BUILDING NEURO-FUZZY INFERENCE SYSTEMS BASED ON INPUTOPTIMAL-FUZZY-SET ESTABLISHMENT

Abstract

This study presents an approach for approximation an unknown function y=f(x ̅) from a numerical data set based on a neuro-fuzzy inference system modeling. The focus of interest in proposed approach is to increase degree of accuracy of the degree of this approximation. New algorithms named CSHL, HLM1 and HLM2, which are used for this target, are presented. The first new algorithm, CSHL, which uses functions named pure function ψ and penalty function τ effecting as direction for input data space partition, is used to build data clusters. The second and the third algorithm based on the Hyperplane Clustering algorithm of [1] and the CSHL algorithm are used to establish adaptive neuro-fuzzy inference systems. A series of numerical experiments are performed to assess the efficiency of the proposed approach.