Attributes reduction algorithms for m-polar fuzzy relation decision systems

Nowadays, attribute reduction has become a significant topic in relation decision systems. Their applications come from different domains of the computer sciences, including machine learning, data mining and pattern recognition, which often involve a large number of attributes in data. Several attribute reduction methods are presented in the literature in order to help solving decision-making problems efficiently. A common characterization for these approaches is still missing, that is, although attribute reduction methods of relation decision systems and fuzzy relation decision systems exist, a common generalization for them is still missing. This study presents a systematic discussion of attribute reduction based on m-polar fuzzy (mF, in short) relation systems and mF relation decision systems, which are respective extensions of fuzzy relation systems and fuzzy relation decision systems. This study provides mathematical results on the attribute reduction algorithms based upon mF relation systems and mF relation decision systems. Both are explained with numerical examples. The resulting algorithms permit to reinterpret the upshots of traditional reduction methods, providing them with larger generality and unification abilities. Afterwards, two real-life applications of the proposed attribute reduction approaches prove their validity and feasibility. Finally, the attribute reduction methods developed here are compared with some existing approaches to show their reliability. (c) 2021 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).