cs.LG（2026-01-22）

📊 共 9 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

#	题目	一句话要点	标签
1	Beat-ssl: Capturing Local ECG Morphology through Heartbeat-level Contrastive Learning with Soft Targets	Beat-SSL：通过心跳级对比学习和软目标捕获局部ECG形态	contrastive learning foundation model
2	Integrating Knowledge Distillation Methods: A Sequential Multi-Stage Framework	提出SMSKD：一种序列多阶段知识蒸馏框架，用于整合异构知识蒸馏方法。	teacher-student distillation
3	Beyond Predictive Uncertainty: Reliable Representation Learning with Structural Constraints	提出结构约束下的可靠表征学习框架，提升表征的稳定性和鲁棒性	representation learning
4	Robust Tool Use via Fission-GRPO: Learning to Recover from Execution Errors	Fission-GRPO：通过分解错误轨迹和在线重采样，提升LLM工具使用中的错误恢复能力	reinforcement learning large language model
5	When Sharpening Becomes Collapse: Sampling Bias and Semantic Coupling in RL with Verifiable Rewards	针对可验证奖励的强化学习，提出逆向成功优势校准和分布级别校准，缓解过拟合问题。	reinforcement learning large language model

#	题目	一句话要点	标签	🔗
6	Provable Robustness in Multimodal Large Language Models via Feature Space Smoothing	提出特征空间平滑方法以增强多模态大语言模型的鲁棒性	large language model multimodal
7	Attributing and Exploiting Safety Vectors through Global Optimization in Large Language Models	提出GOSV框架，通过全局优化识别大语言模型中的安全向量，提升白盒攻击效果。	large language model
8	Next Generation Active Learning: Mixture of LLMs in the Loop	提出Mixture of LLMs in the Loop主动学习框架，提升LLM标注质量并降低标注成本。	large language model
9	Towards Automated Kernel Generation in the Era of LLMs	综述：利用大语言模型实现自动化内核生成与优化	large language model	✅