Partitioning of linearly transformed input space in adaptive network based fuzzy inference system

Abstract

This paper presents a new effective partitioning technique of linearly transformed input space in Adaptive Network based Fuzzy Inference System (ANFIS). Tho ANFIS is thr fuzzy system with a hybrid parameter learning method, which is composed of a gradient and a least square method. The input space can be partitioned flexibly using new modeling inputs, which are the weighted linear combination of the original inputs by the proposed input partitioning technique, thus, the parameter Learning time and the modeling error of ANFIS can Lu reduced, The simulation result illustrates the effectiveness of the proposed technique.