文件名称:coevolutionary-algorithm
- 所属分类:
- 人工智能/神经网络/遗传算法
- 资源属性:
- [PDF]
- 上传时间:
- 2014-05-07
- 文件大小:
- 1.14mb
- 下载次数:
- 0次
- 提 供 者:
- 王*
- 相关连接:
- 无
- 下载说明:
- 别用迅雷下载,失败请重下,重下不扣分!
介绍说明--下载内容均来自于网络,请自行研究使用
协同进化算法[I5] (Co-Evolutionary Algorithm, CEA)是研究者在协同进化理论基础
上提出的一类新算法。这类算法强调了种群与环境之间、种群与种群之间在不断进化过
程中的协调。与传统进化算法相比较,CEA可以对待求问题解空间进行合理的种群划分,
对较大规模的问题求解能有效跳出局部最优点,寻找到更好的优化解虽然CEA研
究起步较晚,但由于它的优越性,目前己成为当前进化计算的一个研究热点。
-Existing coevolutionary techniques can be divided into two
main classes: competitive coevolution and cooperative coevolution. Regardless of the approach adopted, the design of coevolutionary algorithms for MO optimization requires one to address many issues that are unique to the MO problems. In this
aspect, insights such as incorporation of various elitist and diversity mechanisms obtained from the design of MOEAs can
be similarly exploited in the design of MOCAs. On the other
hand, successful implementation of coevolution requires one to
consider various design issues [49], such as problem decomposition, handling of parameter interactions, and credit assignment. The issues of problem decomposition and parameter interactions are often problem dependent, and the approaches for
solving these issues may not be knowna priori. These factors
motivated the work for an alternative coevolutionary model presented in this paper.
上提出的一类新算法。这类算法强调了种群与环境之间、种群与种群之间在不断进化过
程中的协调。与传统进化算法相比较,CEA可以对待求问题解空间进行合理的种群划分,
对较大规模的问题求解能有效跳出局部最优点,寻找到更好的优化解虽然CEA研
究起步较晚,但由于它的优越性,目前己成为当前进化计算的一个研究热点。
-Existing coevolutionary techniques can be divided into two
main classes: competitive coevolution and cooperative coevolution. Regardless of the approach adopted, the design of coevolutionary algorithms for MO optimization requires one to address many issues that are unique to the MO problems. In this
aspect, insights such as incorporation of various elitist and diversity mechanisms obtained from the design of MOEAs can
be similarly exploited in the design of MOCAs. On the other
hand, successful implementation of coevolution requires one to
consider various design issues [49], such as problem decomposition, handling of parameter interactions, and credit assignment. The issues of problem decomposition and parameter interactions are often problem dependent, and the approaches for
solving these issues may not be knowna priori. These factors
motivated the work for an alternative coevolutionary model presented in this paper.
(系统自动生成,下载前可以参看下载内容)
下载文件列表
coevolutionary algorithm.pdf