Contextual and Seasonal LSTMs for Time Series Anomaly Detection

TL;DR

This paper introduces CS-LSTMs, a novel prediction-based framework that combines contextual and seasonal information with noise decomposition to improve detection of subtle anomalies in univariate time series.

Contribution

The paper presents a new prediction-based approach, CS-LSTMs, that effectively captures seasonal and contextual dependencies for anomaly detection in time series.

Findings

CS-LSTMs outperform existing methods on benchmark datasets.

The approach effectively detects subtle and slowly rising anomalies.

Integration of time and frequency domain representations enhances modeling accuracy.

Abstract

Univariate time series (UTS), where each timestamp records a single variable, serve as crucial indicators in web systems and cloud servers. Anomaly detection in UTS plays an essential role in both data mining and system reliability management. However, existing reconstruction-based and prediction-based methods struggle to capture certain subtle anomalies, particularly small point anomalies and slowly rising anomalies. To address these challenges, we propose a novel prediction-based framework named Contextual and Seasonal LSTMs (CS-LSTMs). CS-LSTMs are built upon a noise decomposition strategy and jointly leverage contextual dependencies and seasonal patterns, thereby strengthening the detection of subtle anomalies. By integrating both time-domain and frequency-domain representations, CS-LSTMs achieve more accurate modeling of periodic trends and anomaly localization. Extensive evaluations on public benchmark datasets demonstrate that CS-LSTMs consistently outperform state-of-the-art methods, highlighting their effectiveness and practical value in robust time series anomaly detection.