The Effect of Mutation on Explorative & Exploitative Behaviours with rt-NEAT

Abstract

This thesis aims to explore how different factors can affect the search performance of evolutionary algorithms. Additionally how applying different mutation techniques changes the overall search performance of rtNEAT. This thesis demonstrates how mutation affects exploration and exploitation when optimizing for a 3-input XOR gate as well as optimizing agent movements in a real time environment. This thesis is also provided as a guideline in the development of an evolutionary algorithm, particularly the implementation of rtNEAT algorithm, and a simple game environment in Python.