An Efficient Attribute Reduction Algorithm Designed with Heuristic Function

Abstract

Author comprehensively analyzes the main idea of attribute reduction algorithm about positive region at present. And author also summarizes its advantage and disadvantage. With the main idea of attribute reduction on discernibility object pair and on the basis of simplified decision table, author defines a function, which can measure the number of discernibility object pair produced by conditional attribute set. And with this function, author designs a heuristic function and describes a fast algorithm about this heuristic function in detail. Then author uses this heuristic function to design an efficient algorithm of attribute reduction based on positive region. And at last, author uses an instance to illuminate that this new algorithm can efficiently eliminate the shortages, including too much time consumption, too much space consumption and low efficiency in computing discernibility matrix. Finally, author validates that this new algorithm has high efficiency in practice.