基于互信息粒度的相对约简的矩阵计算方法

On Matrix Computing Approach to Relative Reduction Based on Mutual Information Granular

提出了一种基于互信息粒度的相对约简模型,该模型利用互信息度量决策系统中的条件属性,将互信息对属性的度量映射到布尔矩阵,并能得到完备的相对约简结果;同时给出了基于布尔矩阵属性重要度的度量方法,在此基础上,设计了一种相对约简启发式计算方法,最后通过实验验证了方法的有效性.

Mutual information is one of the important technology in intelligent information processing,and the mutual information is often used for the measurement of attribute in rough sets.In this paper,the rel-ative reduction model based on mutual information granularity has been presented,the proposed model been used to evaluate the condition attributes in decision system,and the mutual information of attribute measure been mapped to Boolean matrix,which guarantees the completeness of the reduction result. What’s more,the measurement method of attribute significance based on Boolean matrix has been given. On this basis,a computing approach to relative reduction has been designed.Finally,the experimental re-sults are verified to the validity of the method.