DualMatch: Promoting Semi-Supervised Learning with Hierarchical Label and Contrastive Learning

Jianlong Wu, Haozhe Yang, Tian Gan, Ning Ding, Feijun Jiang, Liqiang Nie

Keywords: Semi-supervised Learning, Contrastive Learning, Hierarchical Label Matching

Abstract: The recently proposed FixMatch and FlexMatch have achieved remarkable results in the field of semi-supervised learning. But these two methods go to two extremes as FixMatch and FlexMatch use a pre-defined constant threshold for all classes and an adaptive threshold for each category, respectively. By only investigating consistency regularization, they suffer from unstable results and indiscriminative feature representation, especially under the situation of few labeled samples. In this paper, we propose a novel DualMatch method, which can learn an adaptive threshold for all classes to perform instance-level prediction matching as well as discriminative features by graph matching based contrastive learning. We first present a memory-bank based near-global threshold learning strategy to select highly-confident samples. In the meantime, we make full use of the structured information in the hierarchical labels to learn an accurate affinity graph for contrastive learning. DualMatch achieves very stable and superior results on several commonly-used benchmarks. For example, DualMatch achieves 8.44% and 9.02% error rate reduction over FlexMatch on CIFAR-100 under WRN-28-2 with only 4 and 25 labeled samples per class, respectively.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Unsupervised and Self-supervised learning