Potential Flow Generator with $L_2$ Optimal Transport Regularity for Generative Models
Machine Learning
2019-09-02 v1 Computer Vision and Pattern Recognition
Machine Learning
Abstract
We propose a potential flow generator with optimal transport regularity, which can be easily integrated into a wide range of generative models including different versions of GANs and flow-based models. We show the correctness and robustness of the potential flow generator in several 2D problems, and illustrate the concept of "proximity" due to the optimal transport regularity. Subsequently, we demonstrate the effectiveness of the potential flow generator in image translation tasks with unpaired training data from the MNIST dataset and the CelebA dataset.
Cite
@article{arxiv.1908.11462,
title = {Potential Flow Generator with $L_2$ Optimal Transport Regularity for Generative Models},
author = {Liu Yang and George Em Karniadakis},
journal= {arXiv preprint arXiv:1908.11462},
year = {2019}
}