摘要
属性约简问题是粗糙集理论中一个核心的研究课题。本文基于区分矩阵存在大量冗余数据考虑,提出了一种利用吸收算子的快速算法,不用生成庞大的区分矩阵,与现有的属性约简算法相比,节约了大量时间和空间。对于UCI绝大部分数据,在P41.6G的PC机上都可以在1秒以内的时间求出所有的最小约简。实验数据说明了该算法的有效性。
Attribute reduction is a core subject in the domain of rough set theory.Because of the existence of much redundant data in discernibility matrixes,the paper presents a fast algorithm using the absorptive operator.By the algorithm,huge discernibility matrixes need not be generated,and much time and space are saved compared with the existing algorithms of attribute reduction.The algorithm presented in the paper is efficient.For the most data of UCI it can get all of the minimum reductions in less than 1 second on a P4 1.6G PC.
出处
《计算机工程与科学》
CSCD
2007年第1期97-100,共4页
Computer Engineering & Science
基金
国防科技大学基础项目研究基金资助项目(GC03-02-003)
关键词
粗糙集
约简
吸收算子
区分矩阵
rough set
reduction
absorptive operator
discernibility matrix
