基于粗糙集的可变正区域约简

Attribute reduction of various positive region based on rough sets

属性约简是粗糙集理论的研究重点之一.现有的各种粗糙集约简几乎都是保持某种约简准则不变,用这种方法处理一些存在异常点的数据时,在泛化能力方面存在一定的问题.针对此类问题,提出了一种可变正区域的约简方法,该方法在进行属性约简时允许正区域存在一定程度的变化.理论分析和示例表明了该方法的有效性.

Attribute reduction had been one of the hot topics in rough set theory. Almost all of principles for attribute reducts in various kinds of rough set models preserved specified criteria, which revealed shortcomings when they were employed in problems of generalization, if data contain outliers. In order to solve this problem, a reducing method called attribute reducts of various positive regions, was presented, in which the change of positive regions was allowed when attribute reduction was conducted. Theoretical analysis and an example showed that this method was valid.