SimAD

This is the implementation of SimAD: A Simple Dissimilarity-based Approach for Time Series Anomaly Detection

The paper is under review. You can download the paper from arXiv.

Abstract

Despite the prevalence of reconstruction-based deep learning methods, time series anomaly detection remains a challenging task. Existing approaches often struggle with limited temporal contexts, inadequate representation of normal patterns, and flawed evaluation metrics, hindering their effectiveness in identifying aberrant behavior. To address these issues, we introduce SimAD, a Simple dissimilarity-based approach for time series Anomaly Detection. SimAD leverages a feature extractor capable of handling extended time windows, employs the EmbedPatch encoder to effectively integrate normal features, and introduces a novel ContrastFusion module to amplify distributional discrepancies between normal and anomalous data, thereby facilitating robust anomaly discrimination. Additionally, we propose two robust evaluation metrics, UAff and NAff, addressing the limitations of existing metrics and demonstrating their reliability through theoretical and experimental analyses. Experiments across seven diverse time series datasets demonstrate SimAD's superior performance compared to state-of-the-art methods, achieving relative improvements of 19.85% on F1, 4.44% on Aff-F1, 77.79% on NAff-F1, and 9.69% on AUC on six multivariate datasets. Code and pre-trained models are available at https://github.com/EmorZz1G/SimAD.

Architecture

Overall Performance

Model download link

https://drive.google.com/drive/folders/1dDH3JRivRYEU02riHzGFUu74OxiwDax6?usp=sharing ¹

Datasets

You can download all datasets here. (Thanks for DCdetector repo and its authors.)

Installation

You can refer to requirements.txt to install all the packages.

pip install -r requirements.txt

Citation

If you find this repo useful, please cite our paper or leave a star.

@misc{zhong2024simad,
    title={SimAD: A Simple Dissimilarity-based Approach for Time Series Anomaly Detection},
    author={anonymous authors},
    year={2024},
    eprint={2405.11238},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

Footnotes

1. The model trained on MSL dataset is too large. We are willing to upload it in the future. ↩