Coupled Generative Adversarial Networks

TL;DR

This paper introduces CoGAN, a novel generative adversarial network that learns joint distributions of multi-domain images without needing paired samples, enabling applications like domain adaptation and image transformation.

Contribution

CoGAN is the first model to learn joint distributions of multi-domain images using weight-sharing constraints without paired data, advancing multi-domain image generation.

Findings

Successfully learns joint distributions of color and depth images

Learns joint distributions of face images with different attributes

Enables domain adaptation and image transformation

Abstract

We propose coupled generative adversarial network (CoGAN) for learning a joint distribution of multi-domain images. In contrast to the existing approaches, which require tuples of corresponding images in different domains in the training set, CoGAN can learn a joint distribution without any tuple of corresponding images. It can learn a joint distribution with just samples drawn from the marginal distributions. This is achieved by enforcing a weight-sharing constraint that limits the network capacity and favors a joint distribution solution over a product of marginal distributions one. We apply CoGAN to several joint distribution learning tasks, including learning a joint distribution of color and depth images, and learning a joint distribution of face images with different attributes. For each task it successfully learns the joint distribution without any tuple of corresponding images. We also demonstrate its applications to domain adaptation and image transformation.