Reference implementation (example) of the model proposed in the paper:
Adaptive Diffusion in Graph Neural Networks
by Jialin Zhao, Yuxiao Dong, Ming Ding, Evgeny Kharlamov, Jie Tang
Published at NeurIPS 2021.
This repository primarily contains a demonstration of enhancing a graph convolutional network (GCN) with graph diffusion convolution (GDC) in the notebook gdc_demo.ipynb.
The repository uses these packages:
To run ADC:
python ADC.py --config config_Cora_GCN.yaml --t 5
python ADC.py --config config_Cora_JK.yaml --t 1
python ADC.py --config config_Cora_ARMA.yaml --t 5
To run GADC:
python GADC.py --config config_Cora_GCN.yaml
To run GDC:
python GDC.py --config config_Cora_GCN.yaml --preprocessing heat
To run original model (GCN, JKNet, ARMA):
python GDC.py --config config_Cora_GCN.yaml --preprocessing none
Please contact zjl19970607@163.com in case you have any questions.
Please cite our paper if you use the model or this code in your own work:
@inproceedings{NEURIPS2021_c42af2fa,
author = {Zhao, Jialin and Dong, Yuxiao and Ding, Ming and Kharlamov, Evgeny and Tang, Jie},
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 = {23321--23333},
publisher = {Curran Associates, Inc.},
title = {Adaptive Diffusion in Graph Neural Networks},
url = {https://proceedings.neurips.cc/paper_files/paper/2021/file/c42af2fa7356818e0389593714f59b52-Paper.pdf},
volume = {34},
year = {2021}
}