Topological Clustering

About

A code repository written in Python for topological clustering presented in the ICLR 2022 paper:

• Fast Topological Clustering with Wasserstein Distance

This paper proposes a novel and computationally practical topological clustering method that clusters complex networks with intricate topology using principled theory from persistent homology and optimal transport.

Quick Start

1. Prerequisite: install scikit-learn
2. Execute top_clustering.py script for a demo of the topological clustering method

Citation

Please consider citing our paper if you use this code in your research:

@inproceedings{songdechakraiwut2022fast,
    title={Fast topological clustering with {W}asserstein distance},
    author={Tananun Songdechakraiwut and Bryan M Krause and Matthew I Banks and Kirill V Nourski and Barry D Van Veen},
    booktitle={International Conference on Learning Representations},
    year={2022},
    url={https://openreview.net/forum?id=0kPL3xO4R5}
}

Contact

If you have any questions, please feel free to contact Tananun Songdechakraiwut (songdechakra@wisc.edu).