开放尾部的基因表达式程序设计  

Open tail gene expression programming

作　　者：黄智[1] 何锫[2,3] 

机构地区：[1]长沙理工大学计算机与通信工程学院,长沙410114 [2]广州大学计算机科学与教育软件学院,广州510006 [3]武汉大学软件工程国家重点实验室,武汉430072

出　　处：《计算机工程与应用》2016年第9期1-5,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.61170199);湖南省教育厅重点资助科研项目(No.11A004)

摘　　要：基因表达式程序设计(GEP)的染色体由具有特殊限制的头、尾组成,并要求尾部符号严格取自基本的终端集。这一做法作用明了、易于表述,基本为现有GEP所采纳,但不利于语义计算的重用。谋求突破尾部限制条件,探究一种开放尾部的新型GEP算法。该算法将运行过程产生的优良个体动态地引入种群个体的基因,从而实现运算精度的提升。符号回归实验表明,开放尾部的GEP算法在平均精度性能上要优于主流GEP方法。Gene Expression Programming（GEP）genes are structurally organized in a head and a tail with special restrictions and require every symbol in tail must be strictly taken from terminal set. This practice is basically adopted by existing GEP for its perspicuous effect and facility to express, but it is not conducive to semantic computing reuse. This paper seeks to break the restriction on tail and searches a novel open tail GEP algorithm. This algorithm can improve the precision of computing by dynamically introducing the excellent individuals generated during program running to the genes of individuals in a group. The results of symbolic regression experiments show that open tail GEP algorithm outperforms mainstream GEP on average precision performance.

关 键 词：基因表达式程序设计 开放尾部基因表达式程序设计 运算精度 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]