cs.LG（2025-09-13）

📊 共 6 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

#	题目	一句话要点	标签	🔗
1	The Psychogenic Machine: Simulating AI Psychosis, Delusion Reinforcement and Harm Enablement in Large Language Models	Psychosis-bench：评估大型语言模型潜在精神病性风险及危害促成能力	large language model
2	A Traditional Approach to Symbolic Piano Continuation	针对钢琴音乐续创任务，提出一种基于传统方法和优化数据的序列预测方案。	foundation model	✅
3	Rethinking Sparse Autoencoders: Select-and-Project for Fairness and Control from Encoder Features Alone	提出S&P Top-K，一种encoder-centric的稀疏自编码器改进方法，用于提升模型公平性和可控性。	large language model
4	GoldenTransformer: A Modular Fault Injection Framework for Transformer Robustness Research	GoldenTransformer：用于Transformer鲁棒性研究的模块化故障注入框架	large language model

#	题目	一句话要点	标签	🔗	⭐
5	Research on Short-Video Platform User Decision-Making via Multimodal Temporal Modeling and Reinforcement Learning	提出MT-DQN模型，通过多模态时序建模和强化学习优化短视频用户决策预测与推荐。	reinforcement learning MAE multimodal
6	Predictive Free Energy Simulations Through Hierarchical Distillation of Quantum Hamiltonians	提出分层蒸馏量子哈密顿量方法，加速凝聚相化学反应自由能预测。	distillation