RustGraph: Robust Anomaly Detection in Dynamic Graphs by Jointly Learning Structural-Temporal Dependency

The PyTorch implementation of the IEEE TKDE paper "RustGraph: Robust Anomaly Detection in Dynamic Graphs by Jointly Learning Structural-Temporal Dependency".

Link to this paper: https://doi.ieeecomputersociety.org/10.1109/TKDE.2023.3328645

Requirements

python==3.9
pytorch==1.12.1
pytorch-geometric==2.3.0
tensorboard==2.6.0
networkx==3.1
matplotlib==3.7.1

Reproducibility

To reproduce the main results in the paper (Section 5.2), execute bash run.sh $DATASET

To reproduce the results of noisy labels (Section 5.4), execute bash exp_noise_ratio.sh $DATASET

To reproduce the results of sensitivity analysis (Section 5.5), execute bash exp_emb_dim.sh $DATASET, bash exp_train_ratio.sh $DATASET, bash exp_hyperparam.sh $DATASET

Citation

If you find this work interesting, please cite RustGraph using the following Bibtext:

@ARTICLE {rustgraph,
author = {J. Guo and S. Tang and J. Li and K. Pan and L. Wu},
journal = {IEEE Transactions on Knowledge & Data Engineering},
title = {RustGraph: Robust Anomaly Detection in Dynamic Graphs by Jointly Learning Structural-Temporal Dependency},
year = {5555},
volume = {},
number = {01},
issn = {1558-2191},
pages = {1-14},
keywords = {noise measurement;anomaly detection;image edge detection;task analysis;training;representation learning;data models},
doi = {10.1109/TKDE.2023.3328645},
publisher = {IEEE Computer Society},
address = {Los Alamitos, CA, USA},
month = {oct}
}

Contact

Please feel free to contact me through [email protected] if you have any problems:)