Neural Networks

by Herve Abdi; Dominique Valentin; Betty Edelman

• ISBN13: 9780761914402
• ISBN10: 0761914404
• Format: Paperback
• Copyright: 1998-12-09
• Publisher: Sage Publications

Summary

This book provides the first accessible introduction to neural network analysis as a methodological strategy for social scientists. The author details numerous studies and examples which illustrate the advantages of neural network analysis over other quantitative and modeling methods in widespread use. Methods are presented in an accessible style for readers who do not have a background in computer science. The book provides a history of neural network methods, a substantial review of the literature, detailed applications, coverage of the most common alternative models and examples of two leading software packages for neural network analysis.

Table of Contents

Series Editor's Introduction	v	(1)
Notations	vi
1 Introduction	1	(2)
1.1 What Are Neural Networks?	1	(1)
1.2 Overview of This Book	2	(1)
2 The Perceptron	3	(18)
2.1 Overview	3	(1)
2.2 The Building Block: The McCulloch and Pitts Neuron	4	(4)
2.3 Architecture of a Perceptron	8	(1)
2.4 The Widrow-Hoff Learning Rule	9	(4)
2.5 Learning With +1/ - 1 Cells	13	(4)
2.6 Performance Evaluation	17	(2)
2.7 The Perceptron and Discriminant Analysis	19	(1)
2.8 Learning and Testing Sets: The Validation Problem	20	(1)
3 Linear Autoassociative Memories	21	(25)
3.1 Overview	21	(1)
3.2 The Building Block: The Basic Linear Unit	22	(1)
3.3 Architecture of an Autoassociative Memory	23	(2)
3.4 The Hebbian Learning Rule	25	(4)
3.5 Retrieval of a Learned Pattern	29	(2)
3.6 Limitations of Hebbian Learning	31	(4)
3.7 Generalization to New Stimuli	35	(1)
3.8 Reconstruction Error, Error Term, and Global Error	36	(1)
3.9 The Widrow-Hoff Learning Rule	37	(4)
3.10 Eigen and Singular Value Decompositions: PCA	41	(4)
3.11 The Widrow-Hoff Rule and Gradient Descent	45	(1)
4 Linear Heteroassociative Memories	46	(22)
4.1 Overview	46	(1)
4.2 Architecture of a Heteroassociative Memory	47	(1)
4.3 The Hebbian Learning Rule	48	(6)
4.4 The Widrow-Hoff Learning Rule	54	(4)
4.5 Widrow-Hoff, Pseudo-Inverse, and Linear Regression	58	(2)
4.6 The Widrow-Hoff Learning Rule and Gradient Descent	60	(1)
4.7 Discriminant Analysis and Perceptron Revisited	60	(1)
4.8 Radial Basis Function Networks	61	(7)
5 Error Backpropagation	68	(19)
5.1 Overview	68	(1)
5.2 Architecture and Notation	69	(1)
5.3 The Building Block: The Nonlinear Unit	70	(3)
5.4 The Backpropagation Algorithm	73	(10)
5.5 Performance Analysis	83	(2)
5.6 Backpropagation and Gradient Descent	85	(2)
5.7 Backpropagation and Logistic Regression	87	(1)
6 Useful References	87	(3)
References	88	(2)
About the Authors	90