Towards Causal Replay for Knowledge Rehearsal in Continual Learning

Nikhil Churamani, Jiaee Cheong, Sinan Kalkan, Hatice Gunes

Keywords: Continual Learning, Causality, Rehearsal, Pseudo-rehearsal, Replay

TL;DR: This paper proposes the use of causality to improve replay-based Continual Learning of information.

Abstract: Given the challenges associated with the real-world deployment of Machine Learning (ML) models, especially towards efficiently integrating novel information on-the-go, both Continual Learning (CL) and Causality have been proposed and investigated individually as potent solutions. Despite their complimentary nature, the bridge between them is still largely unexplored. In this work, we focus on causality to improve the learning and knowledge preservation capabilities of CL models. In particular, positing Causal Replay for knowledge rehearsal, we discuss how CL-based models can benefit from causal interventions towards improving their ability to replay past knowledge in order to mitigate forgetting.