ER-FSL: Experience Replay with Feature Subspace Learning for Online Continual Learning

TL;DR

ER-FSL introduces a novel approach to online continual learning by learning in feature subspaces and replaying in a larger space, effectively reducing catastrophic forgetting and outperforming existing methods.

Contribution

The paper proposes ER-FSL, a new method that divides feature space into subspaces for learning and replays in an accumulated space, addressing feature interference in continual learning.

Findings

ER-FSL outperforms state-of-the-art methods on three datasets.

Learning in feature subspaces helps retain old knowledge while learning new data.

Replaying in an accumulated feature space effectively mitigates forgetting.

Abstract

Online continual learning (OCL) involves deep neural networks retaining knowledge from old data while adapting to new data, which is accessible only once. A critical challenge in OCL is catastrophic forgetting, reflected in reduced model performance on old data. Existing replay-based methods mitigate forgetting by replaying buffered samples from old data and learning current samples of new data. In this work, we dissect existing methods and empirically discover that learning and replaying in the same feature space is not conducive to addressing the forgetting issue. Since the learned features associated with old data are readily changed by the features related to new data due to data imbalance, leading to the forgetting problem. Based on this observation, we intuitively explore learning and replaying in different feature spaces. Learning in a feature subspace is sufficient to capture novel knowledge from new data while replaying in a larger feature space provides more feature space to maintain historical knowledge from old data. To this end, we propose a novel OCL approach called experience replay with feature subspace learning (ER-FSL). Firstly, ER-FSL divides the entire feature space into multiple subspaces, with each subspace used to learn current samples. Moreover, it introduces a subspace reuse mechanism to address situations where no blank subspaces exist. Secondly, ER-FSL replays previous samples using an accumulated space comprising all learned subspaces. Extensive experiments on three datasets demonstrate the superiority of ER-FSL over various state-of-the-art methods.