Gradient-based Data Augmentation for Semi-Supervised Learning

TL;DR

This paper introduces Gradient-based Data Augmentation (GDA) and MixGDA, combining these with mixup techniques to enhance semi-supervised learning, achieving state-of-the-art results on multiple datasets.

Contribution

It proposes a novel gradient-based data augmentation method and its integration with mixup techniques for improved semi-supervised learning performance.

Findings

MixGDA achieves state-of-the-art results on CIFAR-10, SVHN, and CIFAR-100 datasets.

GDA enhances data diversity, improving model discrimination in SSL.

MixGDA matches or surpasses previous best performances across tested datasets.

Abstract

In semi-supervised learning (SSL), a technique called consistency regularization (CR) achieves high performance. It has been proved that the diversity of data used in CR is extremely important to obtain a model with high discrimination performance by CR. We propose a new data augmentation (Gradient-based Data Augmentation (GDA)) that is deterministically calculated from the image pixel value gradient of the posterior probability distribution that is the model output. We aim to secure effective data diversity for CR by utilizing three types of GDA. On the other hand, it has been demonstrated that the mixup method for labeled data and unlabeled data is also effective in SSL. We propose an SSL method named MixGDA by combining various mixup methods and GDA. The discrimination performance achieved by MixGDA is evaluated against the 13-layer CNN that is used as standard in SSL research. As a result, for CIFAR-10 (4000 labels), MixGDA achieves the same level of performance as the best performance ever achieved. For SVHN (250 labels, 500 labels and 1000 labels) and CIFAR-100 (10000 labels), MixGDA achieves state-of-the-art performance.