Complexity reduction to non-singleton fuzzy-neural network

Abstract

A singular value decomposition (SVD) based reduction technique has been proposed for a singleton-based fuzzy neural network. In fuzzy theory, the use of the non-singleton consequent-based Takagi-Sugeno model is also adopted. By applying a non-singleton-based fuzzy model to fuzzy neural networks, a non-singleton-based network is obtained. The main objective of this work is to extend the SVD-based reduction technique that has been proposed for fuzzy neural networks to non-singleton-based networks.