WGAN with an Infinitely Wide Generator Has No Spurious Stationary Points
Abstract
Generative adversarial networks (GAN) are a widely used class of deep generative models, but their minimax training dynamics are not understood very well. In this work, we show that GANs with a 2-layer infinite-width generator and a 2-layer finite-width discriminator trained with stochastic gradient ascent-descent have no spurious stationary points. We then show that when the width of the generator is finite but wide, there are no spurious stationary points within a ball whose radius becomes arbitrarily large (to cover the entire parameter space) as the width goes to infinity.
Cite
Text
No et al. "WGAN with an Infinitely Wide Generator Has No Spurious Stationary Points." International Conference on Machine Learning, 2021.Markdown
[No et al. "WGAN with an Infinitely Wide Generator Has No Spurious Stationary Points." International Conference on Machine Learning, 2021.](https://mlanthology.org/icml/2021/no2021icml-wgan/)BibTeX
@inproceedings{no2021icml-wgan,
title = {{WGAN with an Infinitely Wide Generator Has No Spurious Stationary Points}},
author = {No, Albert and Yoon, Taeho and Sehyun, Kwon and Ryu, Ernest K},
booktitle = {International Conference on Machine Learning},
year = {2021},
pages = {8205-8215},
volume = {139},
url = {https://mlanthology.org/icml/2021/no2021icml-wgan/}
}