Entropy Causal Graphs for Multivariate Time Series Anomaly Detection

2023-12-15

arXiv.org

AI Summary

Existing multivariate time-series anomaly detection methods often neglect intrinsic causal relationships among variables, limiting detection performance. This paper proposes a novel framework integrating causal inference with dynamic modeling: it introduces transfer entropy to construct a directed causal graph for the first time, and jointly captures structural and temporal dependencies via a weighted graph convolutional network and causal convolutions. Additionally, we design a robust anomaly scoring and normalization mechanism based on the Median Absolute Deviation (MAD). Evaluated on multiple real-world datasets, our method achieves an average 15% improvement in F1-score, precision, and recall over current state-of-the-art approaches. The core contributions lie in (i) a paradigm shift in causal graph construction—leveraging transfer entropy for interpretable, data-driven causal discovery—and (ii) a robust, distribution-agnostic anomaly scoring scheme that enhances generalization across diverse time-series patterns.

Abstract

Many multivariate time series anomaly detection frameworks have been proposed and widely applied. However, most of these frameworks do not consider intrinsic relationships between variables in multivariate time series data, thus ignoring the causal relationship among variables and degrading anomaly detection performance. This work proposes a novel framework called CGAD, an entropy Causal Graph for multivariate time series Anomaly Detection. CGAD utilizes transfer entropy to construct graph structures that unveil the underlying causal relationships among time series data. Weighted graph convolutional networks combined with causal convolutions are employed to model both the causal graph structures and the temporal patterns within multivariate time series data. Furthermore, CGAD applies anomaly scoring, leveraging median absolute deviation-based normalization to improve the robustness of the anomaly identification process. Extensive experiments demonstrate that CGAD outperforms state-of-the-art methods on real-world datasets with a 15% average improvement based on three different multivariate time series anomaly detection metrics.

Problem

• Detects anomalies in multivariate time series data
• Models causal relationships between time series variables
• Improves robustness of anomaly identification process

Innovation

• Transfer entropy constructs causal graph structures
• Graph convolutional networks model temporal patterns
• Median absolute deviation improves anomaly scoring