An Adaptive Multi-phenotype GEP algorithm

Gene Expression Programming (GEP) is a GA and GP based evolutionary computation method. Researchers have improved the efficiency and effectiveness GEP by optimizing gene encoding scheme and inventing new genetic operators. An Adaptive Multi-phenotype GEP (AM-GEP) is proposed in this paper in order to solve the problem of excessive evolutionary generations and massive time consuming. In the evolutionary process of AM-GEP, individuals change the number of genes adaptively, and the convergent rate is greatly improved with a new gene combination mechanism. Experiments showed that running time is reduced significantly, the expression space of chromosomes is extended, and the quality of solutions is improved with AM-GEP.