Wasserstein-2 Generative Networks
Abstract
We propose a novel end-to-end non-minimax algorithm for training optimal transport mappings for the quadratic cost (Wasserstein-2 distance). The algorithm uses input convex neural networks and a cycle-consistency regularization to approximate Wasserstein-2 distance. In contrast to popular entropic and quadratic regularizers, cycle-consistency does not introduce bias and scales well to high dimensions. From the theoretical side, we estimate the properties of the generative mapping fitted by our algorithm. From the practical side, we evaluate our algorithm on a wide range of tasks: image-to-image color transfer, latent space optimal transport, image-to-image style transfer, and domain adaptation.
Cite
Text
Korotin et al. "Wasserstein-2 Generative Networks." International Conference on Learning Representations, 2021.Markdown
[Korotin et al. "Wasserstein-2 Generative Networks." International Conference on Learning Representations, 2021.](https://mlanthology.org/iclr/2021/korotin2021iclr-wasserstein2/)BibTeX
@inproceedings{korotin2021iclr-wasserstein2,
title = {{Wasserstein-2 Generative Networks}},
author = {Korotin, Alexander and Egiazarian, Vage and Asadulaev, Arip and Safin, Alexander and Burnaev, Evgeny},
booktitle = {International Conference on Learning Representations},
year = {2021},
url = {https://mlanthology.org/iclr/2021/korotin2021iclr-wasserstein2/}
}