Dual-CBA: Improving Online Continual Learning via Dual Continual Bias Adaptors from a Bi-level Optimization Perspective

TL;DR

This paper introduces Dual-CBA, a novel framework for online continual learning that combines class-specific and class-agnostic bias adaptors to better handle distribution shifts and improve knowledge retention.

Contribution

It proposes a dual CBA module integrating class-specific and class-agnostic adaptors, along with Incremental Batch Normalization, to address catastrophic forgetting and distribution shifts in online CL.

Findings

Dual-CBA outperforms baseline methods on multiple benchmarks.

The method effectively mitigates catastrophic distribution shifts.

Theoretical analysis supports the empirical improvements.

Abstract

In online continual learning (CL), models trained on changing distributions easily forget previously learned knowledge and bias toward newly received tasks. To address this issue, we present Continual Bias Adaptor (CBA), a bi-level framework that augments the classification network to adapt to catastrophic distribution shifts during training, enabling the network to achieve a stable consolidation of all seen tasks. However, the CBA module adjusts distribution shifts in a class-specific manner, exacerbating the stability gap issue and, to some extent, fails to meet the need for continual testing in online CL. To mitigate this challenge, we further propose a novel class-agnostic CBA module that separately aggregates the posterior probabilities of classes from new and old tasks, and applies a stable adjustment to the resulting posterior probabilities. We combine the two kinds of CBA modules into a unified Dual-CBA module, which thus is capable of adapting to catastrophic distribution shifts and simultaneously meets the real-time testing requirements of online CL. Besides, we propose Incremental Batch Normalization (IBN), a tailored BN module to re-estimate its population statistics for alleviating the feature bias arising from the inner loop optimization problem of our bi-level framework. To validate the effectiveness of the proposed method, we theoretically provide some insights into how it mitigates catastrophic distribution shifts, and empirically demonstrate its superiority through extensive experiments based on four rehearsal-based baselines and three public continual learning benchmarks.