遗传程序设计(GP)的适应性函数设计  

Design of the Fitness in Genetic Programming

作　　者：马欢[1] 关富英[2] 韩战钢[3] 

机构地区：[1]国家知识产权局专利局,北京100088 [2]北京师范大学图书馆,北京100875 [3]北京师范大学系统科学系,北京100875

出　　处：《计算机工程与应用》2004年第30期113-115,共3页Computer Engineering and Applications

基　　金：国家自然科学基金项目(编号:60003018)

摘　　要：适应性函数是遗传算法和遗传程序设计得以实现的关键因素,是评价个体好坏的定量表述,决定着演化过程中群体选择复制及群体整体性的质量。该文叙述了适应性函数的设计方法,解释各设计方法的原因,并提出根据树型遗传算法的个体特点通过控制适应性函数大小对个体进行奖励惩罚。Fitness is the driving force of Darwinian natural selection and of both conventional genetic algorithms and genetic programming,based on which to evaluate individuals of current population and to reproduce the next generation from the current generation.The design for the individual's fitness of the population in genetic programming and the reason for this design| are discussed.Depending on the particular problem,the accessorial fitness design is given to encourage and punish an individual of the population by controlling its fitness value.

关 键 词：适应性函数 遗传程序设计 遗传算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]