Gradient-matching coresets for continual learning

TL;DR

This paper introduces a gradient-matching coreset selection method for continual learning, aiming to create compact rehearsal memories that closely replicate the original data's gradient information, thereby improving learning efficiency.

Contribution

It proposes a novel gradient-matching approach for coreset selection in continual learning, outperforming traditional sampling methods like reservoir sampling.

Findings

Performs competitively across various memory sizes

Effectively matches gradients of original datasets

Enhances rehearsal memory quality in continual learning

Abstract

We devise a coreset selection method based on the idea of gradient matching: The gradients induced by the coreset should match, as closely as possible, those induced by the original training dataset. We evaluate the method in the context of continual learning, where it can be used to curate a rehearsal memory. Our method performs strong competitors such as reservoir sampling across a range of memory sizes.