Overview

This is the implementation for the BeatGAN model architecture described in the paper: "BeatGAN: Anomalous Rhythm Detection using Adversarially Generated Time Series".

In this paper, we propose an unsupervised anomaly detection algorithm for time series data. BeatGAN has the following advantages: 1) Unsupervised: it is applicable even when labels are unavailable; 2) Effectiveness: It outperforms baselines in both accuracy and inference speed, achieving accuracy of nearly 0.95 AUC on ECG data and very fast inference (2.6 ms per beat); 3) Explainability: It pinpoints the time ticks involved in the anomalous patterns, providing interpretable output for visualization and attention routing; 4) Generality: BeatGAN also successfully detects unusual moions in multivariate motion-capture database.

DataSet

• For full MIT-BIH dataset, download from
  https://www.dropbox.com/sh/b17k2pb83obbrkn/AADzJigiIrottyTOyvAEU1LOa?dl=0 （contain preprocessed data） and place them in experiments/ecg/dataset/preprocessed/

• For motion capture dataset, in experiments/mocap/dataset

Usage

• For ecg full experiemnt (need to download full dataset)

  sh run_ecg.sh

• For ecg demo (there are demo data in experiments/ecg/dataset/demo, the output dir is in experiments/ecg/output/beatgan/ecg/demo )

  sh run_ecg_demo.sh

• For motion experiment

  sh run_mocap.sh

Require

• Python 3

Packages

• PyTorch (1.0.0)
• scikit-learn (0.20.0)
• biosppy (0.6.1) # For data preprocess
• tqdm (4.28.1)
• matplotlib (3.0.2)

Reference

If you find this code useful in your research, please, consider citing our paper:

@inproceedings{zhou2019beatgan,
  title={BeatGAN: Anomalous Rhythm Detection using Adversarially Generated Time Series},
  author={Zhou, Bin and Liu, Shenghua and Bryan Hooi and Cheng, Xueqi and Ye, Jing },
  booktitle={International Joint Conference on Artificial Intelligence},
  year={2019},
}