Revisiting Consistency Regularization for Semi-Supervised Learning
International Journal of Computer Vision · 2023-03 · 131 (3) · 626-643
DOI: 10.1007/s11263-022-01723-4 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computer Vision, online erschienen am 07.12.2022. Verfasst von Yue Fan und 3 weiteren Autor:innen. Die Arbeit wurde bisher 118-mal zitiert (Crossref) und nennt 75 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Max Planck Institute for Informatics.
Abstract (Originalsprache)
Consistency regularization is one of the most widely-used techniques for semi-supervised learning (SSL). Generally, the aim is to train a model that is invariant to various data augmentations. In this paper, we revisit this idea and find that enforcing invariance by decreasing distances between features from differently augmented images leads to improved performance. However, encouraging equivariance instead, by increasing the feature distance, further improves performance. To this end, we propose an improved consistency regularization framework by a simple yet effective technique, FeatDistLoss, that imposes consistency and equivariance on the classifier and the feature level, respectively. Experimental results show that our model defines a new state of the art across a variety of standard semi-supervised learning benchmarks as well as imbalanced semi-supervised learning benchmarks. Particularly, we outperform previous work by a significant margin in low data regimes and at large imbalance ratios. Extensive experiments are conducted to analyze the method, and the code will be published.
Förderung
- Max Planck Institute for Informatics
Zitieren
APA
Fan, Y., Kukleva, A., Dai, D., & Schiele, B. (2023). Revisiting Consistency Regularization for Semi-Supervised Learning. International Journal of Computer Vision, 131(3), 626-643. https://doi.org/10.1007/s11263-022-01723-4
Vancouver
Fan Y, Kukleva A, Dai D, Schiele B. Revisiting Consistency Regularization for Semi-Supervised Learning. International Journal of Computer Vision. 2023;131(3):626-643. doi:10.1007/s11263-022-01723-4
BibTeX
@article{Fan2023,
author = {Fan, Yue and Kukleva, Anna and Dai, Dengxin and Schiele, Bernt},
title = {Revisiting Consistency Regularization for Semi-Supervised Learning},
journal = {International Journal of Computer Vision},
year = {2023},
volume = {131},
number = {3},
pages = {626--643},
doi = {10.1007/s11263-022-01723-4}
}
EUVIMED