On-line adaptive RBF network in stationary and nonstationary environment

Abstract

Sequential adaptation of RBF network parameters and structure is achieved using extended Kalman filter. Criterion for network growing is obtained from Kalman filter's consistency test. The Optimal Brain Surgeon and Optimal Brain Damage prunning methods are derived for networks which parameters are estimated by EKF. Criteria for neurons/connections pruning are based on the statistical parameter significance test.