Class-Splitting Generative Adversarial Networks

TL;DR

This paper introduces a class-splitting approach for GANs that enhances sample quality by augmenting class labels through clustering in the learned representation space, applicable in both supervised and unsupervised settings.

Contribution

It proposes a novel class-splitting method that improves GAN performance by dynamically creating subclasses via clustering, applicable even without initial class labels.

Findings

Achieved state-of-the-art Inception scores on CIFAR-10 and STL-10 datasets.

Effective in both supervised and unsupervised GAN training.

Enhanced sample quality through class augmentation.

Abstract

Generative Adversarial Networks (GANs) produce systematically better quality samples when class label information is provided., i.e. in the conditional GAN setup. This is still observed for the recently proposed Wasserstein GAN formulation which stabilized adversarial training and allows considering high capacity network architectures such as ResNet. In this work we show how to boost conditional GAN by augmenting available class labels. The new classes come from clustering in the representation space learned by the same GAN model. The proposed strategy is also feasible when no class information is available, i.e. in the unsupervised setup. Our generated samples reach state-of-the-art Inception scores for CIFAR-10 and STL-10 datasets in both supervised and unsupervised setup.