An Introduction to Natural Computation / Edition 1

by Dana H. Ballard

Overview

This book provides a comprehensive introduction to the computational material that forms the underpinnings of the currently evolving set of brain models.

It is now clear that the brain is unlikely to be understood without recourse to computational theories. The theme of An Introduction to Natural Computation is that ideas from diverse areas such as neuroscience, information theory, and optimization theory have recently been extended in ways that make them useful for describing the brains programs. This book provides a comprehensive introduction to the computational material that forms the underpinnings of the currently evolving set of brain models. It stresses the broad spectrum of learning models--ranging from neural network learning through reinforcement learning to genetic learning--and situates the various models in their appropriate neural context.

To write about models of the brain before the brain is fully understood is a delicate matter. Very detailed models of the neural circuitry risk losing track of the task the brain is trying to solve. At the other extreme, models that represent cognitive constructs can be so abstract that they lose all relationship to neurobiology. An Introduction to Natural Computation takes the middle ground and stresses the computational task while staying near the neurobiology.

Product Details

ISBN-13:	9780262522588
Publisher:	MIT Press
Publication date:	01/22/1999
Series:	Complex Adaptive Systems
Edition description:	Reprint
Pages:	336
Product dimensions:	7.00(w) x 10.00(h) x 0.80(d)
Age Range:	18 Years

About the Author

Dana H. Ballard is Professor of Computer Science at the University of Texas at Austin.

Table of Contents

List of Figures
List of Tables
Preface
1 Natural Computation
1.1 Introduction
1.2 The Brain
1.3 Computational Theory
1.4 Elements of Natural Computation
1.5 Overview
1.6 The Grand Challenge
I Core Concepts
2 Fitness
2.1 Introduction
2.2 Bayes' Rule
2.3 Probability Distributions
2.4 Information Theory
2.5 Classification
2.6 Minimum Description Length
Appendix: Laws of Probability
Notes
Exercises
3 Programs
3.1 Introduction
3.2 Heuristic Search
3.3 Two-Person Games
3.4 Biological State Spaces
Notes
Exercises
4 Data
4.1 Data Compression
4.2 Coordinate Systems
4.3 Eigenvalues and Eigenvectors
4.4 Random Vectors
4.5 High-Dimensional Spaces
4.6 Clustering
Appendix: Linear Algebra Review
Notes
Exercises
5 Dynamics
5.1 Overview
5.2 Linear Systems
5.3 Nonlinear Systems
Appendix: Taylor Series
Notes
Exercises
6 Optimization
6.1 Introduction
6.2 Minimization Algorithms
6.3 The Method of Language Multipliers
6.4 Optimal Control
Notes
Exercises
II Memories
7 Content-Addressable Memory
7.1 Introduction
7.2 Hopfield Memories
7.3 Kanerva Memories
7.4 Radial Basis Functions
7.5 Kalman Filtering
Notes
Exercises
8 Supervised Learning
8.1 Introduction
8.2 Perceptrons
8.3 Continuous Activation Functions
8.4 Recurrent Networks
8.5 Minimum Description Length
8.6 The Activation Function
Notes
Exercises
9 Unsupervised Learning
9.1 Introduction
9.2 Principal Components
9.3 Competitive Learning
9.4 Topological Constraints
9.5 Supervised Competitive Learning
9.6 Multimodal Data
9.7Independent Components
Notes
Exercises
III Programs
10 Markov Models
10.1 Introduction
10.2 Markov Models
10.3 Hidden Markov Models
Notes
Exercises
11 Reinforcement Learning
11.1 Introduction
11.2 Markov Decision Process
11.3 The Core Idea: Policy Improvement
11.4 Q-Learning
11.5 Temporal-Difference Learning
11.6 Learning with a Teacher
11.7 Partially Observable MDPs
11.8 Summary
Notes
Exercises
IV Systems
12 Genetic Algorithms
12.1 Introduction
12.2 Schemata
12.3 Determining Fitness
13 Genetic Programming
13.1 Introduction
13.2 Genetic Operators for Programs
13.3 Genetic Programming
13.4 Analysis
13.5 Modules
13.6 Summary
Notes
Exercises
14 Summary
14.1 Learning to React: Memories
14.2 Learning During a Lifetime: Programs
14.3 Learniing Across Generations: Systems
14.4 The Grand Challenge Revisited
Notes
Index

