GANs beyond divergence minimization

TL;DR

This paper challenges the common view that GANs learn by minimizing divergences, showing instead that they can use a broader range of loss functions, many of which are not divergences, yet still produce high-quality data.

Contribution

The paper introduces a new framework separating the discriminator and generator objectives, and explores diverse loss functions, expanding the understanding of GAN training beyond divergence minimization.

Findings

Most loss functions tested converge well and produce comparable quality data.

Traditional GAN loss functions are not divergences and do not share the same equilibrium.

GANs can effectively use a wide range of loss functions, not limited to divergences.

Abstract

Generative adversarial networks (GANs) can be interpreted as an adversarial game between two players, a discriminator D and a generator G, in which D learns to classify real from fake data and G learns to generate realistic data by "fooling" D into thinking that fake data is actually real data. Currently, a dominating view is that G actually learns by minimizing a divergence given that the general objective function is a divergence when D is optimal. However, this view has been challenged due to inconsistencies between theory and practice. In this paper, we discuss of the properties associated with most loss functions for G (e.g., saturating/non-saturating f-GAN, LSGAN, WGAN, etc.). We show that these loss functions are not divergences and do not have the same equilibrium as expected of divergences. This suggests that G does not need to minimize the same objective function as D maximize, nor maximize the objective of D after swapping real data with fake data (non-saturating GAN) but can instead use a wide range of possible loss functions to learn to generate realistic data. We define GANs through two separate and independent D maximization and G minimization steps. We generalize the generator step to four new classes of loss functions, most of which are actual divergences (while traditional G loss functions are not). We test a wide variety of loss functions from these four classes on a synthetic dataset and on CIFAR-10. We observe that most loss functions converge well and provide comparable data generation quality to non-saturating GAN, LSGAN, and WGAN-GP generator loss functions, whether we use divergences or non-divergences. These results suggest that GANs do not conform well to the divergence minimization theory and form a much broader range of models than previously assumed.