An approach for construction of a satisfactory minimal description rules and parallel testing model

In process of machine learning, many rules generated from training objects are stored in the knowledge base, which affect the efficient of machine learning because of matching in large unrelated rules according to learning objects. Therefore it is important to organize the unrelated rules that were generated by rough set. So this paper is focus on how to simulate some classical algorithm of rough set theory by MATLAB and how to construct the satisfactory minimal description rules from given data by partitioning the data according to the discernibility matrix. Then the algorithm of extracting these satisfactory rules is obtained, which can construct the parallel testing model, for replacing the unrelated rules in knowledge base.