一种基于模糊聚类的区间值属性约简算法  

A Novel Interval-valued Attribution Reduction Algorithm Based on Fuzzy Cluster

作　　者：郭庆[1] 刘文军[2] 焦贤发[1] 吴磊[1] 

机构地区：[1]合肥工业大学数学学院,安徽合肥230009 [2]长沙理工大学数学与计算科学学院,湖南长沙410076

出　　处：《模糊系统与数学》2013年第1期149-153,共5页Fuzzy Systems and Mathematics

基　　金：国家自然科学基金资助项目(11172086)

摘　　要：针对区间值信息系统基于粗糙集理论提出一种新的属性约简算法:首先计算同一属性下对象间的相似度,然后通过合取算子计算出所有属性下对象之间的相似度矩阵,再用模糊聚类中的传递闭包算子得到等价矩阵,将区间值信息系统转化为具有等价关系的信息系统并且进行约简,从而得到λ-核,同时给出了该算法的复杂度。最后通过一个实例表明这种算法的有效性和合理性。In this paper, a novel attribution reduction algorithm of interval-valued information system is proposed based on rough set theory： firstly the resemblance between two objects under the same attribute is calculated then the resemblance matrix of all attributes is calculated through conjunction operator. The equivalent matrix is got by transitive closure operator of fuzzy cluster method and then the interval-valued information system is changed into a equivalent relation system which can be reducted based on algorithm we propose. Socore can be got and the complexity of algorithm is given at the same time. Lastly a numerical example is given to demonstrate the effectiveness and rationality of algorithm we propose.

关 键 词：区间值信息系统 模糊聚类 模糊等价关系 传递闭包 可区分矩阵 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]