A Wasserstein GAN model with the total variational regularization

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.

References in corpus (8)

• Conditional Generative Adversarial Nets
• Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks
• Semi-Supervised Learning with Deep Generative Models
• Improved Training of Wasserstein GANs
• Improved Techniques for Training GANs
• BEGAN: Boundary Equilibrium Generative Adversarial Networks
• Banach Wasserstein GAN
• Primal-Dual Wasserstein GAN

Cited by in corpus (2)

• A Systematic Survey of Regularization and Normalization in GANs
• From EMBER to FIRE: predicting high resolution baryon fields from dark matter simulations with Deep Learning