A fuzzy classifier design method based on the extended hedge algebras quantification

Abstract

The conventional method of quantification of hedge algebras [1-2] has achieved effective successes in its application to the fuzzy classifier design problem of fuzzy sets based semantics of linguistic terms in the form of triangle fuzzy sets [3, 4, 11]. The numeric semantic of a term defined by its semantically quantifying mapping value is a point which is relevant to define the vertex of the triangular fuzzy set. An extended quantification method of hedge algebras proposed in [10] using partitions of the feature spaces allows to model the core semantics of a terms in the form of an interval, which is upper base of a trapezoidal fuzzy set, based on a degree k semantically quantifying mapping intervals. Based on this extended hedge algebras quantification, this paper proposes a method for designing linguistic terms and fuzzy rule based classifiers with trapezoidal fuzzy set semantics and then examines the effectiveness of the new quantification method in solving classification problems. The experimental results over 10 datasets are given to show that the proposed method is more flexible and produces better results