Training recurrent networks using the extended Kalman filter
- 2 January 2003
- proceedings article
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 4, 241-246
- https://doi.org/10.1109/ijcnn.1992.227335
Abstract
The extended Kalman filter (EKF) can be used as anon-line algorithm to determine the weights in a recurrentnetwork given target outputs as it runs. Thispaper notes some relationships between the EKF asapplied to recurrent net learning and some simplertechniques that are more widely used. In particular,making certain simplifications to the EKF gives riseto an algorithm essentially identical to the real-timerecurrent learning (RTRL) algorithm. Since the EKFinvolves adjusting unit...Keywords
This publication has 9 references indexed in Scilit:
- Decoupled extended Kalman filter training of feedforward layered networksPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- An Efficient Gradient-Based Algorithm for On-Line Training of Recurrent Network TrajectoriesNeural Computation, 1990
- Finding structure in timeCognitive Science, 1990
- Identification and control of dynamical systems using neural networksIEEE Transactions on Neural Networks, 1990
- The “moving targets” training algorithmPublished by Springer Nature ,1990
- A Subgrouping Strategy that Reduces Complexity and Speeds Up Learning in Recurrent NetworksNeural Computation, 1989
- Finite State Automata and Simple Recurrent NetworksNeural Computation, 1989
- Experimental Analysis of the Real-time Recurrent Learning AlgorithmConnection Science, 1989
- Optimization of time-varying systemsIEEE Transactions on Automatic Control, 1965