Date of Award

2012

First Advisor

Paul Shields

Second Advisor

Brian Wynne

Abstract

NEAT (NeuroEvolution of Augmenting Topologies) is a genetic algorithm (GA) for artificial neural network (ANN) training. Evolutionary algorithms like NEAT can be used to study evolutionary behaviors. This thesis examines the effects of applying NEAT to the simulation of birds, controlled by ANNs, in a predator-prey environment. The birds are given only a motivation to survive, with the intention of observing whether or not more complex behaviors can emerge. The results positively indicated the emergence of complex behaviors as a symptom of survival.

Share

COinS