A logic method for efficient reduction of the space complexity of the attribute reduction problem

Authors : Mehmet HACIBEYOĞLU, Fatih BAŞÇİFTÇİ, Şirzat KAHRAMANLI

Pages : 643-656

View : 10 | Download : 6

Publication Date : 0000-00-00

Article Type : Research Paper

Abstract :The goal of attribute reduction is to find a minimal subset insert ignore into journalissuearticles values(MS); R of the condition attribute set C of a dataset such that R has the same classification power as C. It was proved that the number of MSs for a dataset with n attributes may be as large as insert ignore into journalissuearticles values(n/2n); and the generation of all of them is an NP-hard problem. The main reason for this is the intractable space complexity of the conversion of the discernibility function insert ignore into journalissuearticles values(DF); of a dataset to the disjunctive normal form insert ignore into journalissuearticles values(DNF);. Our analysis of many DF-to-DNF conversion processes showed that approximately insert ignore into journalissuearticles values(1-2/insert ignore into journalissuearticles values(n/2n); \times 100);% of the implicants generated in the DF-to-DNF process are redundant ones. We prevented their generation based on the Boolean inverse distribution law. Due to this property, the proposed method generates 0.5 \times insert ignore into journalissuearticles values(n/2n); times fewer implicants than other Boolean logic-based attribute reduction methods. Hence, it can process most of the datasets that cannot be processed by other attribute reduction methods.
Keywords : Information system, dataset, attribute reduction, feature selection, discernibility function, computational complexity, reduct