Revisiting class-incremental learning with pre-trained models: generalizability and adaptivity are all you need

• Main Authors: Zhou, Da-Wei, Cai, Zi-Wen, Ye, Han-Jia, Zhan, De-Chuan, Liu, Ziwei
• 其他作者: College of Computing and Data Science
• 格式: Article
• 語言: English
• 出版: 2024
• 主題: Computer and Information Science; Catastrophic forgetting; Class-incremental learning
• 在線閱讀: https://hdl.handle.net/10356/181218

Class-incremental learning (CIL) aims to adapt to emerging new classes without forgetting old ones. Traditional CIL models are trained from scratch to continually acquire knowledge as data evolves. Recently, pre-training has achieved substantial progress, making vast pre-trained models (PTMs) access...