This is the demo code for the Fixed-Support Tree-Sliced Wasserstein Barycenter.
- Python (3.7.4)
- POT (0.7.0)
- scipy (1.3.1)
- numpy (1.19.5)
- treelib (1.6.1)
- sklearn (0.21.3)
- pytorch (1.7.0)
- mat4py (0.4.3)
- gensim (3.8.0)
- nltk (3.4.5)
- tqdm
The sample code is contained in examples.ipynb.
You can reproduce the experimental results by running the following commands.
mkdir exp
bash mnist_time_support.sh
bash mnist_time_sample.sh
bash mnist_memory_support.sh
bash mnist_memory_sample.sh
mnist_memory_support.sh and mnist_memory_sample.sh is only available on Linux.
@inproceedings{takezawa2022fixed,
title = {Fixed Support Tree-Sliced Wasserstein Barycenter},
author = {Yuki Takezawa and Ryoma Sato and Zornitsa Kozareva and Sujith Ravi and Makoto Yamada},
booktitle = {International Conference on Artificial Intelligence and Statistics},
year = {2022}
}