Meta Pseudo Labels

About

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.

Hieu Pham, Zihang Dai, Qizhe Xie, Minh-Thang Luong, Quoc V. Le · 2020

Related benchmarks

Task	Dataset	Result
Image Classification	CIFAR-100 (test)	Accuracy—	3,518
Image Classification	CIFAR-10 (test)	Accuracy—	3,381
Image Classification	ImageNet-1k (val)	Top-1 Accuracy—	1,498
Classification	ImageNet-1K 1.0 (val)	Top-1 Accuracy (%)90.2	1,171
Image Classification	ImageNet-1k (val)	Top-1 Acc90.2	706
Image Classification	CIFAR-100	Accuracy—	691
Image Classification	CIFAR10 (test)	Accuracy—	585
Image Classification	CIFAR-10	Accuracy—	507
Image Classification	SVHN (test)	Accuracy—	470
Image Classification	ImageNet ILSVRC-2012 (val)	Top-1 Accuracy90.2	441