Parameter free determination of optimum time delay

TL;DR

This paper introduces a parameter-free method based on the maximum entropy principle to determine the optimal time delay for sampling data, improving data collection and analysis in time series applications.

Contribution

It presents a novel, parameter-free approach for calculating the optimal time delay using maximum entropy, eliminating the need for traditional parameter tuning.

Findings

Matches results of traditional methods

Independent of free parameters

Suitable for AI and data acquisition processes

Abstract

We show that the same maximum entropy principle applied to recurrence microstates configures a new way to properly compute the time delay necessary to correctly sample a data set. The new method retrieves results obtained using traditional methods with the advantage of being independent of any free parameter. Since all parameters are automatically set, the method is suitable for use in artificial (computational) intelligence algorithms, recovering correct information embedded in time series, and rationalizing the process of data acquisition since only relevant data must be collected.