Improving Generalization and Stability of Generative Adversarial Networks

TL;DR

This paper analyzes the generalization issues of GANs, revealing poor discriminator performance on discrete datasets, and proposes a zero-centered gradient penalty to enhance generalization and convergence.

Contribution

It introduces a zero-centered gradient penalty method that improves GAN discriminator generalization and guarantees convergence, supported by theoretical analysis and experiments.

Findings

Discriminators trained on discrete datasets have poor generalization.

The proposed gradient penalty improves discriminator generalization.

Experiments verify the theoretical guarantees on synthetic and large datasets.

Abstract

Generative Adversarial Networks (GANs) are one of the most popular tools for learning complex high dimensional distributions. However, generalization properties of GANs have not been well understood. In this paper, we analyze the generalization of GANs in practical settings. We show that discriminators trained on discrete datasets with the original GAN loss have poor generalization capability and do not approximate the theoretically optimal discriminator. We propose a zero-centered gradient penalty for improving the generalization of the discriminator by pushing it toward the optimal discriminator. The penalty guarantees the generalization and convergence of GANs. Experiments on synthetic and large scale datasets verify our theoretical analysis.