Learning in the Null Space: Small Singular Values for Continual Learning
在空域中学习:小奇异值用于持续学习
机构 * Mohamed bin Zayed University of Artificial Intelligence (MBZUAI)(穆罕默德·本·扎耶德人工智能大学) ; Massachusetts Institute of Technology (MIT)(麻省理工学院)
AI总结 NESS通过利用小奇异值构建空域,实现持续学习中的正交性约束,从而减少遗忘并提升任务适应性。
Comments 17 pages, accepted as Oral presentation at the Third Conference on Parsimony and Learning (CPAL 2026)