AI资讯日报 - 2026/5/4

👨‍🔬 Ziyang Huang, Yi Cao, Ali K. Shargh, Jing Luo, Ruidong Mei, Mohd Zaki, Zhan Liu, Wyatt Bunstine, William Jurayj, Somdatta Goswami, Tyrel McQueen, Michael Shields, Jaafar El-Awady, Paulette Clancy, Benjamin Van Durme, Nicholas Andrews, William Walden, Daniel Khashabi

Make Your LVLM KV Cache More Lightweight

学术论文 ArXiv 重要度: 8

提出LightKV，通过跨模态消息传递压缩视觉token的KV缓存，减少40%计算量且保持性能。

👨‍🔬 Xihao Chen, Yangyang Guo, Roger Zimmermann

Position: agentic AI orchestration should be Bayes-consistent

学术论文 ArXiv 重要度: 7

主张agentic AI系统的编排层应遵循贝叶斯决策理论，以实现校准信念和效用感知决策。

👨‍🔬 Theodore Papamarkou, Pierre Alquier, Matthias Bauer, Wray Buntine, Andrew Davison, Gintare Karolina Dziugaite, Maurizio Filippone, Andrew Y. K. Foong, Vincent Fortuin, Dimitris Fouskakis, Jes Frellsen, Eyke Hüllermeier, Theofanis Karaletsos, Mohammad Emtiyaz Khan, Nikita Kotelevskii, Salem Lahlou, Yingzhen Li, Fang Liu, Clare Lyle, Thomas Möllenhoff, Konstantina Palla, Maxim Panov, Yusuf Sale, Kajetan Schweighofer, Artem Shelmanov, Siddharth Swaroop, Martin Trapp, Willem Waegeman, Andrew Gordon Wilson, Alexey Zaytsev

To Call or Not to Call: A Framework to Assess and Optimize LLM Tool Calling

学术论文 ArXiv 重要度: 7

提出评估LLM工具调用决策的框架，通过估计需要性和效用性来优化调用策略。

👨‍🔬 Qinyuan Wu, Soumi Das, Mahsa Amani, Arijit Nag, Seungeon Lee, Krishna P. Gummadi, Abhilasha Ravichander, Muhammad Bilal Zafar

GeoContra: From Fluent GIS Code to Verifiable Spatial Analysis with Geography-Grounded Repair

学术论文 ArXiv 重要度: 7

GeoContra框架通过地理合约验证和修复LLM生成的GIS代码，将空间正确率提升至80%以上。

👨‍🔬 Yinhao Xiao, Rongbo Xiao, Yihan Zhang

EASE: Federated Multimodal Unlearning via Entanglement-Aware Anchor Closure

学术论文 ArXiv 重要度: 6

提出EASE框架，通过解缠感知子空间切除实现联邦多模态遗忘学习，效果接近重训练。

👨‍🔬 Zihao Ding, Beining Wu, Jun Huang

Meritocratic Fairness in Budgeted Combinatorial Multi-armed Bandits via Shapley Values

学术论文 ArXiv 重要度: 6

提出K-Shapley值和K-SVFair-FBF算法，在预算组合多臂赌博机中实现功绩公平性。

👨‍🔬 Shradha Sharma, Swapnil Dhamal, Shweta Jain

Directed Social Regard: Surfacing Targeted Advocacy, Opposition, Aid, Harms, and Victimization in Online Media

学术论文 ArXiv 重要度: 6

提出有向社会关注（DSR）方法，对在线媒体中多目标、多情感极性的文本进行跨度级分析。

👨‍🔬 Scott Friedman, Ruta Wheelock, Sonja Schmer-Galunder, Drisana Iverson, Jake Vasilakes, Joan Zheng, Jeffrey Rye, Vasanth Sarathy, Christopher Miller

Unsupervised Denoising of Real Clinical Low Dose Liver CT with Perceptual Attention Networks

学术论文 ArXiv 重要度: 5

提出无监督低剂量CT去噪框架，结合U-Net、注意力机制和感知损失，满足临床需求。

👨‍🔬 Jingxi Pu, Tonghua Liu, Zhilin Guan, Siqiao Li, Yang Ming, Zheng Cong, Wei Zhang, Fangwei Li

Empowering Heterogeneous Graph Foundation Models via Decoupled Relation Alignment

学术论文 ArXiv 重要度: 5

提出解耦关系子空间对齐（DRSA），增强异构图表征的跨域和少样本迁移能力。

👨‍🔬 Ziyu Zheng, Yaming Yang, Zhe Wang, Ziyu Guan, Wei Zhao

🤖 AI资讯日报

📊 今日趋势总结

MIT Non-AI License

Why Boring Businesses Outlast AI Hype Cycles

Ask HN: Anyone concerned about NYC Local Law 144?

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

The AI Crackpot Index

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

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

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

Show HN: Startup Raising capital through Book Sales

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

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

Bioinformatician

When RAG Chatbots Expose Their Backend: An Anonymized Case Study of Privacy and Security Risks in Patient-Facing Medical AI

Persistent Visual Memory: Sustaining Perception for Deep Generation in LVLMs

Can Coding Agents Reproduce Findings in Computational Materials Science?

Make Your LVLM KV Cache More Lightweight

Position: agentic AI orchestration should be Bayes-consistent

To Call or Not to Call: A Framework to Assess and Optimize LLM Tool Calling

GeoContra: From Fluent GIS Code to Verifiable Spatial Analysis with Geography-Grounded Repair

EASE: Federated Multimodal Unlearning via Entanglement-Aware Anchor Closure

Meritocratic Fairness in Budgeted Combinatorial Multi-armed Bandits via Shapley Values

Directed Social Regard: Surfacing Targeted Advocacy, Opposition, Aid, Harms, and Victimization in Online Media

Unsupervised Denoising of Real Clinical Low Dose Liver CT with Perceptual Attention Networks

Empowering Heterogeneous Graph Foundation Models via Decoupled Relation Alignment

📅 历史日报目录