Improving Generative Adversarial Networks with Denoising Feature Matching

David Warde-Farley, Yoshua Bengio

Abstract: We propose an augmented training procedure for generative adversarial networks designed to address shortcomings of the original by directing the generator towards probable configurations of abstract discriminator features. We estimate and track the distribution of these features, as computed from data, with a denoising auto-encoder, and use it to propose high-level targets for the generator. We combine this new loss with the original and evaluate the hybrid criterion on the task of unsupervised image synthesis from datasets comprising a diverse set of visual categories, noting a qualitative and quantitative improvement in the ``objectness'' of the resulting samples.

TL;DR: Use a denoiser trained on discriminator features to train better generators.

Conflicts: umontreal.ca, iro.umontreal.ca, polymtl.ca, google.com

Keywords: Deep learning, Unsupervised Learning