Learning Unified Representations of Normalcy for Time Series Anomaly Detection

TL;DR

This paper introduces U2AD, a novel unsupervised framework for multivariate time series anomaly detection that models normal data distribution using score-based generative modeling and outperforms existing methods.

Contribution

It proposes a unified training approach with a time-dependent score network to effectively learn normal data manifolds considering temporal contexts.

Findings

U2AD outperforms state-of-the-art methods in detection accuracy.

It detects anomalies earlier in their occurrence.

The method effectively models the normal data distribution.

Abstract

The core challenge in unsupervised anomaly detection is identifying abnormal patterns without prior knowledge of their characteristics. While existing methods have addressed aspects of this problem, they often struggle to learn a robust representation of the normal data distribution that is distinct from anomalous patterns. In this paper, we present a novel framework, Unified Unsupervised Anomaly Detection ($\text{U}^2\text{AD}$), that comprehensively addresses anomaly detection in multivariate time series. Our approach learns the underlying data distribution of normal samples by utilizing score-based generative modeling. We introduce a novel time-dependent score network and a unified training objective that together delineate the manifold of normal data while considering both local and global temporal contexts. Reconstruction is then performed via a deterministic sampling process using an ordinary differential equation solver. Our extensive experimental evaluations demonstrate that $\text{U}^2\text{AD}$ not only outperforms current state-of-the-art methods in detection accuracy but also identifies anomalies at significantly earlier stages of their occurrence.