Investigating Under and Overfitting in Wasserstein Generative Adversarial Networks

Amol Kapoor
Ben Adlam
Charles Weill
ICML Understanding and Improving Generalization in Deep Learning Workshop (2019)

Abstract

We investigate under and overfitting in Generative Adversarial Networks (GANs), using discriminators unseen by the generator to measure generalization. We find that the model capacity of the discriminator has a significant effect on the generator's model quality, and that the generator's poor performance coincides with the discriminator underfitting. Contrary to our expectations, we find that generators with large model capacities relative to the discriminator do not show evidence of overfitting on CIFAR10, CIFAR100, and CelebA.

Research Areas