Radial Basis Function Neural Networks With Sequential Learning, Progress In Neural Processing

by Ying Wei Lu, P Saratchandran, Narasimman Sundararajan

Overview

This book presents in detail the newly developed sequential learning algorithm for radial basis function neural networks, which realizes a minimal network. This algorithm, created by the authors, is referred to as Minimal Resource Allocation Networks (MRAN). The book describes the application of MRAN in different areas, including pattern recognition, time series prediction, system identification, control, communication and signal processing. Benchmark problems from these areas have been studied, and MRAN is compared with other algorithms. In order to make the book self-contained, a review of the existing theory of RBF networks and applications is given at the beginning.

Product Details

ISBN-13:	9789810237714
Publisher:	World Scientific Publishing Company, Incorporated
Publication date:	10/05/1999
Series:	Progress In Neural Processing , #11
Pages:	232
Product dimensions:	6.20(w) x 8.60(h) x 0.70(d)

Table of Contents

	Preface	xi
1	A Review of Radial Basis Function (RBF) Neural Networks	1
1.1	An Overview	1
1.2	Review of RBF Neural Networks	3
1.3	Applications of RBF Neural Networks	15
2	A Novel Sequential Learning Algorithm for Minimal Resource Allocation Neural Networks (MRAN)	21
2.1	RANEKF Algorithm	23
2.2	Problems with RANEKF	26
2.3	Pruning Strategy	30
2.4	Sliding Window RMS Criterion for Adding Hidden Neurons	35
2.5	New Learning Algorithm for Minimal Resource Allocation Network	37
3	MRAN for Function Approximation & Pattern Classification Problems	41
3.1	RBF Neural Networks for Function Approximation	41
3.2	Benchmark Problems for Function Approximation	50
3.2.1	A Function of Summation for Six Exponential Functions	50
3.2.2	Hermite Polynomial	52
3.2.3	Benchmark Problems Used by The DI Algorithm	60
3.3	Benchmark Problems from PROBEN1	69
4	MRAN for Nonlinear Dynamic Systems	79
4.1	MRAN for Time Series Prediction	80
4.1.1	Chaotic Time Series Prediction--Mackey-Glass Series	81
4.1.2	Prediction of Currency Exchange Rate	85
4.2	MRAN for Nonlinear Signal Processing	93
4.3	MRAN for System Identification	99
4.3.1	System with Three Exponential Time Functions	102
4.3.2	Nonlinear System Identification - Fixed Dynamics Case	106
4.3.3	Nonlinear System Identification - Varying Dynamics Case	112
4.3.4	Comparison of MRAN with On-line Structural Adaptive Hybrid Learning (ONSAHL) Algorithm	115
5	MRAN for Communication Channel Equalization	125
5.1	A Review of Channel Equalization	125
5.1.1	Channel Equalization	126
5.1.2	Equalization Methods	129
5.2	MRAN for Equalization	143
5.2.1	Background	143
5.2.2	Linear Channel Equalization	144
5.2.3	Non-Linear Channel Equalization	160
5.2.4	Advantage of MRAN in comparison	168
5.3	MRAN for Co-Channel Interference Equalization	170
5.3.1	Co-Channel Interference Problem	173
5.3.2	MRAN application for CCI	179
5.3.3	Advantage of MRAN in comparison	189
6	Concluding Remarks	191
A	Outline Source Code for MRAN in MATLAB	193
	Bibliography	205
	Index	213

Editorial Reviews

The primary objective of this book is to present a sequential learning algorithm for a radial basis function (RBF) neural network and to evaluate its performance on problems from such areas as function approximation, pattern classification, chaotic time series prediction, nonlinear signal processing, nonlinear dynamic system identification, and communication channel equalization. The authors, after demonstrating the development of the algorithm (dubbed the Minimal Resource Allocation Network), apply it to a range of problems in the above areas and argue that it can, with the most compact network size, achieve better or similar performance than other learning algorithms. Annotation c. Book News, Inc., Portland, OR (booknews.com)

Booknews