覆盖信息系统的一类悲观多粒度约简

A pessimistic multigranulation reduction in covering information systems

在覆盖信息系统中,为探讨多粒度约简结构,文中利用辨识矩阵计算覆盖信息系统的悲观多粒度约简.首先,介绍覆盖信息系统中基于多粒度粗糙集的悲观属性约简定义.接着,构建辨识矩阵和辨识函数,通过对辨识矩阵进行析取和合取运算,可以精确地得到覆盖信息系统的所有悲观多粒度近似约简.最后,利用UCI的10个数据集进行了数值实验,验证了所提算法的可行性和有效性.

In covering information systems,to explore multigranularity reduction structures,this paper calculates pessimistic multigranularity reductions for covering information systems through discernibility matrices.First,we introduce the definition of attribute reduction based on pessimistic multigranulation rough sets in covering information systems.Second,we construct the discernibility matrices and discernibility functions.By performing disjunction and conjunction operations on the discernibility matrix,all pessimistic multigranulation reductions of covering information systems can be precisely obtained.Finally,numerical experiments using 10 datasets from UCI are given to verify the feasibility and effectiveness of the proposed algorithms.