Training the Minimal Resource Allocation Network With the Unscented Kalman Filter

摘要

The MARN is a sequential learning RBF network and has the ability to grow and prune the hidden neurons to realize a minimal network structure. This paper proposes the use of Unscented Kalman Filter (UKF) for training the MRAN parameters i.e. centers, radii and weights of all the hidden neurons. In order to reduce UKF computational load, a modification algorithm is then presented. In our simulation, we implemented the MRAN trained with UKF and the MRAN trained with EKF for states estimation. The performance of the MRAN trained with UKF is superior than the MRAN trained with EKF.