Interpolation consistency training for semi-supervised learning

doi:10.1016/j.neunet.2021.10.008

Return

Interpolation consistency training for semi-supervised learning

2022-01-01

207

DOI：10.1016/j.neunet.2021.10.008

Original

Save

View PDF

Abstract

En 中文

We introduce Interpolation Consistency Training (ICT), a simple and computation efficient algorithm for training Deep Neural Networks in the semi-supervised learning paradigm. ICT encourages the prediction at an interpolation of unlabeled points to be consistent with the interpolation of the predictions at those points. In classification problems, ICT moves the decision boundary to low-density regions of the data distribution. Our experiments show that ICT achieves state-of-the-art performance when applied to standard neural network architectures on the CIFAR-10 and SVHN benchmark datasets. Our theoretical analysis shows that ICT corresponds to a certain type of data-adaptive regularization with unlabeled points which reduces overfitting to labeled points under high confidence values. (C) 2021 The Authors. Published by Elsevier Ltd.

Keywords:

Semi-supervised learning

Deep Neural Networks

Mixup

Consistency regularization

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.