118 Genetic Algorithm Using self-Organizing Maps

Abstract

This paper describes Self-Organizing Maps for Genetic Algorithms (SOM-GA), which is the combinational algorithm of Genetic Algorithms (GA) and Self-Organizing Maps (SOM). In the algorithm, the whole population is divided into sub-populations by using SOM clustering. Real-coded genetic algorithm (RCGA) is applied in the sub-populations. The algorithm is applied to the solution search of Rastrigin function. Comparing SOM-GA with RCGA, we notice that the present algorithm has much better search performance than the RCGA. Besides, the discussion on the map-size of SOM indicates that the map-size affects the search performance and the CPU time.