基于相对距离的改进粗K-means方法  

Improved algorithm of rough K-means based on relative distance

作　　者：王明春[1,2] 唐万生[1] 江琪[2] 刘鑫[2] 

机构地区：[1]天津大学系统工程研究所,天津300072 [2]天津工程师范学院数理系,天津300222

出　　处：《计算机应用》2009年第4期1102-1105,共4页journal of Computer Applications

基　　金：国家自然科学基金资助项目(70471049);天津市教委科技发展基金资助项目(20061005)

摘　　要：对现有的两种基于绝对距离的粗K-means方法进行了讨论,指出了各自的不足之处。在此基础之上,讨论了用相对距离替代绝对距离的合理性,从而给出了基于相对距离的粗K-means方法。通过对随机数据、Iris数据和文本数据进行聚类效果比较,验证了基于相对距离的粗K-means方法的可行性和有效性。Two rough K-means algorithms based on absotute distance were discussed in the first place, and then the deficiencies of them were indicated. After that, the rationality of the algorithms was presented when the absolute distance was changed to relative distance, and for the reason, the improved algorithms of rough K-means based on relative distance was given. At last, the feasibility and effectiveness of this algorithm were testified by comparing with random, Iris and text data on clustering effect.

关 键 词：粗糙集 粗K-means方法 聚类 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]