Online continual learning via the knowledge invariant and spread-out properties

作者：

Highlights：

• We proposed a novel online continual learning method via the knowledge invariant and spread-out properties.

• We analyze the learning bias of the replay-based method in online continual learning.

• The experimental results illustrate the effectiveness of our OCLKISP method in a continual learning setting.

摘要

•We proposed a novel online continual learning method via the knowledge invariant and spread-out properties.•We analyze the learning bias of the replay-based method in online continual learning.•The experimental results illustrate the effectiveness of our OCLKISP method in a continual learning setting.

论文关键词：Continual learning,Catastrophic forgetting,Knowledge invariant and spread-out properties,Knowledge transfer

论文评审过程：Received 20 December 2021, Revised 7 October 2022, Accepted 9 October 2022, Available online 15 October 2022, Version of Record 20 October 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.119004