Stabilizing Generative Adversarial Networks: A Survey

TL;DR

This survey reviews various methods developed to stabilize the training of Generative Adversarial Networks, addressing issues like non-convergence and mode collapse, and compares their effectiveness and limitations.

Contribution

It provides a comprehensive overview and comparison of existing stabilization techniques for GAN training, highlighting open challenges.

Findings

Different stabilization methods have varying effectiveness and trade-offs.

No single approach completely solves GAN training instability.

Open problems remain in achieving robust and scalable GAN training.

Abstract

Generative Adversarial Networks (GANs) are a type of generative model which have received much attention due to their ability to model complex real-world data. Despite their recent successes, the process of training GANs remains challenging, suffering from instability problems such as non-convergence, vanishing or exploding gradients, and mode collapse. In recent years, a diverse set of approaches have been proposed which focus on stabilizing the GAN training procedure. The purpose of this survey is to provide a comprehensive overview of the GAN training stabilization methods which can be found in the literature. We discuss the advantages and disadvantages of each approach, offer a comparative summary, and conclude with a discussion of open problems.