Code for Fast Approximation of the Sliced-Wasserstein Distance Using Concentration of Random Projections
This repository contains the implementation of the experiments presented in Fast Approximation of the Sliced-Wasserstein Distance Using Concentration of Random Projections (publication accepted at NeurIPS 2021).
- To reproduce synthetic experiments (Figures 1 and 2): in the directory 'synthetic_exp', run
python3 main.py - To reproduce results on image generation (Table 1, Figure 3): in the directory 'swg', run
./run_mnist.sh(for MNIST dataset) or./run_celeba.sh(for CelebA dataset)
If you use this code in a scientific publication, please cite the following paper.
@inproceedings{nadjahi2021fast,
author = {Nadjahi, Kimia and Durmus, Alain and Jacob, Pierre E and Badeau, Roland and Simsekli, Umut},
booktitle = {Advances in Neural Information Processing Systems},
editor = {M. Ranzato and A. Beygelzimer and Y. Dauphin and P.S. Liang and J. Wortman Vaughan},
pages = {12411--12424},
publisher = {Curran Associates, Inc.},
title = {Fast Approximation of the Sliced-Wasserstein Distance Using Concentration of Random Projections},
url = {https://proceedings.neurips.cc/paper/2021/file/6786f3c62fbf9021694f6e51cc07fe3c-Paper.pdf},
volume = {34},
year = {2021}
}