A Survey of Deep Anomaly Detection in Multivariate Time Series: Taxonomy, Applications, and Directions

Multivariate time series anomaly detection (MTSAD) can effectively identify and analyze anomalous behavior in complex systems, which is particularly important in fields such as financial monitoring, industrial equipment fault detection, and cybersecurity. MTSAD requires simultaneously analyzing temporal dependencies and inter-variable relationships, prompting researchers to develop specialized deep learning models to detect anomalous patterns. This paper presents a structured and comprehensive overview of recent deep learning techniques for MTSAD. It proposes a taxonomy of anomaly detection strategies from the perspectives of learning paradigms and deep learning models and provides a systematic review emphasizing their advantages and drawbacks. Public datasets for time series anomaly detection and their application domains are organized. Finally, open issues for future research on MTSAD are identified.

Fengling Wang, Yiyue Jiang, Rongjie Zhang, Aimin Wei, Jingming Xie, Xiongwen Pang