基于粗糙集边界的V-支持向量聚类算法  

V-Support Vector Clustering Algorithm Based on Margin of Rough Set

作　　者：邹汉斌[1] 黄少年[2] 雷红艳[1] 周慧灿[1] 

机构地区：[1]湖南文理学院计算机科学与技术学院,湖南常德415000 [2]湖南商学院计算机与电子工程学院,湖南长沙410205

出　　处：《无线电工程》2009年第2期20-23,共4页Radio Engineering

基　　金：湖南省教育厅资助科研项目(06C585)

摘　　要：根据粗糙集理论的边界区域和V-支持向量机的优点对支持向量聚类算法进行改进。使用核函数进行特征空间的映射,发现最小粗糙球的包络点。根据上近似集与下近似集,定义粗糙球的内半径r和外半径为R。数据点映射若位于下近似区,则属于一个确定的聚类;若边界的点位于上近似区,属于不确定的聚类,位于球体外的点属于孤立点。实验结果表明,该聚类算法可以不需要额外的计算开销,能够解决任意形状的软聚类问题,有效地处理边界点。According to the border region of rough set theory and the merits of V-support vector machine, the algorithm of support vector clustering is improved. It uses kernel function space for the Characteristics space mapping to find the smallest rough sphere. Based on the upper approximate set and the lower approximate set, the inner radius and the outer radius of the rough sphere are defined. If the data points are mapped to the upper approximate area, it belongs to one identified Clustering. If the border points are located the lower approximate area, it belongs to one certainty clustering. If the border points are located the upper approximate area, it belongs to one uncertainty clustering. If the data points are located out of the sphere, it belongs to the isolated points. Experimental results show that the clustering algorithm can solve the soft cluster of arbitrary shape, that it do not need the additional calculation spending to effectively treat the margin point.

关 键 词：聚类 粗糙集 核方法 支持向量聚类 V-支持向量机 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]