Parent Selection Strategies in Niching Genetic Algorithms

Abstract

In this work, we analyze the effect of three parent selection strategies in a niching genetic algorithm: binary tournament, most similar and most dissimilar parent selection. These strategies were implemented in a steady-state genetic algorithm (GA) that uses the Restricted Tournament Selection (RTS) method for niching formation. The implemented strategies were compared to the random parent selection used by the standard RTS method and tested on the 20 functions of the CEC 2013 benchmark set of the Competition on Niching Methods for Multi-modal Optimization. The results indicate the most similar parent selection as the best strategy, mainly concerning high dimensional functions. Moreover, the proposed GA-RTS algorithm with the most similar parent selection scheme proved to be competitive with other state-of-the-art niching algorithms.