Multivariate Time-Series Anomaly Detection via Graph Attention Network
Обнаружение аномалий в многомерных временных рядах с помощью графовой сети внимания
2020-11-01
SCID: 54.1/st32dva5
Discuss with AI
forecasting and reconstructiongraph attention networkmultivariate time-series anomaly detectionself-supervised learningtemporal and feature dependencies
Figures from the paper
Abstract (AI)
Anomaly detection on multivariate time-series is of great importance in both data mining research and industrial applications. Recent approaches have achieved significant progress in this topic, but there is remaining limitations. One major limitation is that they do not capture the relationships between different time-series explicitly, resulting in inevitable false alarms. In this paper, we propose a novel self-supervised framework for multivariate time-series anomaly detection to address this issue. Our framework considers each univariate time-series as an individual feature and includes two graph attention layers in parallel to learn the complex dependencies of multivariate time-series in both temporal and feature dimensions. In addition, our approach jointly optimizes a forecasting-based model and a reconstruction-based model, obtaining better time-series representations through a combination of single-timestamp prediction and reconstruction of the entire time-series. We demonstrate the efficacy of our model through extensive experiments. The proposed method outperforms other state-of-the-art models on three real-world datasets. Further analysis shows that our method has good interpretability and is useful for anomaly diagnosis.
Key Findings
1
Further analysis indicates that the approach is interpretable and supports anomaly diagnosis.
2
Joint optimization of forecasting and full-series reconstruction produces improved representations for anomaly detection.
3
Parallel graph attention layers learn complex dependencies across both temporal and feature dimensions.
4
The method outperforms other state-of-the-art models on three real-world datasets.
5
The proposed self-supervised framework explicitly models relationships among individual time series to reduce false alarms in multivariate anomaly detection.
Research Object
multivariate time-series
Research Subject
anomalies and inter-series dependencies, including their temporal and feature-dimensional relationships, for anomaly detection and diagnosis
Publication Details
Publication Date
2020-11-01
Journal
Publisher
ISSN
Cited by
841
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai5
Exploiting Generative AI to Scale up Intelligent Tutoring Systems2023
Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling2014
Robust Anomaly Detection for Multivariate Time Series through Stochastic Recurrent Neural Network2019
Unsupervised Anomaly Detection via Variational Auto-Encoder for Seasonal KPIs in Web Applications2018
Time-Series Anomaly Detection Service at Microsoft2019
Cited by8
Anomaly detection in time series: a comprehensive evaluation2022
Deep Learning for Time Series Anomaly Detection: A Survey2024
Deep Learning for Anomaly Detection in Time-Series Data: Review, Analysis, and Guidelines2021
Learning Graph Structures With Transformer for Multivariate Time-Series Anomaly Detection in IoT2021
A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection2024
DCdetector: Dual Attention Contrastive Representation Learning for Time Series Anomaly Detection2023
A Survey of Deep Anomaly Detection in Multivariate Time Series: Taxonomy, Applications, and Directions2025
Time Series Anomaly Detection Leveraging MSE Feedback with AutoEncoder and RNN2024