SGD Learns One-Layer Networks in WGANs
Machine Learning
2020-07-03 v2 Machine Learning
Abstract
Generative adversarial networks (GANs) are a widely used framework for learning generative models. Wasserstein GANs (WGANs), one of the most successful variants of GANs, require solving a minmax optimization problem to global optimality, but are in practice successfully trained using stochastic gradient descent-ascent. In this paper, we show that, when the generator is a one-layer network, stochastic gradient descent-ascent converges to a global solution with polynomial time and sample complexity.
Cite
@article{arxiv.1910.07030,
title = {SGD Learns One-Layer Networks in WGANs},
author = {Qi Lei and Jason D. Lee and Alexandros G. Dimakis and Constantinos Daskalakis},
journal= {arXiv preprint arXiv:1910.07030},
year = {2020}
}
Comments
24 pages, 4 figures, ICML2020