Algorithms and Architectures

Note: Supplemental materials are not guaranteed with Rental or Used book purchases.

ISBN: 9780124438613 | 012443861X
Cover: Hardcover
Copyright: 10/13/1997

Buy New

Print on Demand: 2-4 Weeks. This item cannot be cancelled or returned.

$124.61
eBook

Available Instantly

Online: 1825 Days

Downloadable: Lifetime Access

*To support the delivery of the digital material to you, a digital delivery fee of $3.99 will be charged on each digital item.

$145.20*

This volume is the first diverse and comprehensive treatment of algorithms and architectures for the realization of neural network systems. It presents techniques and diverse methods in numerous areas of this broad subject. The book covers major neural network systems structures for achieving effective systems, and illustrates them with examples. this volume includes Radial Basis Function networks, the Expand-and-Truncate Learning algorithm for the synthesis of Three-Layer Threshold Networks, weight initialization, fast and efficient variants of Hamming and Hopfield neural networks, discrete time synchronous multilevel neural systems with reduced VLSI demands, probabilistic design techniques, time-based techniques, techniques for reducing physical realization requirements, and applications to finite constraint problems. a unique and comprehensive reference for a broad array of algorithms and architectures, this book will be of use to practitioners, researchers, and students in industrial, manufacturing, electrical, and mechanical engineering, as well as in computer science and engineering.

Contributors

(4)

Preface

xix

Statistical Theories of Learning in Radial Basis Function Networks

(61)

Jason A. S. Freeman

Mark J. L. Orr

David Saad

I. Introduction

(3)

A. Radial Basis Function Network

(2)

II. Learning in Radial Basis Function Networks

(17)

A. Supervised Learning

(1)

B. Linear Models

(4)

C. Bias and Variance

(2)

D. Cross-Validation

(2)

E. Ridge Regression

(4)

F. Forward Selection

(2)

G. Conclusion

(2)

III. Theoretical Evaluations of Network Performance

(19)

A. Bayesian and Statistical Mechanics Approaches

(10)

B. Probably Approximately Correct Framework

(6)

C. Approximation Error/Estimation Error

(2)

D. Conclusion

(1)

IV. Fully Adaptive Training--An Exact Analysis

(14)

A. On-Line Learning in Radial Basis Function Networks

(1)

B. Generalization Error and System Dynamics

(1)

C. Numerical Solutions

(2)

D. Phenomenological Observations

(2)

E. Symmetric Phase

(2)

F. Convergence Phase

(1)

G. Quantifying the Variances

(2)

H. Simulations

(1)

I. Conclusion

(2)

V. Summary

(1)

Appendix

(2)

References

(5)

Synthesis of Three-Layer Threshold Networks

(25)

Jung Hwan Kim

Sung-Kwon Park

Hyunseo Oh

Youngnam Han

I. Introduction

(1)

II. Preliminaries

(1)

III. Finding the Hidden Layer

(9)

IV. Learning an Output Layer

(4)

V. Examples

(7)

A. Approximation of a Circular Region

(3)

B. Parity Function

(3)

C. 7-Bit Function

(1)

VI. Discussion

(1)

VII. Conclusion

(1)

References

(1)

Weight Initialization Techniques

(36)

Mikko Lehtokangas

Petri Salmela

Jukka Saarinen

Kimmo Kaski

I. Introduction

(2)

II. Feedforward Neural Network Models

(1)

A. Multilayer Perceptron Networks

(1)

B. Radial Basis Function Networks

(1)

III. Stepwise Regression for Weight Initialization

(2)

IV. Initialization of Multilayer Perceptron Networks

(6)

A. Orthogonal Least Squares Method

(1)

B. Maximum Covariance Method

(1)

C. Benchmark Experiments

(5)

V. Initial Training for Radial Basis Function Networks

(5)

A. Stepwise Hidden Node Selection

(1)

B. Benchmark Experiments

(4)

VI. Weight Initialization in Speech Recognition Application

103

(13)

A. Speech Signals and Recognition

103

(1)

B. Principle of the Classifier

104

(2)

C. Training the Hybrid Classifier

106

(3)

D. Results

109

(7)

VII. Conclusion

116

(1)

Appendix I: Chessboard 4 X 4

116

(1)

Appendix II: Two Spirals

117

(1)

Appendix III: GaAs MESFET

117

(1)

Appendix IV: Credit Card

117

(1)

References

118

(5)

Fast Computation in Hamming and Hopfield Networks

123

(32)

Isaac Meilijson

Eytan Ruppin

Moshe Sipper

I. General Introduction

123

(1)

