Linear Discriminant Generative Adversarial Networks
This addresses training instability in GANs for researchers and practitioners in unsupervised and class-conditional image generation, though it is incremental as it builds on existing GAN frameworks.
The authors tackled the problem of training instability in GANs for image generation by proposing LD-GAN, which uses linear discriminant analysis in the discriminator to maximize separability between real and generated samples, resulting in improved stability without normalization methods and better generalization in class-conditional tasks compared to WGAN.
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.