New fuzzy attribute reduction algorithm based on similarity category clusters

Classical rough set has a limited processing capacity in fuzzy decision table. Combining fuzzy set with classical rough set, attribute reduction algorithm on fuzzy decision table is studied. First, new similarity degree and new similarity category are defined. In the meantime, similarity category clusters which are divided by condition attribute are provided. And then, two theorems are presented. Subsequently, a new attribute reduction algorithm is proposed. Finally, the new attribute reduction algorithm is verified through a performance evaluation decision table of the self-repairing flight-control system. The result shows the proposed attribute reduction algorithm is able to deal with fuzzy decision table to a certain extent.