Double System Gene Expression Programming and It’s Application in Function Finding Problems
Chaoxue Wang, Kai Zhang, Dong Hui, Fangxiao Zhou · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2014
Gene ExpressionProgramming (GEP) is a powerful evolutionary method for knowledge discovery and model learning, which simulates the process of natural evolution, and has the phenomena of low converging speed and readily being premature.After the birth of human society, the evolution processes of nature world and human show a rising acceleration characteristic under the artificial intervention.Inspired by this, this paper proposes a double system gene expression programming (DS-GEP), which consists of natural evolution system and artificial intervention system.The artificial intervention system includes individual intervention operation and population intervention operation.The individual intervention operation aims to repair the unfeasible genes in individuals with superior genes from gene pool.The population intervention operation uses extinction/restart strategy to form new population with high diversity after the evolution has fallen into stagnation.To validate the superiority of DS-GEP, DS-GEP and the standard GEP and an improved GEP in the relevant literatures are compared as regards enough function finding problems.The test results show that DS-GEP can overcome the stagnation and premature convergence phenomenon effectively during the evolutionary process, and promises competitive performance.