Applying Recurrence Plots to Classify Time Series

Abstract

The article describes a new approach to the classification of time series based on the construction of their recurrence plots. After transforming the time series into recurrence plots, two approaches are applied for classification. In the first case, quantitative recurrence characteristics are used for classification as features. In the second case, the time series is presented in the form of a black and white image of its recurrence plot. A convolutional neural network is used as an image classifier. The data for the classification are the electrocardiograms realizations of 100 values, which contained records of healthy people and patients with a diagnosis of ischemia. Research results showed the advantages of classifying images of recurrence plots, indicate a good classification accuracy in comparison with other methods and the potential capabilities of this approach.

Description

Keywords

Time series classification, machine learning classification, recurrence plot, ECG time series, quantitative recurrence characteristic

Citation

Kirichenko L. Applying Recurrence Plots to Classify Time Series / Lyudmyla Kirichenko, Tamara Radivilova, Juliia Stepanenko // Computational linguistics and intelligent systems, 22-23 April 2021, Kharkiv. — Lviv ; Kharkiv, 2021. — Vol Vol. II : Proceedings of the 5th International conference, COLINS 2021, Workshop, Kharkiv, Ukraine, April 22-23. — P. 16–26.

Endorsement

Review

Supplemented By

Referenced By