0

Masked Autoencoders are Efficient Class Incremental Learners

Masked Autoencoders (MAEs) integrated with a bilateral framework enhance Class Incremental Learning (CIL) by efficiently storing exemplars and achieving state-of-the-art performance.

Year
2023
Venue
ICCV 2023 1
Authors
5
Hosting
Abstract onlyARXIV-DEFAULT

Cite

Notes

Only stored in your browser.

Attribution

Abstract & full text
arxiv.org/abs/2308.12510ARXIV-DEFAULT
TL;DR
Semantic Scholar
Attribution policy →

Abstract

Class Incremental Learning (CIL) aims to sequentially learn new classes while avoiding catastrophic forgetting of previous knowledge. We propose to use Masked Autoencoders (MAEs) as efficient learners for CIL. MAEs were originally designed to learn useful representations through reconstructive unsupervised learning, and they can be easily integrated with a supervised loss for classification. Moreover, MAEs can reliably reconstruct original input images from randomly selected patches, which we use to store exemplars from past tasks more efficiently for CIL. We also propose a bilateral MAE framework to learn from image-level and embedding-level fusion, which produces better-quality reconstructed images and more stable representations. Our experiments confirm that our approach performs better than the state-of-the-art on CIFAR-100, ImageNet-Subset, and ImageNet-Full. The code is available at https://github.com/scok30/MAE-CIL .

Authors

5