AI资讯日报 - 2025/9/4

👨‍🔬 Xingxuan Zhang, Gang Ren, Han Yu, Hao Yuan, Hui Wang, Jiansheng Li, Jiayun Wu, Lang Mo, Li Mao, Mingchao Hao, Ningbo Dai, Renzhe Xu, Shuyang Li, Tianyang Zhang, Yue He, Yuanrui Wang, Yunjia Zhang, Zijing Xu, Dongzhe Li, Fang Gao, Hao Zou, Jiandong Liu, Jiashuo Liu, Jiawei Xu, Kaijie Cheng, Kehan Li, Linjun Zhou, Qing Li, Shaohua Fan, Xiaoyu Lin, Xinyan Han, Xuanyue Li, Yan Lu, Yuan Xue, Yuanyuan Jiang, Zimu Wang, Zhenlei Wang, Peng Cui

SafeProtein: Red-Teaming Framework and Benchmark for Protein Foundation Models

学术论文 ArXiv 重要度: 8

SafeProtein是首个蛋白质基础模型红队测试框架，揭示高达70%的安全风险攻击成功率。

👨‍🔬 Jigang Fan, Zhenghong Zhou, Ruofan Jin, Le Cong, Mengdi Wang, Zaixi Zhang

Warming Up for Zeroth-Order Federated Pre-Training with Low Resource Clients

学术论文 ArXiv 重要度: 7

ZOWarmUp算法使低资源设备能参与联邦学习预训练，提升数据多样性和训练效果。

👨‍🔬 Gwen Legate, Irina Rish, Eugene Belilovsky

Strefer: Empowering Video LLMs with Space-Time Referring and Reasoning via Synthetic Instruction Data

学术论文 ArXiv 重要度: 7

Strefer框架生成合成指令数据，增强视频LLM的时空指代和推理能力，无需人工标注。

👨‍🔬 Honglu Zhou, Xiangyu Peng, Shrikant Kendre, Michael S. Ryoo, Silvio Savarese, Caiming Xiong, Juan Carlos Niebles

DPQuant: Efficient and Differentially-Private Model Training via Dynamic Quantization Scheduling

学术论文 ArXiv 重要度: 7

DPQuant动态量化框架减少DP-SGD中的量化方差，实现高效隐私保护训练，精度损失小于2%。

👨‍🔬 Yubo Gao, Renbo Tu, Gennady Pekhimenko, Nandita Vijaykumar

Multi-level SSL Feature Gating for Audio Deepfake Detection

学术论文 ArXiv 重要度: 7

提出多层级特征门控机制检测音频深度伪造，在跨域和 multilingual 数据上表现优异。

👨‍🔬 Hoan My Tran, Damien Lolive, Aghilas Sini, Arnaud Delhay, Pierre-François Marteau, David Guennec

Real-Time Instrument Planning and Perception for Novel Measurements of Dynamic Phenomena

学术论文 ArXiv 重要度: 6

自动化工作流结合检测与轨迹规划，实现动态现象（如火山羽流）的高分辨率测量。

👨‍🔬 Itai Zilberstein, Alberto Candela, Steve Chien

On Entropy Control in LLM-RL Algorithms

学术论文 ArXiv 重要度: 6

提出AEnt方法改进LLM-RL中的熵控制，通过压缩熵奖励提升数学推理任务性能。

👨‍🔬 Han Shen

Robult: Leveraging Redundancy and Modality Specific Features for Robust Multimodal Learning

学术论文 ArXiv 重要度: 6

Robult框架通过信息论方法提升多模态学习鲁棒性，有效处理缺失模态和有限标注数据。

👨‍🔬 Duy A. Nguyen, Abhi Kamboj, Minh N. Do

Continuous Saudi Sign Language Recognition: A Vision Transformer Approach

学术论文 ArXiv 重要度: 6

提出首个沙特手语连续句子数据集KAU-CSSL和Transformer模型，识别准确率达99.02%。

👨‍🔬 Soukeina Elhassen, Lama Al Khuzayem, Areej Alhothali, Ohoud Alzamzami, Nahed Alowaidi

sam-llm: interpretable lane change trajectoryprediction via parametric finetuning

学术论文 ArXiv 重要度: 6

SAM-LLM结合LLM和运动学模型，参数化预测可解释的车道变换轨迹，输出大小减少80%。

👨‍🔬 Zhuo Cao, Yunxiao Shi, Min Xu

🤖 AI资讯日报

📊 今日趋势总结

Why Boring Businesses Outlast AI Hype Cycles

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

Ask HN: Anyone concerned about NYC Local Law 144?

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?

The AI Crackpot Index

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

Ask HN: Dipping my toes with artificial intelligence and what to expect? (CS)

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

Bioinformatician

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

Show HN: Startup Raising capital through Book Sales

Can the Waymo Open Motion Dataset Support Realistic Behavioral Modeling? A Validation Study with Naturalistic Trajectories

LimiX: Unleashing Structured-Data Modeling Capability for Generalist Intelligence

SafeProtein: Red-Teaming Framework and Benchmark for Protein Foundation Models

Warming Up for Zeroth-Order Federated Pre-Training with Low Resource Clients

Strefer: Empowering Video LLMs with Space-Time Referring and Reasoning via Synthetic Instruction Data

DPQuant: Efficient and Differentially-Private Model Training via Dynamic Quantization Scheduling

Multi-level SSL Feature Gating for Audio Deepfake Detection

Real-Time Instrument Planning and Perception for Novel Measurements of Dynamic Phenomena

On Entropy Control in LLM-RL Algorithms

Robult: Leveraging Redundancy and Modality Specific Features for Robust Multimodal Learning

Continuous Saudi Sign Language Recognition: A Vision Transformer Approach

sam-llm: interpretable lane change trajectoryprediction via parametric finetuning

📅 历史日报目录