基于可辨识矩阵的快速粗糙集属性约简算法

Karno Bozi提出的Core Searching算法在向约简中插入候选属性的时候,根据属性出现次数需要循环查找可辨识矩阵中的所有剩余项,直至矩阵为空,导致计算量较大和结果中冗余属性存在的可能.基于Core Searching算法提出通过给属性设立计数器的基于可辨识矩阵的快速属性约简算法,实例分析表明,该算法与Core Searching算法相比,在计算量减少和循环次数减少的同时能得到更简约的结果,是一种快速、高效的属性约简算法.

Fast attribute reduction algorithm of rough set based on discernibility matrix

When core searching algorithm introduced by Karno Bozi extended the reduction,it needed to circularly search all the remanent items in discernibility matrix based on the appearance times of candidate attributes,until the matrix became empty,this induced the exis-tence of redundant attributes in the result.The improved fast algorithm based on the discernibility matrix,it set up counters for different attributes.Theoretical analysis and instance results show that the algorithm,obtaining the more reductive result,requiring less computional effort than core searching algorithm,is a simple and effective attribute reduction algorithm.