Home > Research > Publications & Outputs > HFN

Electronic data

  • revised manuscript R2

    Accepted author manuscript, 5.17 MB, PDF document

    Available under license: CC BY: Creative Commons Attribution 4.0 International License

Links

Text available via DOI:

View graph of relations

HFN: Heterogeneous feature network for multivariate time series anomaly detection

Research output: Contribution to Journal/MagazineJournal articlepeer-review

E-pub ahead of print

Standard

HFN: Heterogeneous feature network for multivariate time series anomaly detection. / Zhan, Jun; Wu, Chengkun; Yang, Canqun et al.
In: Information Sciences, Vol. 670, 120626, 30.06.2024.

Research output: Contribution to Journal/MagazineJournal articlepeer-review

Harvard

APA

Zhan, J., Wu, C., Yang, C., Miao, Q., & Ma, X. (2024). HFN: Heterogeneous feature network for multivariate time series anomaly detection. Information Sciences, 670, Article 120626. Advance online publication. https://doi.org/10.1016/j.ins.2024.120626

Vancouver

Zhan J, Wu C, Yang C, Miao Q, Ma X. HFN: Heterogeneous feature network for multivariate time series anomaly detection. Information Sciences. 2024 Jun 30;670:120626. Epub 2024 Apr 18. doi: 10.1016/j.ins.2024.120626

Author

Zhan, Jun ; Wu, Chengkun ; Yang, Canqun et al. / HFN : Heterogeneous feature network for multivariate time series anomaly detection. In: Information Sciences. 2024 ; Vol. 670.

Bibtex

@article{b541cda0398649b09a7bb61be92b5a6f,
title = "HFN: Heterogeneous feature network for multivariate time series anomaly detection",
abstract = "As the key step of anomaly detection for multivariate time-series (MTS) data, learning the relations among different variables has been explored by many approaches. However, most existing approaches overlook the heterogeneity among variables, that is, different types of variables (continuous numerical variables, discrete categorical variables or hybrid variables) may have different edge distributions. In this paper, we propose a novel semi-supervised anomaly detection framework based on a heterogeneous feature network (HFN) for MTS. Specifically, we first combine the embedding similarity subgraph generated by sensor embedding and the feature value similarity subgraph generated by sensor values to construct a time-series heterogeneous graph, which fully utilizes the rich heterogeneous mutual information among variables. Then, a prediction model containing nodes and channel attentions is jointly optimized to obtain better time-series representations. This approach fuses the state-of-the-art technologies of heterogeneous graph structure learning (HGSL) and representation learning. Experimental results on four sensor datasets from real-world applications demonstrate that our approach achieves more accurate anomaly detection compared to baseline methods, laying a foundation for the rapid positioning of anomalies.",
keywords = "Heterogeneous neural network, Anomaly detection, Multi-sensor data, Multivariate time series, Deep learning",
author = "Jun Zhan and Chengkun Wu and Canqun Yang and Qiucheng Miao and Xiandong Ma",
year = "2024",
month = apr,
day = "18",
doi = "10.1016/j.ins.2024.120626",
language = "English",
volume = "670",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier Inc.",

}

RIS

TY - JOUR

T1 - HFN

T2 - Heterogeneous feature network for multivariate time series anomaly detection

AU - Zhan, Jun

AU - Wu, Chengkun

AU - Yang, Canqun

AU - Miao, Qiucheng

AU - Ma, Xiandong

PY - 2024/4/18

Y1 - 2024/4/18

N2 - As the key step of anomaly detection for multivariate time-series (MTS) data, learning the relations among different variables has been explored by many approaches. However, most existing approaches overlook the heterogeneity among variables, that is, different types of variables (continuous numerical variables, discrete categorical variables or hybrid variables) may have different edge distributions. In this paper, we propose a novel semi-supervised anomaly detection framework based on a heterogeneous feature network (HFN) for MTS. Specifically, we first combine the embedding similarity subgraph generated by sensor embedding and the feature value similarity subgraph generated by sensor values to construct a time-series heterogeneous graph, which fully utilizes the rich heterogeneous mutual information among variables. Then, a prediction model containing nodes and channel attentions is jointly optimized to obtain better time-series representations. This approach fuses the state-of-the-art technologies of heterogeneous graph structure learning (HGSL) and representation learning. Experimental results on four sensor datasets from real-world applications demonstrate that our approach achieves more accurate anomaly detection compared to baseline methods, laying a foundation for the rapid positioning of anomalies.

AB - As the key step of anomaly detection for multivariate time-series (MTS) data, learning the relations among different variables has been explored by many approaches. However, most existing approaches overlook the heterogeneity among variables, that is, different types of variables (continuous numerical variables, discrete categorical variables or hybrid variables) may have different edge distributions. In this paper, we propose a novel semi-supervised anomaly detection framework based on a heterogeneous feature network (HFN) for MTS. Specifically, we first combine the embedding similarity subgraph generated by sensor embedding and the feature value similarity subgraph generated by sensor values to construct a time-series heterogeneous graph, which fully utilizes the rich heterogeneous mutual information among variables. Then, a prediction model containing nodes and channel attentions is jointly optimized to obtain better time-series representations. This approach fuses the state-of-the-art technologies of heterogeneous graph structure learning (HGSL) and representation learning. Experimental results on four sensor datasets from real-world applications demonstrate that our approach achieves more accurate anomaly detection compared to baseline methods, laying a foundation for the rapid positioning of anomalies.

KW - Heterogeneous neural network

KW - Anomaly detection

KW - Multi-sensor data

KW - Multivariate time series

KW - Deep learning

U2 - 10.1016/j.ins.2024.120626

DO - 10.1016/j.ins.2024.120626

M3 - Journal article

VL - 670

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

M1 - 120626

ER -