FROBENIUS-PERRON GRAPHS FOR TEMPORAL DATA CLASSIFICATION1

Barend J. van Wyk and Michaël A. van Wyk

Keywords

Temporal reasoning, modelbased reasoning, self organizing feature map, graphical models, time series classification

Abstract

In this paper Frobenius-Perron Graphs (FPGs), based on the Frobenius-Perron operator, are introduced. An algorithm for the derivation of FPGs and a method to use FPGs for the classification of time series data are presented. Simulation results indicate that the FPG method is not only robust, but also easy to implement. The FPG approach is compared to statistical feature and power spectral density approaches and an approach based on a modification of the normalized cross correlation.

Important Links:



Go Back