Optimization of a subset of features based on fuzzy genetic algorithm

Abstract

To overcome global situation problem tradition genetic algorithm has very strong robustness in finding the solution, but crossover probability and mutation probability is fixed and invariable, it caused premature convergence and running inefficient to the solution on complicated problem at later evolution process of tradition genetic algorithm. To this problem the paper proposed a new algorithm with varying population size based on lifetimes of the chromosomes to realize population size adjust adaptively and crossover probability adjust adaptively and mutation probability adjust adaptively, which called fuzzy genetic algorithm. Compare to tradition genetic algorithm, experiment results show that the approach proposed is effective in the capability of global optimization and significantly improves the convergence rate.