Adaptive Learning Rate for RBFNN

摘要

Radial basis function neural networks are becoming popular in different kinds of classification problems such as pattern recognition, nonlinear identification, adaptive control, etc. In this paper, an adaptive learning rate for the learning algorithm of radial basis function neural networks is presented. The analysis considers the learning algorithm in a feedback structure in the presence of noisy perturbations. A feedback structure is used to model the learning algorithm and small gain theorem is used to inspect the stability. This results in an adaptive learning rate for robust performance.