Deep Co-Training for Semi-Supervised Image Recognition

TL;DR

This paper introduces Deep Co-Training, a semi-supervised learning approach that trains multiple neural networks with diverse views using adversarial examples, significantly improving image recognition performance.

Contribution

It extends the Co-Training framework to deep learning by training multiple networks with enforced view differences, leading to superior semi-supervised image classification results.

Findings

Outperforms previous state-of-the-art methods on SVHN, CIFAR-10/100, and ImageNet.

Uses adversarial examples to maintain diversity among networks.

Demonstrates large margin improvements in semi-supervised image recognition.

Abstract

In this paper, we study the problem of semi-supervised image recognition, which is to learn classifiers using both labeled and unlabeled images. We present Deep Co-Training, a deep learning based method inspired by the Co-Training framework. The original Co-Training learns two classifiers on two views which are data from different sources that describe the same instances. To extend this concept to deep learning, Deep Co-Training trains multiple deep neural networks to be the different views and exploits adversarial examples to encourage view difference, in order to prevent the networks from collapsing into each other. As a result, the co-trained networks provide different and complementary information about the data, which is necessary for the Co-Training framework to achieve good results. We test our method on SVHN, CIFAR-10/100 and ImageNet datasets, and our method outperforms the previous state-of-the-art methods by a large margin.