A new evolutionary model based on family eugenics: the first results
Shaoyan Wu, Qingfu Zhang, Huowang Chen · 2002
Several popular approaches of simulated evolution have been developed. The approaches emphasize different facets of natural evolutionary processes. One has recognized that simulated evolution will benefit from an adequate combination of approaches. The paper views the primary difference between existing approaches as the difference between simulations of genetic link and behavioral link. We propose a new model of simulated evolution, called family eugenics based evolution (FEBE), which combines the genetic link with the behavioral link using the idea of family eugenics. In the FEBE model the orthogonal design technique is introduced into offspring breeding inside a family so as to enhance the behavioral improvement of individuals. We investigate an application of the FEBE model to Goldberg's (1989) deceptive problem that is challenging for most evolutionary algorithms. Exciting experimental results have been achieved.