Approximation and Convergence Properties of Generative Adversarial Learning

TL;DR

This paper investigates the theoretical approximation and convergence properties of GANs, focusing on how discriminator restrictions and objective functions influence their ability to approximate target distributions.

Contribution

It introduces a unified framework of adversarial divergences, analyzing how discriminator restrictions affect approximation and establishing conditions for convergence to the target distribution.

Findings

Restricted discriminators induce moment-matching.

Convergence in adversarial divergence implies weak convergence.

Generalizes previous convergence results.

Abstract

Generative adversarial networks (GAN) approximate a target data distribution by jointly optimizing an objective function through a "two-player game" between a generator and a discriminator. Despite their empirical success, however, two very basic questions on how well they can approximate the target distribution remain unanswered. First, it is not known how restricting the discriminator family affects the approximation quality. Second, while a number of different objective functions have been proposed, we do not understand when convergence to the global minima of the objective function leads to convergence to the target distribution under various notions of distributional convergence. In this paper, we address these questions in a broad and unified setting by defining a notion of adversarial divergences that includes a number of recently proposed objective functions. We show that if the objective function is an adversarial divergence with some additional conditions, then using a restricted discriminator family has a moment-matching effect. Additionally, we show that for objective functions that are strict adversarial divergences, convergence in the objective function implies weak convergence, thus generalizing previous results.