Revisiting Consistency Regularization for Semi-Supervised Learning
This work addresses semi-supervised learning for researchers and practitioners, offering incremental improvements to an existing technique.
The paper tackles the problem of improving consistency regularization in semi-supervised learning by proposing FeatDistLoss, which enforces consistency and equivariance at different levels, resulting in state-of-the-art performance across datasets, especially in low-data settings.
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 for various datasets and settings and outperforms previous work by a significant margin, particularly in low data regimes. Extensive experiments are conducted to analyze the method, and the code will be published.