Generative Multi-Adversarial Networks

TL;DR

GMAN extends GANs by incorporating multiple discriminators, enabling more reliable training with the original objective and producing higher quality images faster than standard GANs.

Contribution

Introduces GMAN, a multi-discriminator framework for GANs that maintains the original objective and improves training stability and image quality.

Findings

GMAN produces higher quality samples than standard GANs.

GMAN achieves faster convergence in image generation tasks.

Training GMAN is reliable with the original GAN objective.

Abstract

Generative adversarial networks (GANs) are a framework for producing a generative model by way of a two-player minimax game. In this paper, we propose the \emph{Generative Multi-Adversarial Network} (GMAN), a framework that extends GANs to multiple discriminators. In previous work, the successful training of GANs requires modifying the minimax objective to accelerate training early on. In contrast, GMAN can be reliably trained with the original, untampered objective. We explore a number of design perspectives with the discriminator role ranging from formidable adversary to forgiving teacher. Image generation tasks comparing the proposed framework to standard GANs demonstrate GMAN produces higher quality samples in a fraction of the iterations when measured by a pairwise GAM-type metric.