DiffClass: Diffusion-Based Class Incremental Learning

TL;DR

DiffClass introduces a diffusion-based approach with multi-distribution matching and synthetic augmentation to effectively mitigate catastrophic forgetting in exemplar-free class incremental learning, achieving state-of-the-art results.

Contribution

The paper proposes a novel exemplar-free CIL method using diffusion models, multi-distribution matching, and synthetic augmentation to bridge domain gaps and improve incremental learning performance.

Findings

Outperforms previous exemplar-free CIL methods on benchmarks.

Effectively reduces catastrophic forgetting through domain gap mitigation.

Achieves state-of-the-art results in class incremental learning scenarios.

Abstract

Class Incremental Learning (CIL) is challenging due to catastrophic forgetting. On top of that, Exemplar-free Class Incremental Learning is even more challenging due to forbidden access to previous task data. Recent exemplar-free CIL methods attempt to mitigate catastrophic forgetting by synthesizing previous task data. However, they fail to overcome the catastrophic forgetting due to the inability to deal with the significant domain gap between real and synthetic data. To overcome these issues, we propose a novel exemplar-free CIL method. Our method adopts multi-distribution matching (MDM) diffusion models to unify quality and bridge domain gaps among all domains of training data. Moreover, our approach integrates selective synthetic image augmentation (SSIA) to expand the distribution of the training data, thereby improving the model's plasticity and reinforcing the performance of our method's ultimate component, multi-domain adaptation (MDA). With the proposed integrations, our method then reformulates exemplar-free CIL into a multi-domain adaptation problem to implicitly address the domain gap problem to enhance model stability during incremental training. Extensive experiments on benchmark class incremental datasets and settings demonstrate that our method excels previous exemplar-free CIL methods and achieves state-of-the-art performance.