II. Threshold Hamming Networks

124

(11)

A. Introduction

124

(2)

B. Threshold Hamming Network

126

(2)

C. Hamming Network and an Optimal Threshold Hamming Network

128

(4)

D. Numerical Results

132

(2)

E. Final Remarks

134

(1)

III. Two-Iteration Optimal Signaling in Hopfield Networks

135

(17)

A. Introduction

135

(2)

B. Model

137

(3)

C. Rationale for Nonmonotone Bayesian Signaling

140

(2)

D. Performance

142

(4)

E. Optimal Signaling and Performance

146

(2)

F. Results

148

(3)

G. Discussion

151

(1)

IV. Concluding Remarks

152

(1)

References

153

(2)

Multilevel Neurons

155

(26)

J. Si

A. N. Michel

I. Introduction

155

(2)

II. Neural System Analysis

157

(10)

A. Neuron Models

158

(2)

B. Neural Networks

160

(2)

C. Stability of an Equilibrium

162

(2)

D. Global Stability Results

164

(3)

III. Neural System Synthesis for Associative Memories

167

(4)

A. System Constraints

168

(2)

B. Synthesis Procedure

170

(1)

IV. Simulations

171

(2)

V. Conclusions and Discussions

173

(1)

Appendix

173

(5)

References

178

(3)

Probabilistic Design

181

(50)

Sumio Watanabe

Kenji Fukumizu

I. Introduction

181

(1)

II. Unified Framework of Neural Networks

182

(7)

A. Definition

182

(3)

B. Learning in Artificial Neural Networks

185

(4)

III. Probabilistic Design of Layered Neural Networks

189

(8)

A. Neural Network That Finds Unknown Inputs

189

(3)

B. Neural Network That Can Tell the Reliability of Its Own Inference

192

(4)

C. Neural Network That Can Illustrate Input Patterns for a Given Category

196

(1)

IV. Probability Competition Neural Networks

197

(21)

A. Probability Competition Neural Network Model and Its Properties

198

(5)

B. Learning Algorithms for a Probability Competition Neural Network

203

(7)

C. Applications of the Probability Competition Neural Network Model

210

(8)

V. Statistical Techniques for Neural Network Design

218

(10)

A. Information Criterion for the Steepest Descent

218

(7)

B. Active Learning

225

(3)

VI. Conclusion

228

(1)

References

228

(3)

Short Time Memory Problems

231

(30)

M. Daniel Tom

Manoel Fernando Tenorio

I. Introduction

231

(1)

II. Background

232

(1)

III. Measuring Neural Responses

233

(1)

IV. Hysteresis Model

234

(3)

V. Perfect Memory

237

(2)

VI. Temporal Precedence Differentiation

239

(2)

VII. Study in Spatiotemporal Pattern Recognition

241

(4)

VIII. Conclusion

245

(1)

Appendix

246

(14)

References

260

(1)

Reliability Issue and Quantization Effects in Optical and Electronic Network Implementations of Hebbian-Type Associative Memories

261

(32)

Pau-Choo Chung

Ching-Tsorng Tsai

I. Introduction

261

(3)

II. Hebbian-Type Associative Memories

264

(2)

A. Linear-Order Associative Memories

264

(2)

B. Quadratic-Order Associative Memories

266

(1)

III. Network Analysis Using a Signal-to-Noise Ratio Concept

266

(2)

IV. Reliability Effects in Network Implementations

268

(10)

A. Open-Circuit Effects

269

(5)

B. Short-Circuit Effects

274

(4)

V. Comparison of Linear and Quadratic Networks

278

(3)

VI. Quantization of Synaptic Interconnections

281

(7)

A. Three-Level Quantization

282

(4)

B. Three-Level Quantization with Conserved Interconnections

286

(2)

VII. Conclusions

288

(1)

References

289

(4)

Finite Constraint Satisfaction

293

(71)

Angelo Monfroglio

I. Constrained Heuristic Search and Neural Networks for Finite Constraint Satisfaction Problems

293

(30)

A. Introduction

293

(2)

B. Shared Resource Allocation Algorithm

295

(5)

C. Satisfaction of a Conjunctive Normal Form

300

(5)

D. Connectionist Networks for Solving n-Conjunctive Normal Form Satisfiability Problems

305

(6)

E. Other Connectionist Paradigms

311

(6)

F. Network Performance Summary

317

(6)

II. Linear Programming and Neural Networks

323

(8)

A. Conjunctive Normal Form Satisfaction and Linear Programming

324

(5)

