Unsupervised Multivariate Time Series Data Anomaly Detection in Industrial IoT: A Confidence Adversarial Autoencoder Network

Anomaly detection of multivariate time series (MTS) is crucial in industrial intelligent systems. To address the challenges of absence of anomaly labels, fast inference time, multi-source and multi-modality in anomaly detection, researchers have primarily investigated unsupervised reconstruction-dri...

Full description

Saved in:
Bibliographic Details
Main Authors: Jiahao Shan, Donghong Cai, Fang Fang, Zahid Khan, Pingzhi Fan
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Open Journal of the Communications Society
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10778252/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850250325474148352
author Jiahao Shan
Donghong Cai
Fang Fang
Zahid Khan
Pingzhi Fan
author_facet Jiahao Shan
Donghong Cai
Fang Fang
Zahid Khan
Pingzhi Fan
author_sort Jiahao Shan
collection DOAJ
description Anomaly detection of multivariate time series (MTS) is crucial in industrial intelligent systems. To address the challenges of absence of anomaly labels, fast inference time, multi-source and multi-modality in anomaly detection, researchers have primarily investigated unsupervised reconstruction-driven methods. However, the existing reconstruction-driven methods mainly focus on minimizing reconstruction errors while neglecting the importance of training methods that increase errors between normal and abnormal classes. Furthermore, accurately constructing the feature space of normal and abnormal classes during the reconstruction process remains a challenge. In this paper, we propose an innovative model, namely the confidence adversarial autoencoder (CAAE). The proposed CAAE combines a confidence network, based on window credibility judgment, with an autoencoder to provide credibility support for anomaly detection. We further introduce fake labels to provide the confidence network with a discriminative knowledge for identifying reconstructed data. Additionally, we implement the confidence adversarial training method to generate fake labels to construct an adversarial loss aiming to expand the decision boundary of anomaly scores. Extensive experimental results on publicly available time series datasets are provided to demonstrate the efficiency of our proposed CAAE. It reveals that excellent generalization ability and superior average performance are achieved on different datasets compared with the state-of-the-art methods.
format Article
id doaj-art-ba9222edf0de4d18a9ad7c8325cdcf77
institution OA Journals
issn 2644-125X
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Open Journal of the Communications Society
spelling doaj-art-ba9222edf0de4d18a9ad7c8325cdcf772025-08-20T01:58:15ZengIEEEIEEE Open Journal of the Communications Society2644-125X2024-01-0157752776610.1109/OJCOMS.2024.351195110778252Unsupervised Multivariate Time Series Data Anomaly Detection in Industrial IoT: A Confidence Adversarial Autoencoder NetworkJiahao Shan0Donghong Cai1https://orcid.org/0000-0001-5350-135XFang Fang2https://orcid.org/0000-0002-6582-6570Zahid Khan3https://orcid.org/0000-0003-4710-4010Pingzhi Fan4https://orcid.org/0000-0002-8281-6251College of Cyber Security, Jinan University, Guangzhou, ChinaCollege of Cyber Security, Jinan University, Guangzhou, ChinaDepartment of Electrical and Computer Engineering, Western University, London, ON, CanadaCollege of Computer and Information Sciences, Prince Sultan University, Riyadh, Saudi ArabiaKey Laboratory of Information Coding and Transmission, Southwest Jiaotong University, Chengdu, ChinaAnomaly detection of multivariate time series (MTS) is crucial in industrial intelligent systems. To address the challenges of absence of anomaly labels, fast inference time, multi-source and multi-modality in anomaly detection, researchers have primarily investigated unsupervised reconstruction-driven methods. However, the existing reconstruction-driven methods mainly focus on minimizing reconstruction errors while neglecting the importance of training methods that increase errors between normal and abnormal classes. Furthermore, accurately constructing the feature space of normal and abnormal classes during the reconstruction process remains a challenge. In this paper, we propose an innovative model, namely the confidence adversarial autoencoder (CAAE). The proposed CAAE combines a confidence network, based on window credibility judgment, with an autoencoder to provide credibility support for anomaly detection. We further introduce fake labels to provide the confidence network with a discriminative knowledge for identifying reconstructed data. Additionally, we implement the confidence adversarial training method to generate fake labels to construct an adversarial loss aiming to expand the decision boundary of anomaly scores. Extensive experimental results on publicly available time series datasets are provided to demonstrate the efficiency of our proposed CAAE. It reveals that excellent generalization ability and superior average performance are achieved on different datasets compared with the state-of-the-art methods.https://ieeexplore.ieee.org/document/10778252/Multivariate time series (MTS)anomaly detectionadversarial trainingunsupervised learningautoencoder
spellingShingle Jiahao Shan
Donghong Cai
Fang Fang
Zahid Khan
Pingzhi Fan
Unsupervised Multivariate Time Series Data Anomaly Detection in Industrial IoT: A Confidence Adversarial Autoencoder Network
IEEE Open Journal of the Communications Society
Multivariate time series (MTS)
anomaly detection
adversarial training
unsupervised learning
autoencoder
title Unsupervised Multivariate Time Series Data Anomaly Detection in Industrial IoT: A Confidence Adversarial Autoencoder Network
title_full Unsupervised Multivariate Time Series Data Anomaly Detection in Industrial IoT: A Confidence Adversarial Autoencoder Network
title_fullStr Unsupervised Multivariate Time Series Data Anomaly Detection in Industrial IoT: A Confidence Adversarial Autoencoder Network
title_full_unstemmed Unsupervised Multivariate Time Series Data Anomaly Detection in Industrial IoT: A Confidence Adversarial Autoencoder Network
title_short Unsupervised Multivariate Time Series Data Anomaly Detection in Industrial IoT: A Confidence Adversarial Autoencoder Network
title_sort unsupervised multivariate time series data anomaly detection in industrial iot a confidence adversarial autoencoder network
topic Multivariate time series (MTS)
anomaly detection
adversarial training
unsupervised learning
autoencoder
url https://ieeexplore.ieee.org/document/10778252/
work_keys_str_mv AT jiahaoshan unsupervisedmultivariatetimeseriesdataanomalydetectioninindustrialiotaconfidenceadversarialautoencodernetwork
AT donghongcai unsupervisedmultivariatetimeseriesdataanomalydetectioninindustrialiotaconfidenceadversarialautoencodernetwork
AT fangfang unsupervisedmultivariatetimeseriesdataanomalydetectioninindustrialiotaconfidenceadversarialautoencodernetwork
AT zahidkhan unsupervisedmultivariatetimeseriesdataanomalydetectioninindustrialiotaconfidenceadversarialautoencodernetwork
AT pingzhifan unsupervisedmultivariatetimeseriesdataanomalydetectioninindustrialiotaconfidenceadversarialautoencodernetwork