CGT: Consistency Guided Training in Semi-Supervised Learning

Nesreen Hasan, Farzin Ghorban, Jörg Velten, Anton Kummert

2022

Abstract

We propose a framework, CGT, for semi-supervised learning (SSL) that involves a unification of multiple image-based augmentation techniques. More specifically, we utilize Mixup and CutMix in addition to introducing one-sided stochastically augmented versions of those operators. Moreover, we introduce a generalization of the Mixup operator that regularizes a larger region of the input space. The objective of CGT is expressed as a linear combination of multiple constituents, each corresponding to the contribution of a different augmentation technique. CGT achieves state-of-the-art performance on the SVHN, CIFAR-10, and CIFAR-100 benchmark datasets and demonstrates that it is beneficial to heavily augment unlabeled training data.

Download


Paper Citation


in Harvard Style

Hasan N., Ghorban F., Velten J. and Kummert A. (2022). CGT: Consistency Guided Training in Semi-Supervised Learning. In Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2022) - Volume 5: VISAPP; ISBN 978-989-758-555-5, SciTePress, pages 55-64. DOI: 10.5220/0010771700003124


in Bibtex Style

@conference{visapp22,
author={Nesreen Hasan and Farzin Ghorban and Jörg Velten and Anton Kummert},
title={CGT: Consistency Guided Training in Semi-Supervised Learning},
booktitle={Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2022) - Volume 5: VISAPP},
year={2022},
pages={55-64},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010771700003124},
isbn={978-989-758-555-5},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2022) - Volume 5: VISAPP
TI - CGT: Consistency Guided Training in Semi-Supervised Learning
SN - 978-989-758-555-5
AU - Hasan N.
AU - Ghorban F.
AU - Velten J.
AU - Kummert A.
PY - 2022
SP - 55
EP - 64
DO - 10.5220/0010771700003124
PB - SciTePress