RBF networks training using a dual extended Kalman filter

@article{Ciocoiu2002RBFNT,
  title={RBF networks training using a dual extended Kalman filter},
  author={Iulian B. Ciocoiu},
  journal={Neurocomputing},
  year={2002},
  volume={48},
  pages={609-622}
}
A new supervised learning procedure for training RBF networks is proposed. It uses a pair of parallel running Kalman $lters to sequentially update both the output weights and the centres of the network. The method o2ers advantages over the joint parameters vector approach in terms of memory requirements and training time. Simulation results for chaotic time… CONTINUE READING