Gene expression programming with a local search operator

Abstract

Gene expression programming (GEP) is one of the newest evolutionary algorithms, the linear model of genetic programming that have been much attention to it, in recent years. In this article this algorithm and memetic algorithms are discussed. Here we are tried to improve its efficiency by combining gene expression programming with a local search method. The proposed algorithm called GEP-LS and it is applicable for all problems in the field of evolutionary computation. Random Mutation Hill-Climbing (RMHC) and Simulated Annealing (SA) methods are separately used to implement local search and their results are compared with each other. Finally, a comparison with the conventional gene expression programming algorithm is performed. These comparisons is performed on problems of symbolic regression, sequence induction with constants creation and robotic planning. The results show that performance of the proposed algorithm with RMHC method is relatively better than other algorithms and is able to solve all problems used here with higher accuracy and lower error.