New Methods for Uncertainty Representations in Neuro-Fuzzy Systems

摘要

In this paper we discuss a new method for uncertainty representations in neuro-fuzzy systems. Expert uncertainty concerning antecedent fuzzy linguistic values are expressed in the form of linguistic values e.g. roughly, more or less. That idea is incorporated into relational neuro-fuzzy systems. In the paper both type-1 and type-2 fuzzy systems are considered. Numerical simulations of the new fuzzy model are presented.