Change Point Detection in Time Series Data using Autoencoders with a Time-Invariant Representation

TL;DR

This paper introduces a novel autoencoder-based change point detection method that learns a time-invariant representation, improving detection of subtle changes in time series data while reducing false alarms.

Contribution

It proposes a new autoencoder loss function and postprocessing technique for more accurate and flexible change point detection in various domains.

Findings

Outperforms baseline methods on simulated and real datasets

Detects changes in slope, mean, variance, autocorrelation, and frequency spectrum

Reduces false alarms with a novel postprocessing procedure

Abstract

Change point detection (CPD) aims to locate abrupt property changes in time series data. Recent CPD methods demonstrated the potential of using deep learning techniques, but often lack the ability to identify more subtle changes in the autocorrelation statistics of the signal and suffer from a high false alarm rate. To address these issues, we employ an autoencoder-based methodology with a novel loss function, through which the used autoencoders learn a partially time-invariant representation that is tailored for CPD. The result is a flexible method that allows the user to indicate whether change points should be sought in the time domain, frequency domain or both. Detectable change points include abrupt changes in the slope, mean, variance, autocorrelation function and frequency spectrum. We demonstrate that our proposed method is consistently highly competitive or superior to baseline methods on diverse simulated and real-life benchmark data sets. Finally, we mitigate the issue of false detection alarms through the use of a postprocessing procedure that combines a matched filter and a newly proposed change point score. We show that this combination drastically improves the performance of our method as well as all baseline methods.