Forecast-based Multi-aspect Framework for Multivariate Time-series Anomaly Detection

被引:1
|
作者
Wang, Lan [1 ]
Lin, Yusan [1 ]
Wu, Yuhang [1 ]
Chen, Huiyuan [1 ]
Wang, Fei [1 ]
Yang, Hao [1 ]
机构
[1] Visa Res, Palo Alto, CA 94306 USA
关键词
Anomaly Detection; Multivariate Time Series; Unsupervised Learning;
D O I
10.1109/BigData52589.2021.9671776
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Today's cyber-world is vastly multivariate. Metrics collected at extreme varieties demand multivariate algorithms to properly detect anomalies. However, forecast-based algorithms, as widely proven approaches, often perform sub-optimally or inconsistently across datasets. A key common issue is they strive to be one-size-fits-all but anomalies are distinctive in nature. We propose a method that tailors to such distinction. Presenting FMUAD - a Forecast-based, Multi-aspect, Unsupervised Anomaly Detection framework. FMUAD explicitly and separately captures the signature traits of anomaly types - spatial change, temporal change and correlation change - with independent modules. The modules then jointly learn an optimal feature representation, which is highly flexible and intuitive, unlike most other models in the category. Extensive experiments show our FMUAD framework consistently outperforms other state-of-the-art forecast-based anomaly detectors.
引用
收藏
页码:938 / 947
页数:10
相关论文
共 50 条
  • [1] Adaptive Multivariate Time-Series Anomaly Detection
    Lv, Jianming
    Wang, Yaquan
    Chen, Shengjing
    INFORMATION PROCESSING & MANAGEMENT, 2023, 60 (04)
  • [2] Multi-Aspect Edge Device Association Based on Time-Series Dynamic Interaction Networks
    Yang, Xiaoteng
    Feng, Jie
    Song, Xifei
    Xu, Feng
    Liu, Yuan
    Pei, Qingqi
    Wu, Celimuge
    IEEE INFOCOM 2024-IEEE CONFERENCE ON COMPUTER COMMUNICATIONS WORKSHOPS, INFOCOM WKSHPS 2024, 2024,
  • [3] Unsupervised Multi-head Attention Autoencoder for Multivariate Time-Series Anomaly Detection
    Kim, Minseok
    Park, Sanghyun
    2024 IEEE INTERNATIONAL CONFERENCE ON BIG DATA AND SMART COMPUTING, IEEE BIGCOMP 2024, 2024, : 1 - 7
  • [4] Time-Series Forecast-Based Endpoint Thickness Compensation for Thinning of Sapphire Wafer
    Lin, Yu-Kun
    Wu, Bing-Fei
    IEEE ACCESS, 2023, 11 : 21252 - 21263
  • [5] Anomaly Detection from Multivariate Time-Series with Sparse Representation
    Takeishi, Naoya
    Yairi, Takehisa
    2014 IEEE INTERNATIONAL CONFERENCE ON SYSTEMS, MAN AND CYBERNETICS (SMC), 2014, : 2651 - 2656
  • [6] Real-Time Deep Anomaly Detection Framework for Multivariate Time-Series Data in Industrial IoT
    Nizam, Hussain
    Zafar, Samra
    Lv, Zefeng
    Wang, Fan
    Hu, Xiaopeng
    IEEE SENSORS JOURNAL, 2022, 22 (23) : 22836 - 22849
  • [7] MSTREAM: Fast Anomaly Detection in Multi-Aspect Streams
    Bhatia, Siddharth
    Jain, Arjit
    Li, Pan
    Kumar, Ritesh
    Hooi, Bryan
    PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE 2021 (WWW 2021), 2021, : 3371 - 3382
  • [8] AttVAE: A Novel Anomaly Detection Framework for Multivariate Time Series
    Liu, Yi
    Han, Yanni
    An, Wei
    SCIENCE OF CYBER SECURITY, SCISEC 2022, 2022, 13580 : 407 - 420
  • [9] AURORA: A Unified fRamework fOR Anomaly detection on multivariate time series
    Lin Zhang
    Wenyu Zhang
    Maxwell J. McNeil
    Nachuan Chengwang
    David S. Matteson
    Petko Bogdanov
    Data Mining and Knowledge Discovery, 2021, 35 : 1882 - 1905
  • [10] AURORA: A Unified fRamework fOR Anomaly detection on multivariate time series
    Zhang, Lin
    Zhang, Wenyu
    McNeil, Maxwell J.
    Chengwang, Nachuan
    Matteson, David S.
    Bogdanov, Petko
    DATA MINING AND KNOWLEDGE DISCOVERY, 2021, 35 (05) : 1882 - 1905