Adversarial-Robust Multivariate Time-Series Anomaly Detection via Joint Information Retention

View PDF HTML (experimental)

Abstract:Time-series anomaly detection (TSAD) is a critical component in monitoring complex systems, yet modern deep learning-based detectors are often highly sensitive to localized input corruptions and structured noise. We propose ARTA (Adversarially Robust multivariate Time-series Anomaly detection via joint information retention), a joint training framework that improves detector robustness through a principled min-max optimization objective. ARTA comprises an anomaly detector and a sparsity-constrained mask generator that are trained simultaneously. The generator identifies minimal, task-relevant temporal perturbations that maximally increase the detector's anomaly score, while the detector is optimized to remain stable under these structured perturbations. The resulting masks characterize the detector's sensitivity to adversarial temporal corruptions and can serve as explanatory signals for the detector's decisions. This adversarial training strategy exposes brittle decision pathways and encourages the detector to rely on distributed and stable temporal patterns rather than spurious localized artifacts. We conduct extensive experiments on the TSB-AD benchmark, demonstrating that ARTA consistently improves anomaly detection performance across diverse datasets and exhibits significantly more graceful degradation under increasing noise levels compared to state-of-the-art baselines.

Comments: 22 pages, 4 figures

Subjects:

Machine Learning (cs.LG)

Cite as: arXiv:2603.25956 [cs.LG]

(or arXiv:2603.25956v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2603.25956

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Hadi Hojjati [view email] [v1] Thu, 26 Mar 2026 22:53:03 UTC (222 KB)