The parameter-free genetic algorithm (PfGA) is attracting interest as a new approach to avoiding the difficulty of parameter setting which has been a problem in genetic algorithms. However, there is a limit to the speed of evolution, since only random N-point crossover and mutation are used to modify the genes. On the other hand, various gene modification procedures have been attempted. Among these, there are some in which parameter setting is entirely or largely unnecessary. It may be useful to combine these with PfGA. This paper proposes a parameter-free genetic algorithm combined with the pseudo-simplex method, in an attempt to improve the power of optimal solution search by the pseudo-simplex method, which is known to be an effective method of improving the speed of genetic algorithms. The proposed algorithm uses the pseudo-simplex method in the process of generating the child individuals in the PfGA, and also in preprocessing to extract parent individuals from the subpopulation, in an attempt to improve the speed of evolution of the subpopulation. The proposed algorithm is compared to other methods in application to a function optimization problem, and it is verified that its performance for various types of functions is greatly superior to that of the ordinary PfGA, thus demonstrating its usefulness.
ASJC Scopus subject areas
- Theoretical Computer Science
- Information Systems
- Hardware and Architecture
- Computational Theory and Mathematics