Title :
On-line adaptive RBF network in stationary and nonstationary environment
Author :
Todorovic, Branimir ; Stankovic, Miomir ; Todorovic-Zarkula, Slavica
Author_Institution :
Fac. of Occupational Safety, Nis Univ., Yugoslavia
Abstract :
Sequential adaptation of RBF network parameters and structure is achieved using extended Kalman filter (EKF). Criterion for network growing is obtained from Kalman filter´s consistency test The optimal brain surgeon and optimal brain damage pruning methods are derived for networks which parameters are estimated by EKF. Criteria for neurons/connections pruning are based on the statistical parameter significance test.
Keywords :
Kalman filters; filtering theory; online operation; optimisation; parameter estimation; radial basis function networks; statistical analysis; EKF; RBF network parameters; RBF network structure; extended Kalman filter; nonstationary environment; online adaptive RBF network; optimal brain damage pruning method; optimal brain surgeon pruning method; sequential adaptation; stationary environment; statistical parameter significance test; Adaptive systems; Artificial neural networks; Intelligent networks; Kalman filters; Neurons; Parameter estimation; Quantum cascade lasers; Radial basis function networks; Surges; Testing;
Conference_Titel :
Electrotechnical Conference, 2000. MELECON 2000. 10th Mediterranean
Print_ISBN :
0-7803-6290-X
DOI :
10.1109/MELCON.2000.879968
