Pros and Cons of Fitness Distance Correlation in Genetic Programming

Pros and Cons of Fitness Distance Correlation in Genetic Programming
复制标题

遗传编程中适应度距离相关性的优缺点

DOI：
--
复制
发表时间：
2003
期刊：
影响因子：
--
通讯作者：
M. Tomassini
M. Tomassini
中科院分区：
--
文献类型：
--
作者：
L. Vanneschi;M. Tomassini

文献摘要

被引文献

相似文献

研究了遗传规划中适应度距离相关性问题。众所周知的结构距离和合适的变异算子的程序空间被定义。实验结果表明，健身距离相关性是一个合理的指数困难的一些问题，包括，第一次，多模态的，无论是新的手工定制的变异算子和标准的交叉。一个反例也显示，从而确认健身距离相关性是一个不可靠的措施问题的难度。
This paper presents a study of fitness distance correlation for genetic programming. The well known structural distance is used and suitable mutation operators for the program space are defined. Experimental results show fitness distance correlation to be a reasonable index of difficulty for a number of problems, including, for the first time, multimodal ones, both for the new hand-tailored mutation operators and standard crossover. A counterexample is also shown, thus confirming fitness distance correlation to be a not infallible measure for problem difficulty.