On Feature Reduction Algorithm of Binary Discernibility Matrix Based on Mutual Information Model

Key words: 
• mutual information，feature reduction，binary discernibility matrix，rough sets /
•  /
•  /

Abstract: Feature reduction of mutual information is one of the important approaches in Filter feature re‐duction;the purpose of mutual information is often used to delete irrelevant features and make the mini‐mum correlation in reduction subset .In this paper ,in view of discernibility between features ,the feature reduction model based on binary discernibility matrix has been presented;it is proved that feature reduc‐tion based on binary discernibility matrix is equivalent to feature reduction based on mutual information , and measure metrics of condition features has been proposed ,a quick feature reduction algorithm with In‐cremental method been designed .Finally ,the experimental results are verified to the feasibility of the algo‐rithm .