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.
@article{arxiv.2112.05025,
title = {Gradient-matching coresets for continual learning},
author = {Lukas Balles and Giovanni Zappella and Cédric Archambeau},
journal= {arXiv preprint arXiv:2112.05025},
year = {2021}
}
Comments
Accepted at the NeurIPS '21 Workshop on Distribution Shifts