| Genetic programming(GP) is one of branches of evolutionary computation. It is a global optimal searching technique stem from genetic algorithm (GA). In comparison with GA, the much more complicated structures of units in GP can therefore be applied to a greater diversity of problems. The thesis is organized as follows: The background and the research actuality of GP are presented in chapter 1. Some problems of GP research are also discussed in this chapter. In Chapter 2 we firstly review the details of GP. Compared with the traditional GP algorithm, an improved algorithm is proposed by modifying the strategy of creating initial population and adjusting mutation-probability and fitness function. To show that our improved algorithm is feasible and more efficient than the traditional algorithm and some other improved algorithms, numerical experiments of symbolic regression are also given in this part. In chapter 3, we shall give the applications of GP in forecast analysis and pattern determination. Then two numerical examples are given: the prediction of the development trends of real estate market in Nanjing and the diagnosis of mammary cancer using genetic programming. In the end, we conclude some special features that GP possess. Some recent research results and future research problems are also presented.
|
PDF Full Text Request