Cellular Neural Networks and Visual Computing: More Information

Cambridge University Press
0521652472 - Cellular Neural Networks and Visual Computing: Foundations and Applications
Leon O. Chua and Tamas Roska
Frontmatter
More information
Cellular neural networks and visual computing

Cellular Nonlinear/neural Network (CNN) technology is both a revolutionary concept and an
experimentally proven new computing paradigm. Analogic cellular computers based on CNNs are
set to change the way analog signals are processed and are paving the way to an entire new analog
computing industry.
This unique undergraduate-level textbook includes many examples and exercises, including CNN
simulator and development software accessible via the Internet. It is an ideal introduction to CNNs
and analogic cellular computing for students, researchers, and engineers from a wide range of
disciplines. Although its prime focus is on visual computing, the concepts and techniques described
in the book will be of great interest to those working in other areas of research, including modeling
of biological, chemical, and physical processes.
Leon Chua is a Professor of Electrical Engineering and Computer Science at the University of
California, Berkeley where he coinvented the CNN in 1988 and holds several patents related to CNN
Technology. He received the Neural Network Pioneer Award, 2000.
Tamas
Roska is a Professor of Information Technology at the Pazmany P. Catholic University of
Budapest and head of the Analogical and Neural Computing Laboratory of the Computer and
Automation Research Institute of the Hungarian Academy of Sciences, Budapest and was an early
pioneer of CNN technology and a coinventor of the CNN Universal Machine as an analogic
supercomputer, He has also spent 12 years as a part-time visiting scholar at the University of
California at Berkeley.
www.cambridge.org

Frontmatter
More information
Cellular neural networks

and visual computing
Foundation and applications
Leon O. Chua
and
Tamas
Roska
www.cambridge.org

Frontmatter
More information
PUBLISHED BY THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge, United Kingdom

CAMBRIDGE UNIVERSITY PRESS
The Edinburgh Building, Cambridge CB2 2RU, UK

40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
Ruiz de Alarcon 13, 28014, Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa
http://www.cambridge.org
c Cambridge University Press 2002

This book is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without
the written permission of Cambridge University Press.
First published 2002
Printed in the United Kingdom at the University Press, Cambridge
Typeface Times 10.5/14pt.
System LATEX 2 [DBD]
A catalogue record of this book is available from the British Library

Library of Congress Cataloguing-in-Publication data
Chua, Leon O., 1936
Cellular neural networks: foundation and primer / Leon O. Chua and Tamas Roska.
p. cm.
Includes bibliographical references and index.
ISBN 0 521 65247 2
1. Neural networks (Computer Science) I. Roska, T. II. Title.
QA.76.87.C494 2002
006.3 2dc21 2001025555
ISBN 0 521 65247 2
hardback
www.cambridge.org

Frontmatter
More information
To our wives, Diana and Zsuzsa
www.cambridge.org

Frontmatter
More information
Contents
Acknowledgements
page xi
Introduction
Notation, definitions, and mathematical foundation
2.1
Basic notation and denitions
2.2
Mathematical foundations
14
Characteristics and analysis of simple CNN templates
35
3.1
Two case studies: the EDGE and EDGEGRAY templates
35
3.2
Three quick steps for sketching the shifted DP plot
49
3.3
Some other useful templates
50
Simulation of the CNN dynamics
100
4.1
Integration of the standard CNN differential equation
100
4.2
Image input
101
4.3
Software simulation
102
4.4
Digital hardware accelerators
110
4.5
Analog CNN implementations
111
4.6
Scaling the signals
113
4.7
Discrete-time CNN (DTCNN)
114
vii
www.cambridge.org

Frontmatter
More information
viii
Contents
Binary CNN characterization via Boolean functions
115
5.1
5.2
5.3
115
122
124
Uncoupled CNNs: unified theory and applications
139
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
6.9
139
140
142
155
156
161
162
166
173
The complete stability phenomenon

Explicit CNN output formula
Proof of completely stable CNN theorem
The primary CNN mosaic
Explicit formula for transient waveform and settling time
Which local Boolean functions are realizable by uncoupled CNNs?
Geometrical interpretations
How to design uncoupled CNNs with prescribed Boolean functions
How to realize non-separable local Boolean functions?
Introduction to the CNN Universal Machine
183
7.1
7.2
7.3
7.4
7.5
184
184
188
190
7.6
7.7
Binary and universal CNN truth table

