Dual-Space Smoothness for Robust and Balanced LLM Unlearning
双空间平滑性用于鲁棒且平衡的LLM反学习
机构 * School of Data Science, The Chinese University of Hong Kong, Shenzhen(香港中文大学(深圳)数据科学学院) ; Department of Computer Science and Engineering, University of Notre Dame(圣母大学计算机科学与工程系)
专题命中 评测与基准 :LLM(title);large language model(abstract);language model(abstract);分类 cs.CL、cs.AI
AI总结 本文提出PRISM框架,通过双空间平滑性提升LLM反学习的鲁棒性和平衡性,有效对抗重学和劫持攻击。
Comments Accepted by ICLR 2026