What People are Saying About This

Andy Barto

Ballard has writtena a lucid introductory text covering a collection of material that is unusual by present standards but that is likely to form an indispensible core of future advances in what Ballard calls natural computation. The great activity over the last decade in biologically-related computation has overrun some of the old disciplinary boundaries, leaving uncertainty as to what one should know to appreciate— as well as to participate— in this active research area. Ballard introduces a collection of topics that would be hard to access without taking a half dozen courses in computer science, applied mathematics, and systems engineering. He shows how these topics all participate in a unified and original view of natural computation. If I had access to this book whn my interest in natural computation was aroused as an undergraduate, it would have saved me a lot of time. I envy today's students chance to study, in one course, this collection of essential material.

Richard Sutton

An Introduction to Natural Computation could seve as an introductory textbook for undergraduate courses surveying computational aspects of biological systems. It covers a lot of important topics in this area. Ballard is an excellent researcher with a broad, powerful view.

Endorsement

I am EXTREMELY enthusiastic about this work. This is the first such book I've seeb that comes even close to covering the topic. I'd love to use it for teahing.

—Alex (Sandy) Pentland, Academic Head, The Media Lab, M.I.T. Toshiba Professor of Media Arts and Sciences

From the Publisher

This is a wonderful book that brings together in one place the modern view of computation as found in nature. It is well written and has soemthing or everyone from the undergraduate to the advanced researcher.

—Terrence J. Sejnowski, Howard Hughes Medical Institute at The Salk Institute for Biological Studies, La Jolla, California

Ballard's text offers clear, direct introductions to the key tools and concepts needed in contemporary approached to AI, including state spaces, dynamics, memory models, reinforcement learning, and evolutionary algorithms. All contribute to the book's central aim: to undestand the computations that permit the survival and successful adaptive behavior of natural systems.

—Stewart W. Wilson, International Society for Adaptive Behavior, and The Rowland Institute for Science

Ballard has writtena a lucid introductory text covering a collection of material that is unusual by present standards but that is likely to form an indispensible core of future advances in what Ballard calls natural computation. The great activity over the last decade in biologically-related computation has overrun some of the old disciplinary boundaries, leaving uncertainty as to what one should know to appreciate—as well as to participate—in this active research area. Ballard introduces a collection of topics that would be hard to access without taking a half dozen courses in computer science, applied mathematics, and systems engineering. He shows how these topics all participate in a unified and original view of natural computation. If I had access to this book whn my interest in natural computation was aroused as an undergraduate, it would have saved me a lot of time. I envy today's students chance to study, in one course, this collection of essential material.

—Andy Barto, Professor of Computer Science, University of Massechusetts at Amherst

An Introduction to Natural Computation could seve as an introductory textbook for undergraduate courses surveying computational aspects of biological systems. It covers a lot of important topics in this area. Ballard is an excellent researcher with a broad, powerful view.

—Richard Sutton, Senior Research Scientist, Department of Computer Science, University of Massechusetts

I am EXTREMELY enthusiastic about this work. This is the first such book I've seeb that comes even close to covering the topic. I'd love to use it for teahing.

—Alex (Sandy) Pentland, Academic Head, The Media Lab, M.I.T. Toshiba Professor of Media Arts and Sciences

Stewart W. Wilson

Ballard's text offers clear, direct introductions to the key tools and concepts needed in contemporary approached to AI, including state spaces, dynamics, memory models, reinforcement learning, and evolutionary algorithms. All contribute to the book's central aim: to undestand the computations that permit the survival and successful adaptive behavior of natural systems.

Terrence J. Sejnowski

This is a wonderful book that brings together in one place the modern view of computation as found in nature. It is well written and has soemthing or everyone from the undergraduate to the advanced researcher.

Alex (Sandy) Pentland

I am EXTREMELY enthusiastic about this work. This is the first such book I've seeb that comes even close to covering the topic. I'd love to use it for teahing.