Triangle Generative Adversarial Networks

TL;DR

The paper introduces $ riangle$-GAN, a novel semi-supervised framework with four neural networks for cross-domain joint distribution matching, effectively handling limited paired data for tasks like image translation and attribute-based generation.

Contribution

It proposes a new $ riangle$-GAN architecture with four networks for semi-supervised cross-domain matching, improving over existing methods in limited supervision scenarios.

Findings

Effective in semi-supervised image classification

Superior in image-to-image translation tasks

Advantageous for attribute-based image generation

Abstract

A Triangle Generative Adversarial Network ($\Delta$-GAN) is developed for semi-supervised cross-domain joint distribution matching, where the training data consists of samples from each domain, and supervision of domain correspondence is provided by only a few paired samples. $\Delta$-GAN consists of four neural networks, two generators and two discriminators. The generators are designed to learn the two-way conditional distributions between the two domains, while the discriminators implicitly define a ternary discriminative function, which is trained to distinguish real data pairs and two kinds of fake data pairs. The generators and discriminators are trained together using adversarial learning. Under mild assumptions, in theory the joint distributions characterized by the two generators concentrate to the data distribution. In experiments, three different kinds of domain pairs are considered, image-label, image-image and image-attribute pairs. Experiments on semi-supervised image classification, image-to-image translation and attribute-based image generation demonstrate the superiority of the proposed approach.