Fisher Information Framework for Time Series Modeling

A robust prediction model invoking the Takens embedding theorem, whose \textit{working hypothesis} is obtained via an inference procedure based on the minimum Fisher information principle, is presented. The coefficients of the ansatz, central to the \textit{working hypothesis} satisfy a time independent Schrödinger-like equation in a vector setting. The inference of i) the probability density function of the coefficients of the \textit{working hypothesis} and ii) the establishing of constraint driven pseudo-inverse condition for the modeling phase of the prediction scheme, is made, for the case of normal distributions, with the aid of the quantum mechanical virial theorem. The well-known reciprocity relations and the associated Legendre transform structure for the Fisher information measure (FIM, hereafter)-based model in a vector setting (with least square constraints) are self-consistently derived. These relations are demonstrated to yield an intriguing form of the FIM for the modeling phase, which defines the \textit{working hypothesis}, solely in terms of the observed data. Cases for prediction employing time series' obtained from the: $(i)$ the Mackey-Glass delay-differential equation, $(ii)$ one ECG sample from the MIT-Beth Israel Deaconess Hospital (MIT-BIH) cardiac arrhythmia database, and $(iii)$ one ECG from the Creighton University ventricular tachyarrhythmia database. The ECG samples were obtained from the Physionet online repository. These examples demonstrate the efficiency of the prediction model. Numerical examples for exemplary cases are provided.