Anomaly detection for time series by using the autoencoder: An empirical study
Ro Jin Pak 1
1Department of Information Statistics, Dankook University
Correspondence to: 1 Professor, Department of Applied Statistics, Dankook University,Gyeonggido 16890, Korea. E-mail: rjpak@dankook.ac.kr
Received June 18, 2023; Revised July 4, 2023; Accepted July 10, 2023.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
- Abstract
- We tried to use an autoencoder to search for abnormal data or outliers, maybe presented in the time series. The autoencoder is a type of artificial neural network and its performance in detecting outliers has been verified by many researchers. In this study, we tried to apply the autoencoder to time series analysis to detect outliers unlike the existing methods by continuously learning the network along the timeline. We selected two time series examples to verify the performance of the proposed method and it was discovered that the proposed method enabled us to detect some abnormal time points in time series.
- Keywords : Abnormal data, artificial neural network, autoencoder, outlier detection, time series
Full Text(PDF) Free
