AI资讯日报 - 2026/1/10

👨‍🔬 Shih-Yang Liu, Xin Dong, Ximing Lu, Shizhe Diao, Peter Belcak, Mingjie Liu, Min-Hung Chen, Hongxu Yin, Yu-Chiang Frank Wang, Kwang-Ting Cheng, Yejin Choi, Jan Kautz, Pavlo Molchanov

Robust Reasoning as a Symmetry-Protected Topological Phase

学术论文 ArXiv 重要度: 9

将鲁棒推理建模为对称性保护拓扑相，提出Holonomic Network，在噪声下保持逻辑一致性，超越Transformer的泛化能力。

👨‍🔬 Ilmo Sung

Internal Representations as Indicators of Hallucinations in Agent Tool Selection

学术论文 ArXiv 重要度: 8

提出利用LLM内部表示实时检测工具调用幻觉的框架，实现高效检测，提升智能体可靠性。

👨‍🔬 Kait Healy, Bharathi Srinivasan, Visakh Madathil, Jing Wu

Mechanisms of Prompt-Induced Hallucination in Vision-Language Models

学术论文 ArXiv 重要度: 8

研究视觉语言模型中提示诱导幻觉机制，发现关键注意力头，通过剪枝可显著减少幻觉，增强视觉证据遵循。

👨‍🔬 William Rudman, Michal Golovanevsky, Dana Arad, Yonatan Belinkov, Ritambhara Singh, Carsten Eickhoff, Kyle Mahowald

SimuAgent: An LLM-Based Simulink Modeling Assistant Enhanced with Reinforcement Learning

学术论文 ArXiv 重要度: 8

推出SimuAgent，基于LLM的Simulink建模助手，采用ReGRPO强化学习优化，提升建模效率与准确性。

👨‍🔬 Yanchang Liang, Xiaowei Zhao

RoboVIP: Multi-View Video Generation with Visual Identity Prompting Augments Robot Manipulation

学术论文 ArXiv 重要度: 7

提出视觉身份提示方法，生成多视角连贯机器人操作视频，增强数据多样性，提升策略模型性能。

👨‍🔬 Boyang Wang, Haoran Zhang, Shujie Zhang, Jinkun Hao, Mingda Jia, Qi Lv, Yucheng Mao, Zhaoyang Lyu, Jia Zeng, Xudong Xu, Jiangmiao Pang

Learning Latent Action World Models In The Wild

学术论文 ArXiv 重要度: 7

研究从野外视频学习潜在动作世界模型，解决环境噪声与缺乏共同体现的挑战，实现动作泛化与规划。

👨‍🔬 Quentin Garrido, Tushar Nagarajan, Basile Terver, Nicolas Ballas, Yann LeCun, Michael Rabbat

Observations and Remedies for Large Language Model Bias in Self-Consuming Performative Loop

学术论文 ArXiv 重要度: 7

研究LLM在自消耗循环中的偏见演化，提出奖励拒绝采样策略以减轻偏见，促进可信自改进系统。

👨‍🔬 Yaxuan Wang, Zhongteng Cai, Yujia Bao, Xueru Zhang, Yang Liu

CAOS: Conformal Aggregation of One-Shot Predictors

学术论文 ArXiv 重要度: 6

提出CAOS框架，聚合多个单样本预测器并进行留一校准，在有限数据下提供有效的不确定性量化与更小预测集。

👨‍🔬 Maja Waldron

FaST: Efficient and Effective Long-Horizon Forecasting for Large-Scale Spatial-Temporal Graphs via Mixture-of-Experts

学术论文 ArXiv 重要度: 6

提出FaST框架，基于混合专家模型实现大规模时空图的长时预测，兼顾预测精度与计算效率。

👨‍🔬 Yiji Zhao, Zihao Zhong, Ao Wang, Haomin Wen, Ming Jin, Yuxuan Liang, Huaiyu Wan, Hao Wu

MineNPC-Task: Task Suite for Memory-Aware Minecraft Agents

学术论文 ArXiv 重要度: 5

发布MineNPC-Task基准，用于评估开放世界Minecraft中记忆感知LLM智能体，支持透明可复现评估。

👨‍🔬 Tamil Sudaravan Mohan Doss, Michael Xu, Sudha Rao, Andrew D. Wilson, Balasaravanan Thoravi Kumaravel

Stock Market Price Prediction using Neural Prophet with Deep Neural Network

学术论文 ArXiv 重要度: 4

提出NP-DNN模型预测股价，结合Neural Prophet与MLP，声称达到高精度，但方法较为传统。

👨‍🔬 Navin Chhibber, Suneel Khemka, Navneet Kumar Tyagi, Rohit Tewari, Bireswar Banerjee, Piyush Ranjan

🤖 AI资讯日报

📊 今日趋势总结

Why Boring Businesses Outlast AI Hype Cycles

Ask HN: Anyone concerned about NYC Local Law 144?

MIT Non-AI License

NLP, AI, ML, bots – a passing trend or much more? What's your take on this?

Ask HN: Is the rate of progress in AI exponential?

Ask HN: What's the pain using current AI algorithms?

The AI Crackpot Index

Ask HN: What would you read to learn about "artificial intelligence"?

Common Lisp + Machine Learning Internship at Google (Mountain View, CA)

Show HN: Startup Raising capital through Book Sales

Bioinformatician

The Next Bill Gates or Albert Einstein in AI “Chris Clark” – Yourobot

GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization

Robust Reasoning as a Symmetry-Protected Topological Phase

Internal Representations as Indicators of Hallucinations in Agent Tool Selection

Mechanisms of Prompt-Induced Hallucination in Vision-Language Models

SimuAgent: An LLM-Based Simulink Modeling Assistant Enhanced with Reinforcement Learning

RoboVIP: Multi-View Video Generation with Visual Identity Prompting Augments Robot Manipulation

Learning Latent Action World Models In The Wild

Observations and Remedies for Large Language Model Bias in Self-Consuming Performative Loop

CAOS: Conformal Aggregation of One-Shot Predictors

FaST: Efficient and Effective Long-Horizon Forecasting for Large-Scale Spatial-Temporal Graphs via Mixture-of-Experts

MineNPC-Task: Task Suite for Memory-Aware Minecraft Agents

Stock Market Price Prediction using Neural Prophet with Deep Neural Network

📅 历史日报目录