Detecting dynamical changes in time series by using the Jensen Shannon Divergence

Abstract

Most of the time series in nature are a mixture of signals with deterministic and random dynamics. Thus the distinction between these two characteristics becomes important. Distinguishing between chaotic and aleatory signals is difficult because they have a common wide-band power spectrum, a delta-like autocorrelation function, and share other features as well. In general signals are presented as continuous records and require to be discretized for being analyzed. In this work we present different schemes for discretizing and for detection of dynamical changes in time series. One of the main motivations is to detect transition from chaotic regime to random regime. The tools used are originated in Information Theory. The schemes proposed are applied to simulated and real life signals, showing in all cases a high proficiency for detecting changes in the dynamics of the associated time series.

14 pages, 7 figures

Cited by in corpus (6)

• Permutation Jensen-Shannon distance: A versatile and fast symbolic tool for complex time series analysis
• Monoparametric family of metrics derived from classical Jensen-Shannon divergence
• Application of the third RIT binary black hole simulations catalog to parameter estimation of gravitational waves signals from the LIGO-Virgo O1/O2 observational runs
• Vela Pulsar: Single Pulses Analysis with Machine Learning Techniques
• Quantifying the irregularity of a time series
• Detecting Stochasticity in Discrete Signals via Nonparametric Excursion Theorem