Attribute Reduction Algorithm Based on the Simplified Discernibility Matrix of Granularity

This paper gives the definition of discernibility matrix of granularity in decision table and the corresponding attribute reduction and the definition of discernibility matrix of granularity in the simplified decision table and the corresponding attribute reduction. Therefore, verifying the equivalence of the definition of attribute reduction in the simplified decision table and the definition of attribute reduction based on relative granularity of decision table. On the basis of the above theories, a new algorithm is designed based on knowledge granulation for attribute reduction in simplified decision table, and the corresponding complexity of time is reduced to O(|C|2|U′ pos ||U|). Finally, an example is given to illustrate the validity of the new algorithm.