Boolean and compressed local rules
Optimizing the truth table
Global clock and global wire

Set inclusion
Translation of sets and binary images
Opening and closing and implementing any morphological operator
Implementing any prescribed Boolean transition function by not more than
256 templates
Minimizing the number of templates when implementing any possible
Boolean transition function
Analog-to-digital array converter
195
198
201
Back to basics: Nonlinear dynamics and complete stability
205
8.1
8.2
8.3
8.4
8.5
205
205
210
214
219
A glimpse of things to come

An oscillatory CNN with only two cells
A chaotic CNN with only two cells and one sinusoidal input
Symmetric A template implies complete stability
Positive and sign-symmetric A template implies complete stability
www.cambridge.org

Frontmatter
More information
ix
Contents
8.6
8.7
A
10
11
Positive and cell-linking A template implies complete stability

Stability of some sign-antisymmetric CNNs
Appendix to Chapter 8
224
231
236
The CNN Universal Machine (CNN-UM)
239
9.1
9.2
9.3
9.4
240
244
246
254
The architecture
A simple example in more detail
A very simple example on the circuit level
Language, compiler, operating system
Template design tools
258
10.1
10.2
10.3
10.4
258
260
264
265
Various design techniques

Binary representation, linear separability, and simple decomposition
Template optimization
Template decomposition techniques
CNNs for linear image processing
267
11.1 Linear image processing with B templates is equivalent to spatial convolution with FIR kernels
11.2 Spatial frequency characterization
11.3 A primer on properties and applications of discrete-space Fourier transform (DSFT)
11.4 Linear image processing with A and B templates is equivalent to spatial
convolution with IIR kernels
12
13
267
269
272
272
Coupled CNN with linear synaptic weights
276
12.1
12.2
12.3
12.4
278
283
284
286
Active and inactive cells, dynamic local rules

Binary activation pattern and template format
A simple propagating type example with B/W symmetrical rule
The connectivity problem
Uncoupled standard CNNs with nonlinear synaptic weights
290
13.1 Dynamic equations and DP plot
291
www.cambridge.org

Frontmatter
More information
Contents
14
Standard CNNs with delayed synaptic weights and motion analysis 296
14.1 Dynamic equations
14.2 Motion analysis discrete time and continuous time image acquisition
15
Visual microprocessors analog and digital VLSI implementation 304

of the CNN Universal Machine
15.1
15.2
15.3
15.4
15.5
16
296
297
The analog CNN core

Analogic CNN-UM cell
Emulated digital implementation
The visual microprocessor and its computational infrastructure
Computing power comparison
304
310
312
313
318
CNN models in the visual pathway and the Bionic Eye
320
16.1 Receptive eld organization, synaptic weights, and cloning template

16.2 Some prototype elementary functions and CNN models of the visual
pathway
16.3 A simple qualitative engineering model of a vertebrate retina
16.4 The Bionic Eye implemented on a CNN Universal Machine
321
322
329
338
Notes
Bibliography
Exercises
Appendices
Index
339
348
361
389
390
www.cambridge.org

Frontmatter
More information
Acknowledgements
We started to teach a formal course devoted entirely to CNN only in 1996, in the Spring
Semester, at Berkeley and in the Autumn Semester in Budapest. Since then, several
versions of Lecture Notes have been iterated. We are indebted to many of our former
students some of whom have become our coworkers who have helped us in various
forms we are thankful to all of them. Dr. Akos

Zarandy, Dr. Ken Crounse, Dr. Csaba
Rekecky, Dr. Chai-Wah Wu, Dr. Laszlo Kek, Dr. Laszlo Nemes, Dr. Andras Radvanyi,
and Dr. Peter Szolgay, as well as Tao Yang, An-Shan Huang, David Balya, Katalin
Keseru, Istvan Petras and Istvan Szatmari made special efforts to help us during the
many years of forming the text to this present version. We are also grateful to Phil
Meyler for his kind initiative to publish this textbook.
BerkeleyBudapest, May 2000
xi
www.cambridge.org

Cellular Neural Networks and Visual Computing: More Information

Uploaded by

Copyright:

Available Formats

Cellular Neural Networks and Visual Computing: More Information

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cellular Neural Networks and Visual Computing: More Information

Uploaded by

Copyright:

Available Formats

Cambridge University Press