Hierarchical Crossover in Genetic Algorithms
Peter J. Bentley, Jonathan P. Wakefield · University of Huddersfield Repository (University of Huddersfield) · 1996
This paper identifies the limitations of conventional crossover in genetic algorithms when operating on two chromosomes of differing lengths. To address these problems, the concept of a Semantic Hierarchy (i.e. tree of meaning) of a genotype within a genetic algorithm is introduced. With this in mind, a new form of crossover operator known as Hierarchical Crossover is presented, capable of performing crossover with genotypes of different sizes, while still being functionally equivalent to standard, single-point (uniform) crossover. Various aspects and advantages of this method are discussed. Finally, an example of some results produced by an implementation is shown. Key words: hierarchical crossover, semantic hierarchy, variable-length chromosome, genetic algorithm, evolutionary design. 1. Introduction The genetic algorithm (GA) is a highly efficient and robust search algorithm based on evolution in nature (Holland, 1975). Today, GAs are widely used to evolve good solutions to hundre...