Reproduction strategy based on self-organizing map for genetic algorithms

A novel reproduction strategy by employing a Self-Organizing Map (SOM) for two types of Genetic Algorithms (GAs) is proposed to maintain genetic diversity of population. In the proposed reproduction strategy, a set of new chromosomes in the next generation is decided by a learning of the SOM with modified updating equation based on fitness values. The approximation ability of the SOM facilitates the preservation of the genetic diversity. The proposed reproduction strategy can be applied to "Bit-String GA" and "Real-Coded GA" by employing the SOM with real value weight vectors and binary weight vectors, respectively.