Learning convergence analysis for Takagi-Sugeno Fuzzy Neural Networks

Abstract

In this paper, we provide a mathematical formulation of the Takagi-Sugeno Fuzzy Neural Network (TS-FNN) to study convergence properties. Note that we describe both information retrieval and learning rules by algebraic equations in matrix form. We then investigate the convergence characteristics and learning behaviors for the TS-FNN by use of these algebraic equations and the eigenvalues of derived matrices. Numerical examples are carried out to further verify the analysis.