jxh4945777 / gsdn

The source code of the paper "Understanding Graph Neural Networks from Graph Signal Denoising Perspectives"

GSDN-F and GSDN-EF

This repository provides a reference implementation of GSDN-F and GSDN-EF as described in the paper "Understanding Graph Neural Networks from Graph Signal Denoising Perspectives".

Requirements

Install the following packages:

• pytorch
• torch_geometric
• scikit-learn
• networkx

Basic Usage

$ python gsdnf.py --input cora --alpha 0.6 --K 4 --epochs 200 --lr 0.02 --hidden_num 16
$ python gsdnef.py --input cora --alpha 0.6 --K 4 --epochs 200 --lr 0.02 --hidden_num 16

noted: your can just checkout gsdnf.py and gsdnef.py to get what you want.

Citing

If you find GSDN-F and/or GSDN-EF useful in your research, please cite our paper:

@misc{2006.04386,
 Author = {Guoji Fu and Yifan Hou and Jian Zhang and Kaili Ma and Barakeel Fanseu Kamhoua and James Cheng},
 Title = {Understanding Graph Neural Networks from Graph Signal Denoising Perspectives},
 Year = {2020}
}