Official Implementation of KDD 2023 paper:
"Pyramid Graph Neural Network: A Graph Sampling and Filtering Approach for Multi-scale Disentangled Representations"
We propose Pyramid Graph Neural Network framework(PyGNN), a multi-scale framework for node representations on graphs.
More details are shown in this project.
* pytorch 1.7.0
* numpy 1.18.1
* torch-geometric 1.7.0
* torch-cluster 1.6.0
* torch-scatter 2.0.5
* torch-sparse 0.6.8
* torch-spline-conv 1.2.0
* scipy 1.6.2
(The currently used dataset may not support later torch-geometric versions.)
- set "TOP_DIR" as dataset directory in src/dataloader.py (provided in "./dataset/")
- Generate Pyramid subgraphs (provided)
python src/proc/Downsample.py -d ${dataname}
python src/proc/post_proc.py -d ${dataname} -s Downsample
- run:
bash node.sh
If you use our implementation in your works, we would appreciate citations to the paper:
@inproceedings{geng2023pyramid,
author = {Geng, Haoyu and Chen, Chao and He, Yixuan and Zeng, Gang and Han, Zhaobing and Chai, Hua and Yan, Junchi},
title = {Pyramid Graph Neural Network: A Graph Sampling and Filtering Approach for Multi-Scale Disentangled Representations},
year = {2023},
isbn = {9798400701030},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3580305.3599478},
doi = {10.1145/3580305.3599478},
booktitle = {Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining},
pages = {518–530},
numpages = {13},
keywords = {graph algorithms, spectral graph theory, graph neural networks},
location = {Long Beach, CA, USA},
series = {KDD '23}
}