LOGAN: Latent Optimisation for Generative Adversarial Networks

Yan Wu, Jeff Donahue, David Balduzzi, Karen Simonyan, Timothy Lillicrap

TL;DR: Latent optimisation improves adversarial training dynamics. We present both theoretical analysis and state-of-the-art image generation with ImageNet 128x128.

Abstract: Training generative adversarial networks requires balancing of delicate adversarial dynamics. Even with careful tuning, training may diverge or end up in a bad equilibrium with dropped modes. In this work, we introduce a new form of latent optimisation inspired by the CS-GAN and show that it improves adversarial dynamics by enhancing interactions between the discriminator and the generator. We develop supporting theoretical analysis from the perspectives of differentiable games and stochastic approximation. Our experiments demonstrate that latent optimisation can significantly improve GAN training, obtaining state-of-the-art performance for the ImageNet (128 x 128) dataset. Our model achieves an Inception Score (IS) of 148 and an Frechet Inception Distance (FID) of 3.4, an improvement of 17% and 32% in IS and FID respectively, compared with the baseline BigGAN-deep model with the same architecture and number of parameters.

Keywords: GAN, adversarial training, generative model, game theory