遗传程序设计中的赋值函数和节点分类  

The Assignment Function and Node Classification in Genetic Programming

作　　者：齐越胜 王保中 康立山[1] 

机构地区：[1]武汉大学软件工程国家重点实验室,武汉430072

出　　处：《计算机科学》1998年第3期93-94,共2页Computer Science

摘　　要：遗传程序设计(GP)是用演化方法让计算机自己发现解决问题的算法,生成可以解决实际问题的程序。一个程序可以用树结构很好地表示出来,因此．演化一个程序也就是演化一个合适的树的过程。To make sure that the crossover operator produces meaningful offsprings,it is always a must that every function node of the tree has the property of closeness. This paper points out the problems of this closeness requirement ,especially when assignment function is needed in the function set. In this paper,we propose a node classification method based on the return data type of the node so as to cancel this closeness requirement. Using this method we evolved a program to model a multioutput problem. The computationary result is quite satisfactory.

关 键 词：遗传程序设计 赋值函数 节点分类 遗传算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论] O242.23[自动化与计算机技术—计算机科学与技术]