/PC-GNN

(WWW 2021) Source code of PC-GNN

Primary LanguagePython

PC-GNN

This is the author implementation of "Pick and Choose: A GNN-based Imbalanced Learning Approach for Fraud Detection" (WebConf 2021).

Slides and video can be found here.

Yang Liu, Xiang Ao, Zidi Qin, Jianfeng Chi, Jinghua Feng, Hao Yang and Qing He.

Requirements

argparse          1.1.0
networkx          1.11
numpy             1.16.4
scikit_learn      0.21rc2
scipy             1.2.1
torch             1.4.0

Dataset

YelpChi and Amazon can be downloaded from here or dgl.data.FraudDataset.

Put them in /data directory and run unzip /data/Amazon.zip and unzip /data/YelpChi.zip to unzip the datasets.

Run python src/data_process.py to pre-process the data.

Kindly note that there may be two versions of node features for YelpChi. The old version has a dimension of 100 and the new version is 32. In our paper, the results are reported based on the old features.

Usage

python main.py --config ./config/pcgnn_yelpchi.yml

Citation

@inproceedings{liu2021pick,
  title={Pick and Choose: A GNN-based Imbalanced Learning Approach for Fraud Detection},
  author={Liu, Yang and Ao, Xiang and Qin, Zidi and Chi, Jianfeng and Feng, Jinghua and Yang, Hao and He, Qing},
  booktitle={Proceedings of the Web Conference 2021},
  pages={3168--3177},
  year={2021}
}

Other implementations

Thanks for Jack Huang and Ronald D. R. Pereira for their kind implementations.