Evaluating And Strategy For Diversifying Population Of Gene Expression Programming By

Genetic Algorithm(GA) and Genetic Programming(GP) are the most important computation models in Evolutionary computation.They hava become a hot point in the research of Artificial Intelligence. GA uses linear string as chromosome code, and solves simple problems. GP uses tree structure as chromosome code, and solves complex problems.In 2001,F.Candida proposes a new Evolutionary Computation method named Gene Expression Programming(GEP),basing on GA and GP. GEP has overcome the shortcoming of GA and GP,integrateing their advantages. It can be using simple code expression to solve complex application problems. It has easier genetic operation and more powerful genetic operator. As a new branch of Evolutionary computation,GEP needs more solid theoretical foundation to perfect itself,and more fields to prove itself.This paper mainly develops work on present research condition ,principle, insufficiency ,improvement and application of GEP.The main content of this article is as follows:(1) discuss the bad point of standard GEP.(2) A novel GEP decoding method named Stack Decoding was proposed. It did not need to transform the chromosome into expression tree,but directly used stack for decoding and evaluating chromosome to acclerate the evolution rate. The result shows that it outperforms the conventional GEP in evolutionary efficiency.(3) Inorder to keep diversity of the population ,it improted cellular automata to avoid the problem of premature convergence.The result shows that it outperforms the conventional GEP in prediction accuracy.(4) In the field of combinatorial optimization, described the key technology of solving TSP problems by GEP.An Improved Gene Expression Programmin(IGEP) was proposed.It imported Gene Space Balance Strategy and the global convergence strategy based on the probability of mutation operators.The results shows that IGEP has better performance.