Twin Graph-based Anomaly Detection via Attentive Multi-Modal Learning for Microservice System

This is the Pytorch implementation of MSTGAD in the ASE 2023: Twin Graph-based Anomaly Detection via Attentive Multi-Modal Learning for Microservice System. You may refer to our paper for more details.

Environment

The repository has some important dependencies below

Ubuntu 18.04
Python 3.8
Pytorch 1.12.0
Pytorch_geometric == 2.2.0

Install other dependencies can be installed by:

pip install -r requirements.txt

Dataset

The MSDS datasets used in this paper can be downloaded from the Multi-Source Distributed System Data for AI-powered Analytics | Zenodo

The other dataset that we don't have a permission to publish

The downloaded datasets can be put in the 'data' directory. The directory structure looks like:

${CODE_ROOT}
    ......
    |-- data
        |-- MSDS
            |-- concurrent_data

Training

To preprocess the data, run:

python util/pre_MSDS.py

To start training, run:

python main.py

Architecture

Citation

@inproceedings{Huang2023MSTGAD,
  author       = {Huang, Jun and Yang, Yang and Yu, Hang and Li,Jianguo and Zheng, Xiao},
  title        = {Twin Graph-based Anomaly Detection via Attentive Multi-Modal Learning for Microservice System},
  booktitle    = {38th {IEEE/ACM} International Conference on Automated Software Engineering, {ASE} 2023},
  year         = {2023},
  page         = {66 - 78},
}

Contact

For any questions w.r.t. MSTGAD, please submit them to Github Issues .