Directed Evolutionary Programming: Towards an Improved Performance of Evolutionary Programming
Directed Evolutionary Programming: Towards an Improved Performance of Evolutionary Programming
复制标题
DOI：
10.1109/cec.2006.1688489
复制
发表时间：
2006-09
期刊：
影响因子：
--
通讯作者：
A. Hedar;M. Fukushima
中科院分区：
文献类型：
--
作者：
A. Hedar;M. Fukushima
进化规划（EP）是进化算法（EA）的主要类型之一.改进现有的EA是必要的，以实现更好的结果，并克服其昂贵的计算复杂性。在本文中，我们提出了一个新版本的EP称为定向进化规划（DEP），其中更多的指导策略与学习终止标准调用，以克服EP的一些缺点。在DEP中，突变的孩子有机会在父母的指导下提高自己。DEP中的搜索过程由多样化和强化计划支持，以保持多样性，实现更快的收敛，并为搜索提供自动终止标准。计算实验表明，DEP是有效的，比一些著名的版本的EP便宜。
Evolutionary programming (EP) is one of the main classes of evolutionary algorithms (EAs). Improving existing EAs is necessary in order to achieve better results and overcome their costly computational complexity. In this paper, we present a new version of EP called Directed Evolutionary Programming (DEP) in which more directing strategies with learned termination criteria are invoked to overcome some drawbacks of EP. In DEP, the mutated children are given the chance to improve themselves with the guidance of their parents. The search process in DEP is supported by diversification and intensification schemes in order to keep the diversity, achieve faster convergence and equip the search with an automatic termination criteria. The computational experiments show that DEP is efficient and cheaper than some well-known versions of EP.