Anomaly Detection for Univariate Time Series with Statistics and Deep Learning

Jian Bin Kao; Jehn Ruey Jiang

doi:10.1109/ECICE47484.2019.8942727

Anomaly Detection for Univariate Time Series with Statistics and Deep Learning

Jian Bin Kao, Jehn Ruey Jiang

Department of Computer Science and Information Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

28 Scopus citations

Abstract

This paper proposes an anomaly detection framework for univariate time series data. Based on the Dickey-Fuller test, fast Fourier transform (FFT), and Pearson product-moment correlation coefficient, data are classified into three classes, namely (1) stationary, (2) periodic and (3) non-stationary and non-periodic time series. Different schemes using statistics and gated recurrent unit (GRU) deep learning concepts are applied to different class of time series for performing anomaly detection. The proposed framework outperforms related methods, namely STL, SARIMA, LSTM, LSTM with STL, and ADSaS, in almost all measurements for five Numenta Anomaly Benchmark(NAB) datasets.

Original language	English
Title of host publication	2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019
Editors	Teen-Hang Meen
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	404-407
Number of pages	4
ISBN (Electronic)	9781728125015
DOIs	https://doi.org/10.1109/ECICE47484.2019.8942727
State	Published - Oct 2019
Event	2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019 - Yunlin, Taiwan Duration: 3 Oct 2019 → 6 Oct 2019

Publication series

Name	2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019

Conference

Conference	2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019
Country/Territory	Taiwan
City	Yunlin
Period	3/10/19 → 6/10/19

Keywords

anomaly detection
deep learning
gated recurrent unit neural network
time series

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/ECICE47484.2019.8942727

Cite this

Kao, J. B., & Jiang, J. R. (2019). Anomaly Detection for Univariate Time Series with Statistics and Deep Learning. In T.-H. Meen (Ed.), 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019 (pp. 404-407). Article 8942727 (2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ECICE47484.2019.8942727

Kao, Jian Bin ; Jiang, Jehn Ruey. / Anomaly Detection for Univariate Time Series with Statistics and Deep Learning. 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019. editor / Teen-Hang Meen. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 404-407 (2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019).

@inproceedings{6dbac46a7cf14a47995163af87209fba,

title = "Anomaly Detection for Univariate Time Series with Statistics and Deep Learning",

abstract = "This paper proposes an anomaly detection framework for univariate time series data. Based on the Dickey-Fuller test, fast Fourier transform (FFT), and Pearson product-moment correlation coefficient, data are classified into three classes, namely (1) stationary, (2) periodic and (3) non-stationary and non-periodic time series. Different schemes using statistics and gated recurrent unit (GRU) deep learning concepts are applied to different class of time series for performing anomaly detection. The proposed framework outperforms related methods, namely STL, SARIMA, LSTM, LSTM with STL, and ADSaS, in almost all measurements for five Numenta Anomaly Benchmark(NAB) datasets.",

keywords = "anomaly detection, deep learning, gated recurrent unit neural network, time series",

author = "Kao, {Jian Bin} and Jiang, {Jehn Ruey}",

note = "Publisher Copyright: {\textcopyright} 2019 IEEE.; 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019 ; Conference date: 03-10-2019 Through 06-10-2019",

year = "2019",

month = oct,

doi = "10.1109/ECICE47484.2019.8942727",

language = "???core.languages.en_GB???",

series = "2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "404--407",

editor = "Teen-Hang Meen",

booktitle = "2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019",

}

Kao, JB & Jiang, JR 2019, Anomaly Detection for Univariate Time Series with Statistics and Deep Learning. in T-H Meen (ed.), 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019., 8942727, 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019, Institute of Electrical and Electronics Engineers Inc., pp. 404-407, 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019, Yunlin, Taiwan, 3/10/19. https://doi.org/10.1109/ECICE47484.2019.8942727

Anomaly Detection for Univariate Time Series with Statistics and Deep Learning. / Kao, Jian Bin; Jiang, Jehn Ruey.
2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019. ed. / Teen-Hang Meen. Institute of Electrical and Electronics Engineers Inc., 2019. p. 404-407 8942727 (2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Anomaly Detection for Univariate Time Series with Statistics and Deep Learning

AU - Kao, Jian Bin

AU - Jiang, Jehn Ruey

PY - 2019/10

Y1 - 2019/10

N2 - This paper proposes an anomaly detection framework for univariate time series data. Based on the Dickey-Fuller test, fast Fourier transform (FFT), and Pearson product-moment correlation coefficient, data are classified into three classes, namely (1) stationary, (2) periodic and (3) non-stationary and non-periodic time series. Different schemes using statistics and gated recurrent unit (GRU) deep learning concepts are applied to different class of time series for performing anomaly detection. The proposed framework outperforms related methods, namely STL, SARIMA, LSTM, LSTM with STL, and ADSaS, in almost all measurements for five Numenta Anomaly Benchmark(NAB) datasets.

AB - This paper proposes an anomaly detection framework for univariate time series data. Based on the Dickey-Fuller test, fast Fourier transform (FFT), and Pearson product-moment correlation coefficient, data are classified into three classes, namely (1) stationary, (2) periodic and (3) non-stationary and non-periodic time series. Different schemes using statistics and gated recurrent unit (GRU) deep learning concepts are applied to different class of time series for performing anomaly detection. The proposed framework outperforms related methods, namely STL, SARIMA, LSTM, LSTM with STL, and ADSaS, in almost all measurements for five Numenta Anomaly Benchmark(NAB) datasets.

KW - anomaly detection

KW - deep learning

KW - gated recurrent unit neural network

KW - time series

UR - http://www.scopus.com/inward/record.url?scp=85078575736&partnerID=8YFLogxK

U2 - 10.1109/ECICE47484.2019.8942727

DO - 10.1109/ECICE47484.2019.8942727

M3 - 會議論文篇章

AN - SCOPUS:85078575736

T3 - 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019

SP - 404

EP - 407

BT - 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019

A2 - Meen, Teen-Hang

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019

Y2 - 3 October 2019 through 6 October 2019

ER -

Kao JB, Jiang JR. Anomaly Detection for Univariate Time Series with Statistics and Deep Learning. In Meen TH, editor, 2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019. Institute of Electrical and Electronics Engineers Inc. 2019. p. 404-407. 8942727. (2019 IEEE Eurasia Conference on IOT, Communication and Engineering, ECICE 2019). doi: 10.1109/ECICE47484.2019.8942727

Anomaly Detection for Univariate Time Series with Statistics and Deep Learning

Abstract

Publication series

Conference

Keywords

UN SDGs

Access to Document

Fingerprint

Cite this