粗糙集理论中属性相对约简算法  

Algorithms of Attribute Relative Reduction in Rough Set Theory

作　　者：张腾飞[1] 肖健梅[1] 王锡淮[1] 

机构地区：[1]上海海事大学电气自动化系,上海200135

出　　处：《电子学报》2005年第11期2080-2083,共4页Acta Electronica Sinica

基　　金：国家自然科学基金(No.60074004);上海市教委科学研究重点项目(No.04FA02);上海市重点学科建设项目(No.T0602)

摘　　要：粗糙集理论是近年来发展起来的一种有效地处理模糊和不确定性知识的数学工具,而求核与约简是粗糙集理论中的两个重要问题,现已证明求决策表所有约简和最小约简是一个典型的NP难题.本文在分析粗糙集理论的基础上,发现了正区域的一些有用性质,提出了一种利用正区域直接求核的方法,并利用正区域的启发式信息给出了两种相对约简算法.Rough set is a valid mathematical theory developed in recent years,which has the abihty to deal with imprecise,uncertain,and vague information.The core and reduction of attributes are two important topics in the research on rough set theory. It has been proven that computing all the reductions and the optimal （minimal） reduction of decision table is a NP-hard problem. In this paper, Rough set theory is deeply investigated; a number of useful properties of the positive region are discovered. Based on the above findings, we present a calculation algorithm for core directly. And then, two algorithms for relative reduction based on the positive region are designed.

关 键 词：粗糙集 求核 相对约简 决策表 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]