CensNet

Code release for "Co-embedding of Nodes and Edges with Graph Neural Networks" (IEEE PAMI 2020) and "CensNet: Convolution with Edge-Node Switching in Graph Neural Networks" (IJCAI 2019)

Prerequisites

• Python3
• Pytorch == 1.0.0 (with suitable CUDA and CuDNN version)
• Numpy
• argparse
• tqdm

Dataset

The datasets (Cora, Citeseer and PubMed) are in GoogleDrive and BaiduPan (pw:frvg).
You need to move the dataset file into CensNet file.

Training

You can run python train.py to train and evaluate.

Citation

If you use this code for your research, please consider citing:

@article{jiang2020co,
  title={Co-embedding of Nodes and Edges with Graph Neural Networks},
  author={Jiang, Xiaodong and Zhu, Ronghang and Ji, Pengsheng and Li, Sheng},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year={2020},
  publisher={IEEE}
}  
@inproceedings{jiang2019censnet,
  title={CensNet: Convolution with Edge-Node Switching in Graph Neural Networks.},
  author={Jiang, Xiaodong and Ji, Pengsheng and Li, Sheng},
  booktitle={IJCAI},
  pages={2656--2662},
  year={2019}
}