Gene expression programming based on symbiotic evolutionary algorithm

Gene expression programming (GEP) is a recently developed evolutionary computation method for model learning and data mining. Sometimes it is not easy when use GEP to solve too complex problem, and in the term of evolvability and learning capability, GEP is far from perfect. So enhancing the algorithm is necessary. Based on symbiotic algorithm, clonal selection algorithm and estimation of distribution algorithm (EDA), this paper proposes a new approach called symbiotic gene expression programming (SGEP). In this approach, the evolutionary process is split into two steps: symbiotic evolution and EDA evolution and the population is composed of three sub population: the set of symbionts and the set of assembly and the set of individuals. In symbiotic evolution, the immune clonal strategy is introduced, hoping to further improve the search efficiency of the algorithm. EDA evolution is an appropriate tool for building schemata in such algorithm. The experimental results on predicting the amount of gas emitted from coal face show that SGEP outperforms the standard GEP.