This project implements the paper "Heterophily Learning and Global-local Dependencies Enhanced Multi-view Representation Learning for Graph Anomaly Detection" submitted to Knowledge-Based Systems.
This project is tested on cuda 11.6 with several dependencies listed below:
pytorch=1.11.0
torch-geometric=2.0.4Public datasets weibo and toloker used for graph anomaly detection are available for evaluation.
python benchamrk.py --datasets 0/1