Dynamic Residual Classifier for Class Incremental Learning

TL;DR

This paper introduces a Dynamic Residual Classifier (DRC) that effectively addresses data imbalance in class incremental learning, significantly improving performance across benchmarks by handling the dynamic nature of data imbalance.

Contribution

The paper proposes a novel DRC that adapts to data imbalance in CIL and integrates with existing pipelines to enhance their effectiveness.

Findings

Achieves state-of-the-art results on CIL benchmarks.

Effectively handles data imbalance with the DRC.

Compatible with various CIL pipelines.

Abstract

The rehearsal strategy is widely used to alleviate the catastrophic forgetting problem in class incremental learning (CIL) by preserving limited exemplars from previous tasks. With imbalanced sample numbers between old and new classes, the classifier learning can be biased. Existing CIL methods exploit the long-tailed (LT) recognition techniques, e.g., the adjusted losses and the data re-sampling methods, to handle the data imbalance issue within each increment task. In this work, the dynamic nature of data imbalance in CIL is shown and a novel Dynamic Residual Classifier (DRC) is proposed to handle this challenging scenario. Specifically, DRC is built upon a recent advance residual classifier with the branch layer merging to handle the model-growing problem. Moreover, DRC is compatible with different CIL pipelines and substantially improves them. Combining DRC with the model adaptation and fusion (MAF) pipeline, this method achieves state-of-the-art results on both the conventional CIL and the LT-CIL benchmarks. Extensive experiments are also conducted for a detailed analysis. The code is publicly available.