[!IMPORTANT] We are still working on the readme, stay tunned...

[TADNet] Unravel Anomalies: An End-to-End Seasonal-Trend Decomposition Approach for Time Series Anomaly Detection

This repository This repository contains the code for the paper "Unravel Anomalies: An End-to-End Seasonal-Trend Decomposition Approach for Time Series Anomaly Detection" by Zhenwei Zhang; Ruiqi Wang; Ran Ding; Yuantao Gu, published in the IEEE ICASSP 2024 (International Conference on Acoustics, Speech, and Signal Processing).

Introduction

:triangular_flag_on_post: Presentation Slides for this paper can be found on IEEE SigPort (Download).

Traditional Time-series Anomaly Detection (TAD) methods often struggle with the composite nature of complex time-series data and a diverse array of anomalies. We introduce TADNet, an end-to-end TAD model that leverages Seasonal-Trend Decomposition to link various types of anomalies to specific decomposition components, thereby simplifying the analysis of complex time-series and enhancing detection performance. Our training methodology, which includes pre-training on a synthetic dataset followed by fine-tuning, strikes a balance between effective decomposition and precise anomaly detection. Experimental validation on real-world datasets confirms TADNet’s state-of-the-art performance across a diverse range of anomalies.

Datasets

For more details on the datasets used in the paper, please refer to this repo.

• UCR:
• SMD:
• SWaT:
• PSM:
• WADI:

Preparation

<!-- Preprocess all datasets using the command -->

Generate the synthetic dataset using the command:

python run.py --mode synthetic 

Training & Evaluation

Train the model using the command:

python run.py --mode pretrain --loss 2
python run.py --file_dir xxx.npy --mode finetune --loss 5 --number xxx --exists 1

Evaluate the model using the command:

python run.py --file_dir xxx.npy --mode test --number xxx --exists 1

Citation

If you find this work useful, please consider citing the following paper:

@INPROCEEDINGS{10446482,
  author={Zhang, Zhenwei and Wang, Ruiqi and Ding, Ran and Gu, Yuantao},
  booktitle={ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)}, 
  title={Unravel Anomalies: an End-to-End Seasonal-Trend Decomposition Approach for Time Series Anomaly Detection}, 
  year={2024},
  volume={},
  number={},
  pages={5415-5419},
  keywords={Training;Analytical models;Time series analysis;Data visualization;Signal processing;Data models;Arrays;time-series anomaly detection;seasonal-trend decomposition;time-series analysis;end-to-end},
  doi={10.1109/ICASSP48485.2024.10446482}}

License

This project is licensed under the MIT License - see the LICENSE file for details.