Parameter Identification in a Predator-Prey System using Persistent Homology

Abstract

The present work uses persistent homology combined with machine learning to identify (classify) parameters of system of equations producing complex patterns. Persistent homology is used as a tool to extract topological information from the patterns. This topological information is in turn used as features for the machine learning methods used for the classification. The method is applied to patterns generated by a predator-prey system using the SVM, PLS-DA, and the Naive Bayes machine learning methods.