基于改进型期望值选择的GA及程序实现

Genetic Algorithms and Programming Implementation Based on Revised Expected Value Selection

遗传算法(Genetic Algorithms,GAs)作为一种新的全局优化搜索算法,在各学科中有着广泛的应用,选择策略在GA的进化中具有重要的意义,直接决定GA进化结果的效率和效果,该文指出基于轮盘赌选择的遗传算法和基于传统期望值选择的遗传算法的不足,并在此基础上提出了基于改进型期望值选择的遗传算法(RevisedExpected Value Selection-GA,REVS-GA),提高了程序的简洁度,同时通过实验证明了REVS-GA有效地克服了“早熟”现象,并且执行的效率比传统算法高.

Genetic Algorithms is used widely in the many subjects as a new global optimal algorithms, such as Machine Learning,Artificial Intelligence, Image Processing etc. In the Genetic Algorithms, Selection Operator is one of the important operators. The traditional Selection Operator based on Roulette Wheel and Expected Value may result in ＂prematuration＂ and low efficiency of programming. To solve the problem, this paper puts forward a new selection stratagem： Revised Expected Value Selection- GA（REVS- GA）. The comparison experiments show that REVS- GA increases the global convergent times, reduces generations of evolving, furthermore improving greater efficiency of programming than traditional methods.