AUTONOMOUS GENETIC ALGORITHM FOR FUNCTIONAL OPTIMIZATION

Genetic algorithm (GA) is effective for global optimiza- tions, but needs the user to define several parameters. Unless these parameters are defined appropriately, search efficiency drops signifi- cantly. There are, however, no clear rules for the defining, and almost all users have considerable difficulty to use GA efficiently. A good al- gorithm must be use-friendly. It should not, if possible, need the user to define such parameters and can play high performance for any opti- mization problem. This paper proposes an autonomous GA addressing these problems.