Interpolation consistency training for semi-supervised learning

Vikas Verma*, Kenji Kawaguchi, Alex Lamb, Juho Kannala, Arno Solin, Yoshua Bengio, David Lopez-Paz

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

28 Downloads (Pure)

Abstract

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.
Original languageEnglish
Pages (from-to)90-106
JournalNeural Networks
Volume145
Early online date2021
DOIs
Publication statusPublished - 2022
MoE publication typeA1 Journal article-refereed

Fingerprint

Dive into the research topics of 'Interpolation consistency training for semi-supervised learning'. Together they form a unique fingerprint.

Cite this