Reasoning Methods Based on OWA Operators under Fuzzy Mayority in Fuzzy Rule-Based Classification Systems

In Fuzzy Rule Based Classification Systems (FRBCSs), the classic Fuzzy Reasoning Method (FRM) classifies an example with the class indicated in the consequent of the rule which has a higher association degree. Other FRMs, that use the information provided by all the rules compatible with the example to be classified, increase the generalisation capacity of the classification system. Nevertheless, in FRBCSs composed of a Rule Base (RB) with a high cardinality, this prediction ability can be drastically decreased because many rules with a lower association degree can exert a stronger influence than others with a higher degree, so this fact that may lead to errors in the classification. In this paper, we present FRMs that select the information to be added in the inference process. This selection is modelled throught several operators belonging to the OWA family, under the fuzzy majority concept. The results obtained with two well-known example bases and RBs generated by two different lea..