Session:

EEC - Evolvability in Evolutionary Computation

Title:

Chemical Genetic Algorithms: Enhancing Evolvability of GAs via Genotype-Phenotype Mapping

Authors:

Hidefumi Sawai
Hideaki Suzuki
Wojciech Piaseczny

Abstract:

We propose the concept of a chemical genetic algorithm (CGA), in which several types of molecules (information units) react with each other in a cell. Not only the information in DNA but also that of the smaller molecules responsible for the transcription and translation of DNA into amino acids is changed adaptively during evolution. This mechanism optimizes the fundamental mapping from binary substrings in DNA (genotypes) to real values for a parameter set (phenotypes). Through the struggle between cells containing a DNA unit and small molecular units, the codes (DNA) and the interpreter (the small molecular units) co-evolve, and a specific output function, from which a cell's fitness is evaluated, is optimized. To demonstrate the effectiveness of the CGA, it is applied to some problems including a set of deceptive problems and benchmark problems such as Shekel's foxholes and a generalized Langermann's function. To ascertain the validity of the genotype-phenotype mapping by the CGA, some analytical experiments were conducted while observing the basin size of a global optimum solution in the binary genotypic space. The results show that the CGA effectively broadens the basin size, making it easier to find a path to a global optimum solution, while enhancing the GA's evolvability during evolution.