Computing and Analyzing the Sensitivity of Radial-Basis Function Neural Network

This paper proposes an algorithm to compute the sensitivity of the Radial-Basis Function Neural Network (RBFNN) due to the errors of the inputs and others parameters of the net works. For simplicity and practicality, all inputs and weights are assumed to be independent and identically distributed (i.i.d) with uniform distribution. A number of simulations are conducted and the good agreement between the experimental results and the theoretical results verifies the reliability and feasibility of the proposed algorithm. The relationship between the sensitivity of RBFNN and input error and the perturbation of others parameters is given.