Real-time Change-Point Detection: A deep neural network-based adaptive approach for detecting changes in multivariate time series data

作者：

Highlights：

• End-to-end adaptive and pipelined change point detection has been proposed.

• The model uses deep adaptive input normalization for real-time analysis.

• The recursive version of Singular Spectrum Analysis has been developed for outlier removal.

• A deep learning-based autoencoder is used for quick detection of change points.

摘要

•End-to-end adaptive and pipelined change point detection has been proposed.•The model uses deep adaptive input normalization for real-time analysis.•The recursive version of Singular Spectrum Analysis has been developed for outlier removal.•A deep learning-based autoencoder is used for quick detection of change points.

论文关键词：Adaptive Change-Point Detection,Data normalization,Deep learning,Recursive singular spectrum analysis,Lagrangian multiplier,Overcomplete autoencoder

论文评审过程：Received 15 February 2022, Revised 23 June 2022, Accepted 20 July 2022, Available online 26 July 2022, Version of Record 9 August 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118260