B. Connectionist Networks That Learn to Choose the Position of Pivot Operations

329

(2)

III. Neural Networks and Genetic Algorithms

331

(10)

A. Neural Network

332

(4)

B. Genetic Algorithm for Optimizing the Neural Network

336

(1)

C. Comparison with Conventional Linear Programming Algorithms and Standard Constraint Propagation and Search Techniques

337

(3)

D. Testing Data Base

340

(1)

IV. Related Work, Limitations, Further Work, and Conclusions

341

(1)

Appendix I. Formal Description of the Shared Resource Allocation Algorithm

342

(4)

Appendix II. Formal Description of the Conjunctive Normal Form Satisfiability Algorithm

346

(2)

A. Discussion

348

(1)

Appendix III. A 3-CNF-SAT Example

348

(2)

Appendix IV. Outline of Proof for the Linear Programming Algorithm

350

(9)

A. Preliminary Considerations

350

(7)

B. Interior Point Methods

357

(1)

C. Correctness and Completeness

358

(1)

References

359

(5)

Parallel, Self-Organizing, Hierarchical Neural Network Systems

364

(38)

O. K. Ersoy

I. Introduction

364

(2)

II. Nonlinear Transformations of Input Vectors

366

(1)

A. Binary Input Data

366

(1)

B. Analog Input Data

366

(1)

C. Other Transformations

367

(1)

III. Training, Testing, and Error-Detection Bounds

367

(4)

A. Training

367

(1)

B. Testing

368

(1)

C. Detection of Potential Errors

368

(3)

IV. Interpretation of the Error-Detection Bounds

371

(2)

V. Comparison between the Parallel, Self-Organizing, Hierarchical Neural Network, the Backpropagation Network, and the Maximum Likelihood Method

373

(6)

A. Normally Distributed Data

374

(5)

B. Uniformly Distributed Data

379

(1)

VI. PNS Modules

379

(2)

VII. Parallel Consensual Neural Networks

381

(4)

A. Consensus Theory

382

(1)

B. Implementation

383

(1)

C. Optimal Weights

384

(1)

D. Experimental Results

385

(1)

VIII. Parallel, Self-Organizing, Hierarchical Neural Networks with Competitive Learning and Safe Rejection Schemes

385

(7)

A. Safe Rejection Schemes

387

(2)

B. Training

389

(1)

C. Testing

390

(2)

D. Experimental Results

392

(1)

IX. Parallel, Self-Organizing, Hierarchical Neural Networks with Continuous Inputs and Outputs

392

(3)

A. Learning of Input Nonlinearities by Revised Backpropagation

393

(1)

B. Forward-Backward Training

394

(1)

X. Recent Applications

395

(4)

A. Fuzzy Input Signal Representation

395

(2)

B. Multiresolution Image Compression

397

(2)

XI. Conclusions

399

(1)

References

399

(3)

Dynamics of Networks of Biological Neurons: Simulation and Experimental Tools

402

(23)

M. Bove

M. Giugliano

M. Grattarola

S. Martinoia

G. Massobrio

I. Introduction

402

(1)

II. Modeling Tools

403

(15)

A. Conductance-Based Single-Compartment Differential Model Neurons

403

(6)

B. Integrate-and-Fire Model Neurons

409

(3)

C. Synaptic Modeling

412

(6)

III. Arrays of Planar Microtransducers for Electrical Activity Recording of Cultured Neuronal Populations

418

(3)

A. Neuronal Cell Cultures Growing on Substrate Planar Microtransducers

419

(1)

B. Example of a Multisite Electrical Signal Recording from Neuronal Cultures by Using Planar Microtransducer Arrays and Its Simulations

420

(1)

VI. Concluding Remarks

421

(1)

References

422

(3)

Estimating the Dimensions of Manifolds Using Delaunay Diagrams

425

(32)

Yun-Chung Chu

I. Delaunay Diagrams of Manifolds

425

(10)

II. Estimating the Dimensions of Manifolds

435

(20)

III. Conclusions

455

(1)

References

456

(1)

Index

457

Please wait while the item is added to your bag...

FREE SHIPPING

Rent More, Save More! Use code: BBRENTAL

5% off 1 book, 7% off 2 books, 10% off 3+ books

Algorithms and Architectures

Summary

Table of Contents

Supplemental Materials

FREE SHIPPING

Rent More, Save More! Use code: BBRENTAL

5% off 1 book, 7% off 2 books, 10% off 3+ books

Algorithms and Architectures

Summary

Table of Contents

Supplemental Materials

Digital License