同可区分度属性约简的研究

Research of attribute reduction algorithm based on common discernibility degree

通过具体算例指出基于同可区分度属性约简不是王国胤等提出的条件信息熵约简。理论上证明了基于同可区分度属性约简仅是基于差别矩阵的HU属性约简,它是一种基于梁吉业等提出的新条件信息熵属性约简。证明了同可区分度约简协调集一定是代数协调集和信息熵协调集,从而代数约简核属性和信息熵核属性一定是同可区分度约简核属性的子集。

It is showed that the attribute reduction based on common discernibility degree is different from the one based on conditional information entropy proposed by Wang Guoyin et al.it is proved that the attribute reduction based on com-mon discernibility degree is just an HU’s attribute reduction based on discernibility matrix,which is equivalent to the reduc-tion that is based on the information entropy presented by Liang Jiye et.al.It is pointed out that the core attributes of alge-braic and conditional information entropy reductions must be included by the one based on common discernibility degree.