Co$^2$L: Contrastive Continual Learning

TL;DR

This paper introduces Co$^2$L, a contrastive continual learning method that enhances transferability and robustness of learned representations, outperforming existing approaches in image classification benchmarks.

Contribution

It demonstrates that contrastive learning improves continual learning by reducing catastrophic forgetting and proposes a novel rehearsal-based algorithm with self-supervised distillation.

Findings

Achieves state-of-the-art results on benchmark datasets.

Contrastive representations are more robust against forgetting.

The proposed method outperforms joint-training baselines.

Abstract

Recent breakthroughs in self-supervised learning show that such algorithms learn visual representations that can be transferred better to unseen tasks than joint-training methods relying on task-specific supervision. In this paper, we found that the similar holds in the continual learning con-text: contrastively learned representations are more robust against the catastrophic forgetting than jointly trained representations. Based on this novel observation, we propose a rehearsal-based continual learning algorithm that focuses on continually learning and maintaining transferable representations. More specifically, the proposed scheme (1) learns representations using the contrastive learning objective, and (2) preserves learned representations using a self-supervised distillation step. We conduct extensive experimental validations under popular benchmark image classification datasets, where our method sets the new state-of-the-art performance.