From Reconstruction to Generation: A Survey of Multivariate Time Series Anomaly Detection

DOI

10.2991/978-94-6239-721-7_8How to use a DOI?

Keywords

Multivariate Time Series; Anomaly Detection; Diffusion Models; Graph Neural Networks; Large Language Models; Contrastive Learning

Abstract

Multivariate time series (MTS) anomaly detection is a core task for ensuring system reliability in industrial AIOps. Traditional methods typically rely on reconstruction or prediction errors, which often fail to capture complex data distributions. This paper reviews the transition of detection paradigms from discriminative to generative approaches. We categorize existing techniques into four groups: prediction, reconstruction, explicit relationship modeling, and generative frameworks. Specifically, we analyze the capability of Graph Neural Networks (GNNs) in capturing variable dependencies and discuss the trade-offs of diffusion models in distribution fitting. Additionally, we evaluate modern backbones, such as ModernTCN, regarding their feature extraction efficiency, and benchmark these paradigms across standard datasets. Recent research has increasingly focused on generative models, particularly diffusion models, due to their superior performance in modeling complex data distributions.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Danyang Cao
AU  - Hanyu Cui
AU  - Jiayi Fu
PY  - 2026
DA  - 2026/07/06
TI  - From Reconstruction to Generation: A Survey of Multivariate Time Series Anomaly Detection
BT  - Proceedings of the 2026 6th International Conference on Public Management and Intelligent Society (PMIS 2026)
PB  - Atlantis Press
SP  - 75
EP  - 82
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6239-721-7_8
DO  - 10.2991/978-94-6239-721-7_8
ID  - Cao2026
ER  -