analiza aiqbench
This paper argues that for time series anomaly detection, a simple supervised baseline (STAND) can match or beat complex unsupervised models when historical incident labels exist. The results come from five benchmarks under a strict out-of-sample protocol, with label-scaling and few-shot tests. Readers building monitoring systems should weigh the cost of labeling past incidents against the cost of scaling unsupervised architectures, though the advantage depends on label availability and the authors' chosen evaluation setup.
Rethinking the Unsupervised Paradigm in Time Series Anomaly Detection: An Empirical Study and Minimalist Baseline
arXiv:2511.16145v3 Announce Type: replace Abstract: Time series anomaly detection (TSAD) research predominantly focuses on unsupervised architectures to model normal data dynamics under label scarcity. However, practical deployments frequently archive historical operational incidents, questioning whether continuous unsupervised architectural scaling remains the most cost-effective pursuit. In this paper, we empirically quantify the trade-offs between supervisory signals and model complexity. We introduce STAND, a deliberately minimalist supervised baseline based on bidirectional recurrent encoding. Across five benchmark datasets evaluated under a strictly disjoint out-of-sample temporal protocol, minimalist supervised models achieve competitive or superior accuracy relative to complex unsupervised alternatives. Furthermore, controlled label-scaling and few-shot evaluations confirm that these supervisory advantages persist under constrained historical event budgets, while STAND exhibits higher prediction certainty and sharper event delineation than reconstruction-based baselines. These findings encourage the TSAD community to complement algorithmic complexity with data-centric label utilization when annotations are accessible. The source code, benchmark datasets, and an extended Supplementary Document are publicly available at https://emorzz1g.github.io/STAND.
Abstract:Time series anomaly detection (TSAD) research predominantly focuses on unsupervised architectures to model normal data dynamics under label scarcity. However, practical deployments frequently archive historical operational incidents, questioning whether continuous unsupervised architectural scaling remains the most cost-effective pursuit. In this paper, we empirically quantify the trade-offs between supervisory signals and model complexity. We introduce STAND, a deliberately minimalist supervised baseline based on bidirectional recurrent encoding. Across five benchmark datasets evaluated under a strictly disjoint out-of-sample temporal protocol, minimalist supervised models achieve competitive or superior accuracy relative to complex unsupervised alternatives. Furthermore, controlled label-scaling and few-shot evaluations confirm that these supervisory advantages persist under constrained historical event budgets, while STAND exhibits higher prediction certainty and sharper event delineation than reconstruction-based baselines. These findings encourage the TSAD community to complement algorithmic complexity with data-centric label utilization when annotations are accessible. The source code, benchmark datasets, and an extended Supplementary Document are publicly available at this https URL.
25 pages, 10 figures, 7 tables. Accepted for publication in Information Sciences
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2511.16145 [cs.LG]
(or arXiv:2511.16145v3 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2511.16145 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 262 cuvinte
Subiecte conexe
Understanding Confabulation and Rethinking Reconstruction in Activation Explanations
arXiv · 30 sept. 2026
Why Do LLMs Corrupt Your Documents When You Delegate?
KDnuggets · 8 iun. 2026
Rethinking Object-Centric Representations for Video Dynamics Modeling
arXiv · 30 sept. 2026
Beyond Static Cues: Unified Dynamic Detection of MLLM Backdoors via Dual-Flow Trajectory Analysis
arXiv · 15 oct. 2026