COMPLEX-VALUED RADIAL BASIS FUNCTION NETWORK .1. NETWORK ARCHITECTURE AND LEARNING ALGORITHMS

被引：71

作者：

CHEN, S

MCLAUGHLIN, S

MULGREW, B

机构：

[1] Department of Electrical Engineering, University of Edinburgh, Edinburgh, EH9 3JL Scotland, King's Buildings

来源：

SIGNAL PROCESSING | 1994年 / 35卷 / 01期

关键词：

COMPLEX-VALUED SIGNAL; RADIAL BASIS FUNCTION NETWORK; ORTHOGONAL LEAST SQUARES ALGORITHM; UNSUPERVISED CLUSTERING;

D O I：

10.1016/0165-1684(94)90187-2

中图分类号：

TM [电工技术]; TN [电子技术、通信技术];

学科分类号：

0808 ; 0809 ;

摘要：

The paper proposes a complex radial basis function network. The network has complex centres and connection weights, but the nonlinearity of its hidden nodes remains a real-valued function. This kind of network is able to generate complicated nonlinear decision surfaced or to approximate an arbitrary nonlinear function in complex multi-dimensional space, and it provides a powerful tool for nonlinear signal processing involving complex signals. The paper is divided into two parts. The first part introduces the network architecture and derives both block-data and recursive learning algorithms for this complex radial basis function network. The complex orthogonal least squares algorithm is a batch learning algorithm capable of constructing an adequate network structure, while a complex version of the hybrid clustering and least squares algorithm offers real-time adaptation capability. The identification of a nonlinear communications channel model is used to illustrate these two learning algorithms. In the second part of the paper, a practical application of this complex radial basis function network is demonstrated using digital communications channel equalisation.

引用

页码：19 / 31

页数：13

共 36 条

[1]

Benvenuto, Marchesi, Piazza, Uncini, Complex neural networks for equalizing nonlinear digital radio links, Adaptive Algorithms: Applications and Non-classical Schemes, pp. 162-170, (1991)

[2]

Billings, Chen, Extended model set, global data and threshold model identification of severely non-linear systems, Internat. J. Control, 50, 5, pp. 1897-1923, (1989)

[3]

Bjorck, Solving linear least squares problems by Gram-Schmidt orthogonalization, Nordisk Tidskr. Informations-Behandling, 7, pp. 1-21, (1967)

[4]

Broomhead, Lowe, Multivariable functional interpolation and adaptive network, Complex Systems, 2, pp. 321-355, (1988)

[5]

Chen, Billings, Neural networks for non-linear dynamic system modelling and identification, Internat. J. Control, Special Issue on Intelligent Control, 56, 2, pp. 319-346, (1992)

[6]

Chen, Billings, Luo, Orthogonal least squares methods and their application to non-linear system identification, Internat. J. Control, 50, 5, pp. 1873-1896, (1989)

[7]

Chen, Cowan, Billings, Grant, Parallel recursive prediction error algorithm for training layered neural networks, Internat. J. Control, 51, 6, pp. 1215-1228, (1990)

[8]

Chen, Cowan, Grant, Orthogonal least squares learning algorithm for radial basis function networks, IEEE Trans. Neural Networks, 2, 2, pp. 302-309, (1991)

[9]

Chen, Grant, Cowan, Orthogonal least squares algorithm for training multi-output radial basis function networks, Proc. IEEE 2nd Internat. Conf. Artificial Neural Networks, pp. 336-339, (1991)

[10]

Chen, McLaughlin, Mulgrew, Complex-valued radial basis function network, Part II: Application to digital communications channel equalisation, Signal Processing, 36, (1994)

← 1 2 3 4 →