首页 | 本学科首页   官方微博 | 高级检索  
     检索      

演化算法中各种选择算子的选择压力比较
引用本文:林广明,黄程波,陆芸婷.演化算法中各种选择算子的选择压力比较[J].深圳信息职业技术学院学报,2009,7(2):24-29.
作者姓名:林广明  黄程波  陆芸婷
作者单位:深圳信息职业技术学院信息技术研究所,深圳,518029
基金项目:国家自然科学基金项目,广东省自然科学基金,深圳市科技计划 
摘    要:对于演化计算,变异和杂交这样的算子是在完全没有导向的方式产生的新个体。选择算子在演化过程中起着导向的作用,使得个体沿着适应值优良的方向演化。选择算子是在演化过程唯一受适应值影响的组件。本文将讨论现在流行的选择算子的选择压力并比较他们的影响。选择算子在演化过程中应用较好适应值的个体信息并传递到后代,为算法设计中挑选选择算子提供理论依据。

关 键 词:演化算法  选择算子  选择压力  接管时间

Comparison selective pressure of different selection operators in evolutionary algorithms
LIN Guangming,HUANG Chengbo,LU Yunting.Comparison selective pressure of different selection operators in evolutionary algorithms[J].Journal of Shenzhen Institute of Information Technology,2009,7(2):24-29.
Authors:LIN Guangming  HUANG Chengbo  LU Yunting
Institution:(Shenzhen Institute of Information Technology, Shenzhen 518029, China)
Abstract:In Evolutionary Algorithms, the genetic operators such as mutation and crossover create new individuals in a completely undirected way. The selection operator plays a major role by imposing a direction on the search process that is a clear preference for those individuals which perform better according to the fitness measurements. Selection is the only component in Evolutionary Algorithms where the fitness of individuals has an impact on the evolution process. In this paper we discuss the selective pressure of different well-known selection operators and compared their effects.The selection operator uses the fitness information of higher quality individuals for transferring to the next generation in the evolution process. It provides a mechanism to make a balance between exploitation - by increasing emphasis on the better individuals and exploration - by providing similar chances to survive even for worse individuals. It can he used to theoretically guide the choice of selection strategy in algorithm design.
Keywords:evolutionary algorithm: selection operators: selective pressure  takeover time
本文献已被 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号