Time series anomaly detection adapts using specialized expert models
Generalist Representation, Specialist Detection: TS-Router for Time-Series Anomaly Detection
Machine LearningArtificial Intelligence
Summary
Detecting unusual patterns in time-based data from different sources is hard because what counts as normal varies a lot. The authors propose a system called TS-Router that uses a general representation model combined with multiple expert detectors, choosing the best ones for each dataset. Instead of needing labeled examples from the actual data, TS-Router learns which experts perform well from simulated tasks and then adapts unsupervised to new data. This approach improves anomaly detection across many real-world datasets compared to previous methods.
What this means in practice
- •For data engineering teams: Improve anomaly detection pipelines by automatically selecting specialized detectors tailored to each time-series dataset, without needing labeled anomaly data.
- •For industrial monitoring teams: Detect equipment faults in diverse sensor data by combining pretrained models with adaptive specialist detectors that fit new operational data unsupervised.
Authors
Tian Lan, Yifei Gao, Yimeng Lu, Xuming An, Meng Wang, Yue Pan, Wenjun He, Chenghao Liu, Chen Zhang
Abstract
Time-series anomaly detection (TSAD) is difficult to generalize across datasets because heterogeneous temporal dynamics imply different notions of normality and favor different detection criteria. While time-series foundation models provide transferable representations, coupling them with a fixed anomaly-scoring mechanism can overlook this variation. This motivates a different perspective on foundation-model-based TSAD: using foundation models to coordinate specialized anomaly criteria rather than directly imposing a universal one. Based on this view, we propose \textbf{TS-Router}, a generalist-representation, specialist-detection framework that estimates the relative competence of heterogeneous anomaly detectors from pretrained temporal representations and selects suitable specialists for each target series. To avoid relying on specialist-performance labels from real tasks, we derive soft competence supervision from specialists' relative performance on labeled simulated tasks. At deployment, routing requires no target anomaly labels, and only the selected specialists are fitted unsupervisedly on the target series. We bound Top-\(k\) set-competence regret under representation coverage and conditional competence stability. Across 16 real-world benchmarks and four complementary evaluation metrics, TS-Router achieves the best overall average rank. Controlled ablations with multiple frozen TSFM encoders further support the use of pretrained representations for competence estimation and adaptive specialist selection. The code is available at https://anonymous.4open.science/r/TS-Router-D8FF.