MR-GAN: Manifold Regularized Generative Adversarial Networks

TL;DR

This paper introduces a manifold regularizer for GANs that leverages data geometry to improve training stability, reduce mode collapse, and enhance data quality, supported by theoretical proofs and preliminary experiments.

Contribution

It proposes a novel manifold regularization technique for GANs that improves stability and performance, with theoretical validation and empirical evidence.

Findings

Reduces mode collapse during training

Enhances stability of GAN training process

Improves data quality and generalization

Abstract

Despite the growing interest in generative adversarial networks (GANs), training GANs remains a challenging problem, both from a theoretical and a practical standpoint. To address this challenge, in this paper, we propose a novel way to exploit the unique geometry of the real data, especially the manifold information. More specifically, we design a method to regularize GAN training by adding an additional regularization term referred to as manifold regularizer. The manifold regularizer forces the generator to respect the unique geometry of the real data manifold and generate high quality data. Furthermore, we theoretically prove that the addition of this regularization term in any class of GANs including DCGAN and Wasserstein GAN leads to improved performance in terms of generalization, existence of equilibrium, and stability. Preliminary experiments show that the proposed manifold regularization helps in avoiding mode collapse and leads to stable training.