Xiuyu-Li / LFR

Not All Low-Pass Filters are Robust in Graph Convolutional Networks

GCN-LFR

This repository is the official PyTorch implementation of "Not All Low-Pass Filters are Robust in Graph Convolutional Networks".

Heng Chang, Yu Rong, Tingyang Xu, Yatao Bian, Shiji Zhou, Xin Wang, Junzhou Huang, Wenwu Zhu, Not All Low-Pass Filters are Robust in Graph Convolutional Networks, NeurIPS 2021.

Requirements

The script has been tested running under Python 3.6.9, with the following packages installed (along with their dependencies):

• pytorch (tested on 1.7.1)
• torch_geometric (tested on 1.6.3)
• scipy (tested on 1.5.4)
• numpy (tested on 1.19.5)
• networkx (tested on 2.5.1)
• sklearn (tested on 0.24.2)
• deeprobust (tested on 0.1.1)

Datasets

The datasets are from PyG, which can be referred to the docs.

Run

• For the defense experiment on Cora dataset, one-edge targeted attack under Nettack (default setting):

python LFR_test.py

Acknowledgement

Part of this implementation is modified from DeepRobust, and we sincerely thank them for their contributions.

Reference

• If you find GCN-LFR useful in your research, please cite the following in your manuscript:

@article{chang2021not,
  title={Not All Low-Pass Filters are Robust in Graph Convolutional Networks},
  author={Chang, Heng and Rong, Yu and Xu, Tingyang and Bian, Yatao and Zhou, Shiji and Wang, Xin and Huang, Junzhou and Zhu, Wenwu},
  journal={Advances in Neural Information Processing Systems},
  volume={34},
  year={2021}
}