Robust Group Anomaly Detection for Quasi-Periodic Network Time Series

TL;DR

This paper introduces seq2GMM, a novel machine learning framework for detecting anomalies in quasi-periodic network time series, demonstrating superior performance and theoretical convergence analysis.

Contribution

The paper presents seq2GMM, a new sequence-to-Gaussian Mixture Model framework for robust anomaly detection in quasi-periodic network time series, with an efficient training algorithm and theoretical guarantees.

Findings

Outperforms existing anomaly detection methods on benchmark datasets

Provides theoretical analysis of the convergence of the training algorithm

Demonstrates strong empirical results across multiple public datasets

Abstract

Many real-world multivariate time series are collected from a network of physical objects embedded with software, electronics, and sensors. The quasi-periodic signals generated by these objects often follow a similar repetitive and periodic pattern, but have variations in the period, and come in different lengths caused by timing (synchronization) errors. Given a multitude of such quasi-periodic time series, can we build machine learning models to identify those time series that behave differently from the majority of the observations? In addition, can the models help human experts to understand how the decision was made? We propose a sequence to Gaussian Mixture Model (seq2GMM) framework. The overarching goal of this framework is to identify unusual and interesting time series within a network time series database. We further develop a surrogate-based optimization algorithm that can efficiently train the seq2GMM model. Seq2GMM exhibits strong empirical performance on a plurality of public benchmark datasets, outperforming state-of-the-art anomaly detection techniques by a significant margin. We also theoretically analyze the convergence property of the proposed training algorithm and provide numerical results to substantiate our theoretical claims.