An LMS algorithm for training single layer globally recursive neural networks

Author

Stubberud, Peter ; Bruce, J.W.

Author_Institution

Dept. of Electr. & Comput. Eng., Nevada Univ., Las Vegas, NV, USA

Volume

3

fYear

1998

fDate

4-9 May 1998

Firstpage

2214

Abstract

Unlike feedforward neural networks which can act as universal function approximators, recursive neural networks have the potential to act as both universal function approximators and universal system approximators. In this paper, a globally recursive neural network least mean square gradient descent or a real time recursive backpropagation algorithm is developed for a single layer globally recursive neural network that has multiple delays in its feedback path

Keywords

backpropagation; feedback; function approximation; least mean squares methods; matrix algebra; real-time systems; recurrent neural nets; backpropagation; delays; feedback; function approximators; globally recursive neural network; gradient descent; least mean squares; real time systems; recurrent neural networks; system approximators; Backpropagation; Cost function; Feedforward neural networks; Finite impulse response filter; IIR filters; Least squares approximation; Neural networks; Neurofeedback; Neurons; Recurrent neural networks;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks Proceedings, 1998. IEEE World Congress on Computational Intelligence. The 1998 IEEE International Joint Conference on

Conference_Location

Anchorage, AK

ISSN

1098-7576

Print_ISBN

0-7803-4859-1

Type

conf

DOI

10.1109/IJCNN.1998.687204

Filename

687204