ConMatch: Semi-Supervised Learning with Confidence-Guided Consistency Regularization

TL;DR

ConMatch introduces a confidence-guided consistency regularization framework for semi-supervised learning, leveraging pseudo-label confidence measures from strongly-augmented views to improve model performance.

Contribution

It proposes novel confidence measures for pseudo-labels and end-to-end learning of confidence within the network, enhancing semi-supervised learning effectiveness.

Findings

ConMatch outperforms recent semi-supervised methods in experiments.

Confidence learning improves pseudo-label reliability.

Stage-wise training boosts convergence.

Abstract

We present a novel semi-supervised learning framework that intelligently leverages the consistency regularization between the model's predictions from two strongly-augmented views of an image, weighted by a confidence of pseudo-label, dubbed ConMatch. While the latest semi-supervised learning methods use weakly- and strongly-augmented views of an image to define a directional consistency loss, how to define such direction for the consistency regularization between two strongly-augmented views remains unexplored. To account for this, we present novel confidence measures for pseudo-labels from strongly-augmented views by means of weakly-augmented view as an anchor in non-parametric and parametric approaches. Especially, in parametric approach, we present, for the first time, to learn the confidence of pseudo-label within the networks, which is learned with backbone model in an end-to-end manner. In addition, we also present a stage-wise training to boost the convergence of training. When incorporated in existing semi-supervised learners, ConMatch consistently boosts the performance. We conduct experiments to demonstrate the effectiveness of our ConMatch over the latest methods and provide extensive ablation studies. Code has been made publicly available at https://github.com/JiwonCocoder/ConMatch.