---

Here is a repo for our paper published at ICSE 2023: Heterogeneous Anomaly Detection for Software Systems via Semi-supervised Cross-modal Attention.

Data

Our data are at https://doi.org/10.5281/zenodo.7609780.

Environment

We support python3.x $\geq$ 3.7. The environment can be built by: $ pip install -r requirements.txt

Reproducing

$ cd codes && python run.py --data ../data/chunk_10

Thanks to nni (https://nni.readthedocs.io/zh/stable/index.html), we can automate hyper-parameter tuning.

Architecture

Tree

.
├── README.md
├── codes
│   ├── common
│   │   ├── __init__.py
│   │   ├── data_loads.py
│   │   ├── semantics.py
│   │   └── utils.py
│   ├── models
│   │   ├── __init__.py
│   │   ├── base.py
│   │   ├── fuse.py
│   │   ├── kpi_model.py
│   │   ├── log_model.py
│   │   └── utils.py
│   └── run.py
├── data
│   └── chunk_10
│       ├── test.pkl
│       ├── train.pkl
│       └── unlabel.pkl
├── preprocess
│   ├── README.md
│   ├── get_chunks.py
│   └── split_data.py
├── requirements.txt

Result

The trained model and the final result records are in the result directory.

Preview