A Wasserstein GAN model with the total variational regularization

TL;DR

This paper introduces a novel Wasserstein GAN variant that incorporates total variational regularization, enhancing training stability and allowing control over image diversity and quality without extra computational cost.

Contribution

It proposes replacing gradient penalty with total variational regularization in WGANs, improving stability and flexibility across architectures.

Findings

More stable training compared to GP-WGANs

Effective across different GAN architectures

No additional computational burden

Abstract

It is well known that the generative adversarial nets (GANs) are remarkably difficult to train. The recently proposed Wasserstein GAN (WGAN) creates principled research directions towards addressing these issues. But we found in practice that gradient penalty WGANs (GP-WGANs) still suffer from training instability. In this paper, we combine a Total Variational (TV) regularizing term into the WGAN formulation instead of weight clipping or gradient penalty, which implies that the Lipschitz constraint is enforced on the critic network. Our proposed method is more stable at training than GP-WGANs and works well across varied GAN architectures. We also present a method to control the trade-off between image diversity and visual quality. It does not bring any computation burden.