EdgeConvFormer: Dynamic Graph CNN and Transformer based Anomaly Detection in Multivariate Time Series

TL;DR

EdgeConvFormer is a novel model combining dynamic graph CNN and Transformer techniques, designed to improve anomaly detection in multivariate time series by capturing complex spatial-temporal correlations more effectively.

Contribution

It introduces a new architecture integrating Time2vec, dynamic graph CNN, and Transformer to better model interdependencies in multivariate time series data.

Findings

Outperforms state-of-the-art methods on multiple real-world datasets.

Effectively captures complex spatial-temporal correlations.

Demonstrates improved anomaly detection accuracy.

Abstract

Transformer-based models for anomaly detection in multivariate time series can benefit from the self-attention mechanism due to its advantage in modeling long-term dependencies. However, Transformer-based anomaly detection models have problems such as a large amount of data being required for training, standard positional encoding is not suitable for multivariate time series data, and the interdependence between time series is not considered. To address these limitations, we propose a novel anomaly detection method, named EdgeConvFormer, which integrates Time2vec embedding, stacked dynamic graph CNN, and Transformer to extract global and local spatial-time information. This design of EdgeConvFormer empowers it with decomposition capacities for complex time series, progressive spatiotemporal correlation discovery between time series, and representation aggregation of multi-scale features. Experiments demonstrate that EdgeConvFormer can learn the spatial-temporal correlations from multivariate time series data and achieve better anomaly detection performance than the state-of-the-art approaches on many real-world datasets of different scales.