New method of dynamic system identification and neural networks learning
1999
Abstract A new approximating sensitivity algorithm in prediction error methods is derived for a class of continuous nonlinear dynamic system identification including training multi-layer neural networks. The algorithm can be used to approximate the gradient of output with respect to unknown parameter in a wide class of continuous-discrete nonlinear systems. The comparison between new and conventional algorithm, and simulated example are included to demonstrate the effectiveness of the new algorithm.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
3
References
0
Citations
NaN
KQI