Evolutionary multi-objective optimization of robustness and innovation in redundant genetic representations

Robustness and innovation are two essential facets for biological evolution, where robustness means the relative insensitivity of an organism's phenotype to mutations, while innovation (evolvability) denotes the individual's ability to evolve novel phenotypes that help its survival and reproduction. Although much research has been conducted on robustness and evolvability of both biological and computational evolutionary systems, little work on the quantitative analysis of the relationship between robustness and evolvability has been reported. In this work, a measure for innovation called local variability has been suggested. Based on a neutrality degree borrowed from literature [1] and local variability, a multi-objective evolutionary algorithm has been employed to maximize the robustness and innovation by optimizing the genotype-phenotype mapping of the redundant representation. The obtained Pareto-optimal solutions are then analyzed to reveal the trade-off relationship between robustness and innovation of the redundant representation.