Multi-Class Incremental Learning (MCIL) aims to learn new concepts by incrementally updating a model trained on previous concepts. However, there is an inherent trade-off to effectively learning new concepts without catastrophic forgetting of previous ones. To alleviate this issue, it has been proposed to keep around a few examples of the previous concepts but the effectiveness of this approach heavily depends on the representativeness of these examples. This paper proposes a novel and automatic framework we call mnemonics, where we parameterize exemplars and make them optimizable in an end-to-end manner. We train the framework through bilevel optimizations, i.e., model-level and exemplar-level. We conduct extensive experiments on three MCIL benchmarks, CIFAR-100, ImageNet-Subset and ImageNet, and show that using mnemonics exemplars can surpass the state-of-the-art by a large margin. Interestingly and quite intriguingly, the mnemonics exemplars tend to be on the boundaries between different classes.
@article{arxiv.2002.10211,
title = {Mnemonics Training: Multi-Class Incremental Learning without Forgetting},
author = {Yaoyao Liu and Yuting Su and An-An Liu and Bernt Schiele and Qianru Sun},
journal= {arXiv preprint arXiv:2002.10211},
year = {2021}
}
Comments
Experiment results updated (different from the conference version). Code is available at https://github.com/yaoyao-liu/mnemonics-training