Evolutionary Algorithms Using Artificial Embryogeny

Abstract

In evolutionary computation, direct encoding methods have been commonly used for genotype-phenotype mapping. It is difficult to evolve very complex phenotypes with these methods because such phenotypes require very long chromosomes, which cannot be evolved efficiently. To be able to evolve more complex solutions, approaches called artificial embryogeny (AE) have been proposed employing indirect encoding that is inspired from development of embryos in nature. Here we demonstrate the emergence of gene reuse as a result of introduction of developmental encodings, showing the possibility of AE systems.