Hyperbolic Neural Networks in Node-Level Graph Anomaly Detection

This repository provides official implementation of the model from the following paper.

Three Revisits to Node-Level Graph Anomaly Detection: Outliers, Message Passing and Hyperbolic Neural Networks

Jing Gu and Dongmian Zou, Duke Kunshan University, 2023

OpenReview: https://openreview.net/forum?id=fNsU9gi1Fy&noteId=fNsU9gi1Fy

Training

In the folder HNN_GAD, can specify parameters in config.py and run the model via

python run.py

Environment

• torch==1.9.1+cu111
• torch_sparse==0.6.12
• torch_scatter==2.0.9
• torch_geometric==2.1.0
• python==3.7.13
• scikit-learn
• networkx
• ogb
• geoopt
• jupyter
• nb_conda_kernels

For more specific information, please see environment.yml.

Citation

Gu, Jing, and Dongmian Zou. "Three Revisits to Node-Level Graph Anomaly Detection: Outliers, Message Passing and Hyperbolic Neural Networks." Proceedings of the Second Learning on Graphs Conference (LoG 2023), PMLR 231, Virtual Event, November 27–30, 2023.

or

@inproceedings{gu2023three,
  title={Three Revisits to Node-Level Graph Anomaly Detection: Outliers, Message Passing and Hyperbolic Neural Networks},
  author={Gu, Jing and Zou, Dongmian},
  booktitle={The Second Learning on Graphs Conference},
  year={2023}
}

Reference

For construction of hyperbolic models, we utilized code available at https://github.com/HazyResearch/hgcn.