Generative Multi-Adversarial Networks

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.

Accepted as a conference paper (poster) at ICLR 2017

Cited by in corpus (17)

• MD-GAN: Multi-Discriminator Generative Adversarial Networks for Distributed Datasets
• Multi-Adversarial Domain Adaptation
• Decentralized Learning of Generative Adversarial Networks from Non-iid Data
• CrystalGAN: Learning to Discover Crystallographic Structures with Generative Adversarial Networks
• Reusing Discriminators for Encoding: Towards Unsupervised Image-to-Image Translation
• GDPP: Learning Diverse Generations Using Determinantal Point Process
• LaSO: Label-Set Operations networks for multi-label few-shot learning
• CDE-GAN: Cooperative Dual Evolution Based Generative Adversarial Network
• Generative Zero-shot Network Quantization
• Generative Adversarial Network with Multi-Branch Discriminator for Cross-Species Image-to-Image Translation
• Robust GANs against Dishonest Adversaries
• DO-GAN: A Double Oracle Framework for Generative Adversarial Networks
• Generative Adversarial Networks and Adversarial Autoencoders: Tutorial and Survey
• Independent Encoder for Deep Hierarchical Unsupervised Image-to-Image Translation
• IID-GAN: an IID Sampling Perspective for Regularizing Mode Collapse
• Coulomb Autoencoders
• Conditional Transferring Features: Scaling GANs to Thousands of Classes with 30% Less High-quality Data for Training