Nonlinear system identification using neural networks trained with natural gradient descent

Authors:
Mohamed Ibnkahla
Affiliations:
Electrical and Computer Engineering Department, Queen's University, Kingston, Ontario, Canada
Venue:
EURASIP Journal on Applied Signal Processing
Year:
2003

Citing 16
Cited 1

System identification: theory for the user

System identification: theory for the user
Learning internal representations by error propagation

Parallel distributed processing: explorations in the microstructure of cognition, vol. 1
Nonlinear black-box modeling in system identification: a unified overview

Automatica (Journal of IFAC) - Special issue on trends in system identification
Adaptive filter theory (3rd ed.)

Adaptive filter theory (3rd ed.)
Natural gradient works efficiently in learning

Neural Computation
Applications of neural networks to digital communications: a survey

Signal Processing - Special issue on emerging techniques for communication terminals
Neural Networks: A Comprehensive Foundation

Neural Networks: A Comprehensive Foundation
Principles of Digital Transmission: With Wireless Applications

Principles of Digital Transmission: With Wireless Applications
Semiparametric model and superefficiency in blind deconvolution

Signal Processing
Adaptive Method of Realizing Natural Gradient Learning for Multilayer Perceptrons

Neural Computation
Neural network modeling and identification of nonlinear channelswith memory: algorithms, applications, and analytic models

IEEE Transactions on Signal Processing
Blind identification of LTI-ZMNL-LTI nonlinear channel models

IEEE Transactions on Signal Processing
Analysis of stochastic gradient tracking of time-varying polynomialWiener systems

IEEE Transactions on Signal Processing
Identification of a class of nonlinear systems under stationarynon-Gaussian excitation

IEEE Transactions on Signal Processing
Statistical analysis of neural network modeling and identificationof nonlinear systems with memory

IEEE Transactions on Signal Processing
Stochastic analysis of gradient adaptive identification ofnonlinear systems with memory for Gaussian data and noisy input andoutput measurements

IEEE Transactions on Signal Processing

Quality modeling of chemical product based on a new chaotic Elman neural network

ICNC'09 Proceedings of the 5th international conference on Natural computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

We use natural gradient (NG) learning neural networks (NNs) for modeling and identifying nonlinear systems with memory. The nonlinear system is comprised of a discrete-time linear filter H followed by a zero-memory nonlinearity g(ċ). The NN model is composed of a linear adaptive filter Q followed by a two-layer memoryless nonlinear NN. A Kalman filter-based technique and a search-and-converge method have been employed for the NG algorithm. It is shown that the NG descent learning significantly outperforms the ordinary gradient descent and the Levenberg-Marquardt (LM) procedure in terms of convergence speed and mean squared error (MSE) performance.