Function Finding Based on Gene Expression Programming

This paper proposes a new evolution algorithm, M_GEP, based on the concept of multi-gene chromosome in gene expression programming. The algorithm has two characters: (1) a chromosome is composed of more than one gene; (2) the sub-genes are linked together according to the linking gene which may conclude more than one kind of function. We give two examples, whose results show that the models set up by M_GEP are better than the models set up by GEP.