Multivariate Time Series Anomaly Detection by Capturing Coarse-Grained Intra- and Inter-Variate Dependencies
Yongzheng Xie, Hongyu Zhang, Muhammad Ali Babar
TL;DR
MtsCID tackles semi-supervised multivariate time series anomaly detection by learning coarse-grained intra-variate temporal dependencies and inter-variate relationships through a dual-network design that operates in both time and frequency domains. It combines a temporal autoencoder with multi-scale patch attention and an inter-variate encoder augmented by a sinusoidal prototypes interaction module, optimized with a temporal reconstruction loss and a prototype-oriented entropy loss. Across seven public datasets, MtsCID achieves state-of-the-art or near-best results, with notable improvements on challenging GECCO data, and ablation studies confirm the value of dual branches, frequency-domain processing, and coarse-grained representations. The approach offers a scalable, robust solution with practical applicability to failure detection in complex web and IoT environments, while suggesting avenues for further enhancement via self-supervised learning and deeper frequency-domain analysis.
Abstract
Multivariate time series anomaly detection is essential for failure management in web application operations, as it directly influences the effectiveness and timeliness of implementing remedial or preventive measures. This task is often framed as a semi-supervised learning problem, where only normal data are available for model training, primarily due to the labor-intensive nature of data labeling and the scarcity of anomalous data. Existing semi-supervised methods often detect anomalies by capturing intra-variate temporal dependencies and/or inter-variate relationships to learn normal patterns, flagging timestamps that deviate from these patterns as anomalies. However, these approaches often fail to capture salient intra-variate temporal and inter-variate dependencies in time series due to their focus on excessively fine granularity, leading to suboptimal performance. In this study, we introduce MtsCID, a novel semi-supervised multivariate time series anomaly detection method. MtsCID employs a dual network architecture: one network operates on the attention maps of multi-scale intra-variate patches for coarse-grained temporal dependency learning, while the other works on variates to capture coarse-grained inter-variate relationships through convolution and interaction with sinusoidal prototypes. This design enhances the ability to capture the patterns from both intra-variate temporal dependencies and inter-variate relationships, resulting in improved performance. Extensive experiments across seven widely used datasets demonstrate that MtsCID achieves performance comparable or superior to state-of-the-art benchmark methods.