A population diversity-oriented gene expression programming for function finding

Abstract

Gene expression programming (GEP) is a novel evolutionary algorithm, which combines the advantages of simple genetic algorithm (SGA) and genetic programming (GP). Owing to its special structure of linear encoding and nonlinear decoding, GEP has been applied in various fields such as function finding and data classification. In this paper, we propose a modified GEP (Mod-GEP), in which, two strategies including population updating and population pruning are used to increase the diversity of population. Mod-GEP is applied into two practical function finding problems, the results show that Mod-GEP can get a more satisfactory solution than that of GP, GEP and GEP based on statistical analysis and stagnancy (AMACGEP). © 2010 Springer-Verlag.