Achieving High-Level Functionality through Complexification
Kenneth Owen Stanley, Risto P Miikkulainen · 2003
An appropriate but challenging goal for evolutionary com-putation (EC) is to evolve systems of biological com-plexity. However, specifying complex structures requires many genes, and searching for a solution in such a high-dimensional space can be intractable. In this paper, we pro-pose a method for finding high-dimensional solutions in-crementally, by starting with an initial population of very small genomes and gradually complexifying those genomes by adding new genes over generations. That way, search begins in an easily-optimized low-dimensional space and in-crements into increasingly high-dimensional spaces. We de-scribe an existing method for implementing complexification, and further propose that combining complexification with an indirect genetic encoding, in which genes are reused in the specification of the phenotype, can lead to the discovery of highly complex solutions.