Boosting Graph Pooling with Persistent Homology

This is the repo for our paper: Boosting Graph Pooling with Persistent Homology presented at NeurlPS 2024.

Setup

Our code can run smoothly on Ubuntu 20.04. The basic environment settings are as follows:

Python=3.7
PyTorch=1.12.1+cu113
torch_geometric=2.3.1
gudhi=3.8.0
GraphRicciCurvature=0.5.3.1
ogb=1.2.1
networkx=2.6.3

Then setup the environment using the following command:

python setup.py build_ext --inplace

How to run

python main_diffpool_tip.py --device=cuda:0

Citation

@inproceedings{
ying2024boosting,
title={Boosting Graph Pooling with Persistent Homology},
author={Chaolong Ying and Xinjian Zhao and Tianshu Yu},
booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems},
year={2024},
url={https://openreview.net/forum?id=WcmqdY2AKu}
}