Semi-Supervised Learning with Context-Conditional Generative Adversarial Networks

TL;DR

This paper presents a semi-supervised image learning method using a context-conditional GAN that inpaints missing image regions, improving training efficiency and performance on benchmark datasets.

Contribution

It introduces a novel in-painting based semi-supervised learning approach with adversarial training for large neural networks.

Findings

Achieves comparable or superior results to existing methods on STL-10 and PASCAL datasets.

Enables direct semi-supervised training of large VGG-style networks.

Uses in-painting as a regularizer for standard supervised training.

Abstract

We introduce a simple semi-supervised learning approach for images based on in-painting using an adversarial loss. Images with random patches removed are presented to a generator whose task is to fill in the hole, based on the surrounding pixels. The in-painted images are then presented to a discriminator network that judges if they are real (unaltered training images) or not. This task acts as a regularizer for standard supervised training of the discriminator. Using our approach we are able to directly train large VGG-style networks in a semi-supervised fashion. We evaluate on STL-10 and PASCAL datasets, where our approach obtains performance comparable or superior to existing methods.