Pseudo-hill climbing genetic algorithm (PHGA) for function optimization
Masafumi Hagiwara · 2005
In general, one of the shortcomings in GAs as search methods is their lack of local search ability. The main objective of this paper is to combine the ideas of simplex method with the genetic algorithms (GAs). In order to give a hill-climbing ability to the conventional GAs, like neural networks, we propose PHGA for function optimization. Computer simulation results using De Jong's five-function test bed (1975) are shown. According to our simulation, all of the results by the proposed PHGA are better than those by the conventional GAs.