Wasserstein-2 Generative Networks

Alexander Korotin, Vage Egiazarian, Arip Asadulaev, Alexander Safin, Evgeny Burnaev
9/28/2019
cs.LGcs.CVstat.ML

Official code: iamalexkorotin/Wasserstein2GenerativeNetworks (67 stars, MIT, PyTorch, last commit 2023-02-03).

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.

View on arXivView PDF

Code Implementations(1)

iamalexkorotin/Wasserstein2GenerativeNetworksOfficial100%
PyTorchMIT

PyTorch implementation of "Wasserstein-2 Generative Networks" (ICLR 2021)

675Python, Jupyter NotebookJun 22, 20203 years agoMIT
deep-learninggangenerative-adversarial-networkoptimal-transportpytorch+1 more

Similar Papers

Do Neural Optimal Transport Solvers Work? A Continuous Wasserstein-2 Benchmark

90% match2 repos672021

OT-Net: A Reusable Neural Optimal Transport Solver

83% match2023

Improving Neural Optimal Transport via Displacement Interpolation

83% match2024

Neural Optimal Transport

83% match1 repo2392022

Sparse Partial Optimal Transport via Quadratic Regularization

82% match2025

ENOT: Expectile Regularization for Fast and Accurate Training of Neural Optimal Transport

82% match2024

Cite this paper

@article{korotin2019wasserstein,
  title  = {Wasserstein-2 Generative Networks},
  author = {Alexander Korotin and Vage Egiazarian and Arip Asadulaev and Alexander Safin and Evgeny Burnaev},
  year   = {2019},
  eprint = {1909.13082},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1909.13082}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1909.13082.svg)](https://codeofpaper.com/paper/1909.13082)

Discussion