Family genetic algorithms based on gene exchange and its application

1. INTRODUCTION Genetic algorithms (GA) are a search techniques bas-ed on mechanics of nature selection and were initiallyproposed by Holland1]. GA have already beensuccessfully applied in many diverse areas, such asfunction optimization, the traveling salesman proble-ms, multiobje-ctive optimization problems, schedulingneural network design, system identification, visioncontrol and machine learning. A detailed review ofthese applications is provided in Refs. 2] and 3]. The strength …

Family genetic algorithms based on gene exchange and its application

Genetic Algorithms (GA) are a search techniques based on mechanics of nature selection and have already been successfully applied in many diverse areas. However, increasing samples show that GA's performance is not as good as it was expected to be. Criticism of this algorithm includes the slow speed and premature result during convergence procedure. In order to improve the performance, the population size and individuals' space is emphatically described. The influence of individuals' space and population size on the operators is analyzed. And a novel family genetic algorithm (FGA) is put forward based on this analysis. In this novel algorithm, the optimum solution families closed to quality individuals is constructed, which is exchanged found by a search in the world space. Search will be done in this microspace. The family that can search better genes in a limited period of time would win a new life. At the same time, the best gene of this micro space with the basic population in the world space is exchanged. Finally, the FGA is applied to the function optimization and image matching through several experiments. The results show that the FGA possessed high performance.