We present Meta Pseudo Labels, a semi-supervised learning method that achieves a new state-of-the-art top-1 accuracy of 90.2% on ImageNet, which is 1.6% better than the existing state-of-the-art. Like Pseudo Labels, Meta Pseudo Labels has a teacher network to generate pseudo labels on unlabeled data to teach a student network. However, unlike Pseudo Labels where the teacher is fixed, the teacher in Meta Pseudo Labels is constantly adapted by the feedback of the student's performance on the labeled dataset. As a result, the teacher generates better pseudo labels to teach the student. Our code will be available at https://github.com/google-research/google-research/tree/master/meta_pseudo_labels.
@article{arxiv.2003.10580,
title = {Meta Pseudo Labels},
author = {Hieu Pham and Zihang Dai and Qizhe Xie and Minh-Thang Luong and Quoc V. Le},
journal= {arXiv preprint arXiv:2003.10580},
year = {2021}
}