Linear Discriminant Generative Adversarial Networks

TL;DR

This paper introduces LD-GAN, a new GAN training method using linear discriminant analysis to improve stability and performance in unsupervised and class-conditional image generation.

Contribution

The paper proposes LD-GAN, which uses LDA-based decision hyper-planes for discriminator training, enhancing stability and generalization without normalization or weight constraints.

Findings

Stable training achieved by calibrating update frequencies.

Improved generalization over WGAN with auxiliary classifier.

Effective for both unsupervised and class-conditional generation.

Abstract

We develop a novel method for training of GANs for unsupervised and class conditional generation of images, called Linear Discriminant GAN (LD-GAN). The discriminator of an LD-GAN is trained to maximize the linear separability between distributions of hidden representations of generated and targeted samples, while the generator is updated based on the decision hyper-planes computed by performing LDA over the hidden representations. LD-GAN provides a concrete metric of separation capacity for the discriminator, and we experimentally show that it is possible to stabilize the training of LD-GAN simply by calibrating the update frequencies between generators and discriminators in the unsupervised case, without employment of normalization methods and constraints on weights. In the class conditional generation tasks, the proposed method shows improved training stability together with better generalization performance compared to WGAN that employs an auxiliary classifier.