Difference-Based Mutation Operation for Neuroevolution of Augmented Topologies

In this paper, a novel search operation is proposed for the neuroevolution of augmented topologies, namely difference-based mutation. This operator uses differences between individuals in population to perform more efficient optimal weights and structure model. The difference determined according innovation numbers assigned each node connection, allowing tracking changes. implemented algorithm allows backward connections loops topology, set mutation operators, including merging deletion. tested on classification problems rotary inverted pendulum control problem. comparison performed basic approach modified versions. sensitivity parameter values examined. experimental results prove that newly developed delivers significant improvements quality several cases, allow finding better algorithms.