TeamWork: Multivariate Time Series Anomaly Detection via Asymmetric Role-aware Channel Modeling

Multivariate time series anomaly detection remains challenging as it requires the joint modeling of variable relationships and temporal dependencies. Existing methods often struggle to balance channel relationship modeling and overlook the relative importance of different variables within multivariate time series. To address this, we propose TeamWork, an asymmetric role-aware channel modeling framework that decouples variables into dominant and auxiliary roles according to their contributions to uncertainty reduction. Dominant variables drive system evolution and their deviations more strongly disrupt normal patterns, while auxiliary variables provide complementary cues. These variables with different roles are integrated through a role-aware gated interaction module. Moreover, point and subsequence anomalies can exist in multiple periodic systems, and the same anomaly type may behave differently across short- and long-period series. To capture such variations, we introduce a period-aware masked modeling mechanism. It employs multiple specialized masking mechanisms spanning short to long periods to facilitate comprehensive temporal dependency learning. Extensive experiments on multiple real-world datasets demonstrate that TeamWork achieves superior performance compared with state-of-the-art methods.

Many real-world systems, such as industrial equipment, data centers, spacecraft, and financial platforms, are monitored through multiple signals collected over time. When something unusual happens, these signals may change in subtle or complicated ways. Detecting such anomalies is important, but it is challenging because different signals do not contribute equally to the system, and abnormal patterns may appear at different time scales. In this paper, we propose TeamWork, a method for multivariate time series anomaly detection. The key idea is to distinguish between signals that are more representative of the overall system behavior and signals that provide additional supporting information. TeamWork then combines these two types of signals in a role-aware manner. It also learns from different temporal patterns, so that both short, sudden changes and longer abnormal segments can be better captured. Experiments on several real-world datasets show that TeamWork performs better than existing methods. These results suggest that TeamWork may be useful for building more reliable monitoring tools in areas such as industrial maintenance, server operation, and risk detection.