Stabilizing Generative Adversarial Networks: A Survey
It addresses a critical issue for researchers and practitioners in machine learning by synthesizing existing solutions, but it is incremental as it reviews rather than proposes new methods.
This survey tackles the problem of training instability in Generative Adversarial Networks (GANs), such as non-convergence and mode collapse, by providing a comprehensive overview of stabilization methods from the literature, including a comparative summary and discussion of open problems.
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.