Hybridization of local search with self-configuring genetic programming algorithm for automated fuzzy classifier design

A fuzzy classifier is one of the intelligent information technologies allowing the generation of a fuzzy rule base suitable for interpretation by human experts. For a fuzzy classifier automated design the hybrid self-configuring evolutionary algorithm is proposed. The self-configuring genetic programming algorithm is suggested for the choice of effective fuzzy rule bases. For the tuning of linguistic variables the self-configuring genetic algorithm is used. A hybridization of self-configuring genetic programming algorithms (SelfCGPs) with a local search in the space of trees is fulfilled to improve their performance for fuzzy rule bases automated design. The local search is implemented with two neighborhood systems (1-level and 2-level neighborhoods), three strategies of a tree scanning (“full”, “incomplete” and “truncated”) and two ways of a movement between adjacent trees (transition by the first improvement and the steepest descent). The Lamarckian local search is applied on each generation to ten percent of best individuals. The performance of all developed memetic algorithms is estimated on a representative set of test problems of the functions approximation as well as on real-world classification problems. It is shown that developed memetic algorithm requires comparable amount of computational efforts but outperforms the original SelfCGP for the fuzzy rule bases automated design. The best variant of the local search always uses the steepest descent and full scanning for fuzzy classifier design. Additional advantage of the approach proposed is a possibility of the automated features selection. The numerical experiment results show the competitiveness of the approach proposed.