Wasserstein Proximal of GANs
Machine Learning
2021-02-16 v1 Artificial Intelligence
Numerical Analysis
Numerical Analysis
Abstract
We introduce a new method for training generative adversarial networks by applying the Wasserstein-2 metric proximal on the generators. The approach is based on Wasserstein information geometry. It defines a parametrization invariant natural gradient by pulling back optimal transport structures from probability space to parameter space. We obtain easy-to-implement iterative regularizers for the parameter updates of implicit deep generative models. Our experiments demonstrate that this method improves the speed and stability of training in terms of wall-clock time and Fr\'echet Inception Distance.
Cite
@article{arxiv.2102.06862,
title = {Wasserstein Proximal of GANs},
author = {Alex Tong Lin and Wuchen Li and Stanley Osher and Guido Montufar},
journal= {arXiv preprint arXiv:2102.06862},
year = {2021}
}