一种改进的小生境遗传聚类算法  

An Improved Niche Genetic Clustering Algorithm

作　　者：孙红艳[1] 王英博[1] 

机构地区：[1]辽宁工程技术大学电子与信息工程学院,辽宁葫芦岛125105

出　　处：《计算机系统应用》2010年第2期37-40,共4页Computer Systems & Applications

摘　　要：传统的遗传算法具有早熟收敛和后期收敛速度慢的缺点,采用改进的小生境技术解决这一问题,同时根据具体问题改进了遗传算子,并将改进后的小生境遗传算法应用于聚类挖掘中。由于聚类挖掘算法中的K-means算法对初始值K的选取敏感,选取值的不同会导致聚类结果的不同,很容易陷入局部最优,使得聚类结果很差。因此,将改进的小生境遗传算法和K-means算法相结合,得出一种改进的小生境遗传聚类算法。验证表明优该算法对提高聚类分析质量是有效的。The traditional genetic algorithm has the shortcomings of premature convergence and slow convergence. This paper adopts improved niche technology to solve this problem. It also uses the specific issues to improve the genetic operators, and the improved niche genetic algorithm is applied to Clustering Mining. As the K-means algorithm in the clustering algorithm for mining has the problem of the selection of the initial value of K-senstive and if we select a different value, it will lead to a different clustering result. It is easy to fall into local optimum. So it will make poor clustering results. Therefore, this article combines the improved niche genetic algorithm with K-means algorithm to produce a new improved algorithm named an improved niche genetic clustering algorithm. It is verified that the algorithm is valid in improving the quality of clustering analysis.

关 键 词：小生境技术 聚类挖掘 K-MEANS算法 小生境遗传算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]