基于最大决策邻域粗糙集的不确定性度量方法

Uncertainty Measurement Method for Information System Based on Max-decision Neighborhood Rough Set

邻域粗糙集模型是经典粗糙集模型的变型,对处理数值型数据具有较好的优势性.本文引入最大决策邻域粗糙集模型,该模型密切关注边界样本,通过增加与决策类有最大交集的邻域样本来扩大正域,并在该模型上定义了最大决策粗糙度的概念.为了能够反映正域、负域的同时变化,提出一种基于边界域的不确定性度量方法.为了能够更全面的度量,在最大决策邻域粗糙集模型中定义了最大决策邻域粒结构,并基于该粒结构提出了最大决策邻域粒度概念,该粒度是对信息系统的分类能力的度量.文章最后提出一种基于最大决策邻域粗糙集的混合不确定性度量方法,将两种度量方法进行结合.实验结果表明,所提出的度量方法在邻域信息系统中具有较好的分类效果.

Neighborhood rough set model is a variant of classical rough set model.It has better advantages in dealing with numerical data.The Max-decision neighborhood rough set model is introduced in this paper.The model pays close attention to the boundary samples.It enlarges the positive domain by adding the samples with the maximum intersection of neighborhood and decision class.The concept of maximum decision roughness is defined on the model in order to reflect the positive and negative domains.A method of uncertainty measurement based on boundary domain is proposed.In order to measure more comprehensively,the maximum decision granular structure is defined in the max-decision neighborhood rough set model.Based on the granular structure,the concept of Max-decision neighborhood granularity is proposed,which is a measure of the classification ability of information systems.Finally,the method of mixed uncertainty measurement method based on max-decision neighborhood rough set is proposed,which combines the two measurement methods.The experimental results show that the proposed method has good classification effect in the neighborhood information system.