨江）区块链与数据安全研究院） ； Nanyang Technological University（南洋理工大学）
Comments The first three authors contributed equally to this work. To appear in NeurIPS 2025. 35 pages
DexFlyWheel: A Scalable and Self-improving Data Generation Framework for Dexterous Manipulation
机构 * Harbin Institute of Technology（哈尔滨工业大学） ； PsiBot ； Peking University（北京大学） ； PKU-Psibot Lab（北京大学-PsiBot实验室） ； Harbin Institute of Technology Suzhou Research Institute（哈尔滨工业大学苏州研究所）
Comments NeurIPS 2025, Spotlight
STAIR: Addressing Stage Misalignment through Temporal-Aligned Preference Reinforcement Learning
机构 * Beijing Key Laboratory of Embodied Intelligence Systems, Department of Automation, Tsinghua University, Beijing, China（北京智能体智能系统重点实验室，自动化系，清华大学，北京，中国） ； The Key Laboratory of Cognition and Decision Intelligence for Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China（复杂系统认知与决策智能重点实验室，自动化研究所，中国科学院，北京，中国）
Comments NeurIPS 2025
GroupCoOp: Group-robust Fine-tuning via Group Prompt Learning
机构 * Pohang University of Science and Technology (POSTECH)（釜山科学技术大学） ； Tübingen AI Center（图宾根人工智能中心） ； Universität Tübingen（图宾根大学）
Comments This paper was first submitted to NeurIPS 2024 in May 2024
ResAD++: Towards Class Agnostic Anomaly Detection via Residual Feature Learning
机构 * School of Information Science and Electronic Engineering, Shanghai Jiao Tong University（上海交通大学信息科学与电子工程学院） ； School of Information Science and Technology, Beijing Forestry University（北京林业大学信息科学与技术学院）
Comments This paper is an extended version of our NeurIPS 2024 paper, ResAD. arXiv admin note: substantial text overlap with arXiv:2410.20047
Measuring Sparse Autoencoder Feature Sensitivity
机构 * The Harker School（哈克尔学校） ； Stanford University（斯坦福大学）
Comments NeurIPS 2025 Workshop on Mechanistic Interpretability Camera Ready
Text-to-Code Generation for Modular Building Layouts in Building Information Modeling
Comments Accepted at NeurIPS 2025 Main Track
Beyond Greedy Exits: Improved Early Exit Decisions for Risk Control and Reliability
机构 * Department of Industrial Engineering and Operations Research（工业工程与运营管理系） ； Indian Institute of Technology Bombay（印度理工学院孟买）
Comments Accepted as poster in NeurIPS 2025
Orientation-anchored Hyper-Gaussian for 4D Reconstruction from Casual Videos
Comments NeurIPS 2025. Code: \href{https://github.com/adreamwu/OriGS}{OriGS}
Factor Decorrelation Enhanced Data Removal from Deep Predictive Models
机构 * Wuhan University of Technology（武汉理工大学） ； University of Southern Queensland（南方科技术大学）
Comments accepted by NeurIPS 2025
PARROT: A Benchmark for Evaluating LLMs in Cross-System SQL Translation
Comments To appear in NeurIPS 2025. Welcome your submission to challenge our leaderboard at: https://code4db.github.io/parrot-bench/. Also visit our code repository at: https://github.com/weAIDB/PARROT
Statistical Inference for Gradient Boosting Regression
机构 * School of Economics, Fudan University（复旦大学经济学院） ； Department of Statistics and Data Science, The Wharton School, University of Pennsylvania（宾夕法尼亚大学沃顿商学院统计与数据科学系）
Comments Accepted to NeurIPS 2025
Meta-Learning Fourier Neural Operators for Hessian Inversion and Enhanced Variational Data Assimilation
机构 * School of Computational Science and Engineering, McMaster University（计算科学与工程学院，麦斯特大学） ； Mathematics and Statistics, McMaster University（数学与统计学，麦斯特大学） ； Department of Chemistry and Chemical Biology, McMaster University（化学与生物化学系，麦斯特大学）
Comments 6 pages, 2 figures, Machine Learning and the Physical Sciences Workshop, (NeurIPS 2025)
Model Context Protocol for Vision Systems: Audit, Security, and Protocol Extensions
Comments Accepted to NeurIPS 2025 Workshop on Bridging Language, Agent, and World Models for Reasoning and Planning (LAW 2025)
DEFT: Decompositional Efficient Fine-Tuning for Text-to-Image Models
机构 * Mohamed bin Zayed University of Artificial Intelligence（穆罕默德·本·扎耶德人工智能大学）
Comments 13 Figures, 21 pages, accepted in NeurIPS 2025
LayoutAgent: A Vision-Language Agent Guided Compositional Diffusion for Spatial Layout Planning
机构 * Personalization Team, Walmart Global Tech（Walmart全球科技个性化团队）
Comments NeurIPS 2025 Workshop on SPACE in Vision, Language, and Embodied AI
(When) Should We Delegate AI Governance to AIs? Some Lessons from Administrative Law
Comments 4 pages, Regulatable ML Workshop at NeurIPS '25
SQL-of-Thought: Multi-agentic Text-to-SQL with Guided Error Correction
机构 * Max Planck Institute for Software Systems（马克斯·普朗克软件系统研究所） ； AWS GenAI（亚马逊AWS生成式人工智能）
Comments Accepted at NeurIPS 2025, DL4C "Deep Learning for Code" workshop. Code is available at: https://github.com/shollercoaster/SQL-of-Thought
Overcoming Long-Context Limitations of State-Space Models via Context-Dependent Sparse Attention
机构 * Mila - Québec AI Institute（魁北克AI研究所） ； University of Montréal（蒙特利尔大学） ； HEC Montréal（蒙特利尔HEC学院） ； CIFAR AI Chair（CIFAR人工智能席位）
Comments Proceedings of the Thirty-Ninth Annual Conference on Neural Information Processing Systems, 21 pages, 9 figures
Constant Bit-size Transformers Are Turing Complete
机构 * Shenzhen International Center For Industrial And Applied Mathematics（深圳国际工业与应用数学中心） ； Shenzhen Research Institute of Big Data（深圳大数据研究院） ； CRRC Zhuzhou Institute（CRRC 长沙研究所） ； Tengen Intelligence Institute（腾格智能研究院）
Comments Accepted by NeurIPS 2025
Why Masking Diffusion Works: Condition on the Jump Schedule for Improved Discrete Diffusion
机构 * New York University（纽约大学）
Comments Published at Neurips 2025. Code available at: https://github.com/AlanNawzadAmin/SCUD
MineAnyBuild: Benchmarking Spatial Planning for Open-world AI Agents
机构 * Shenzhen Campus of Sun Yat-sen University（中山大学深圳校区） ； Shanghai Jiao Tong University（上海交通大学） ； Mohamed bin Zayed University of Artificial Intelligence（穆罕默德·本·扎耶德人工智能大学） ； Huawei Noah’s Ark Lab（华为诺亚实验室）
Comments Accepted by NeurIPS 2025 Datasets and Benchmarks Track
Fine-tuning Diffusion Policies with Backpropagation Through Diffusion Timesteps
机构 * Tsinghua University（清华大学）
Comments 9 pages for main text, 23 pages in total, submitted to Neurips, 13 figures
