計算智能：從概念到實現

《計算智能：從概念到實現》全書面向智能系統學科的前沿領域，系統地討論了計算智能的理論、技術及其應用，比較全面地反映了計算智能研究和應用的最新進展。書中涵蓋了模糊控制、神經網絡控制、進化計算以及其他一些技術及應用的內容。全書提供了大量的實用案例，重點強調實際的應用和計算工具，這些對於計算智能領域的進一步發展是非常有意義的。《計算智能:從概念到實現(英文版)》取材新穎，內容深入淺出，材料豐富，理論密切結合實際，具有較高的學術水平和參考價值。全書可作為高等院校相關專業高年級本科生或研究生的教材及參考用書，也可供從事智能科學、自動控制、系統科學、計算機科學、應用數學等領域研究的教師和科研人員參考。

作者：(美國)埃伯哈特 (Russell C.Eberhart) (美國)史玉回 (Yuhui Shi)

Russell C.Eberhart，普度大學電子與計算機工程系主任，IEEE會士。與James Kennedy共同提出了粒子羣優化算法。曾任IEEE神經網絡委員會的主席。除了本書之外。他還著有《羣體智能》（影印版由人民郵電出版社出版）等。

Yuhui Shi（史玉回），國際計算智能領域專家，現任Journal of Swarm Intelligence編委，IEEE CIS羣體智能任務組主席，西交利物浦大學電子與電氣工程系教授。1992年獲東南大學博士學位，先後在美國、韓國、澳大利亞等地從事研究工作，曾任美國電子資訊系統公司專家長達9年。他還是《羣體智能》一書的作者之一。

“這是第一部如此全面的計算智能教科書，包括了大量的實踐示例。”

——Shun-ichi Amari，日本理化研究所腦科學研究機構

“本書強調計算智能的基礎是演化計算，這種全新的視角使其獨樹一幟。本書有非常豐富的實際應用和計算工具，對於計算智能領域的發展意義重大。”

——Xin Yao，伯明翰計算智能與應用研究中心

chapter one Foundations

Definitions

Biological Basis for Neural Networks

Neurons

Biological versus Artificial Neural Networks

Biological Basis for Evolutionary Computation

Chromosomes

Biological versus Artificial Chromosomes

Behavioral Motivations for Fuzzy Logic

Myths about Computational Intelligence

Computational Intelligence Application Areas

Neural Networks

Evolutionary Computation

Fuzzy Logic

Summary

Exercises

chapter two Computational Intelligence

Adaptation

Adaptation versus Learning

Three Types of Adaptation

Three Spaces of Adaptation

Self-organization and Evolution

Evolution beyond Darwin

Historical Views of Computational Intelligence

Computational Intelligence as Adaptation and Self-organization

The Ability to Generalize

Computational Intelligence and Soft Computing versus Artificial Intelligence and Hard Computing

Summary

Exercises

chapter three Evolutionary Computation Concepts and Paradigms

History of Evolutionary Computation

Genetic Algorithms

Evolutionary Programming

Evolution Strategies

Genetic Programming

Particle Swarm Optimization

Toward Unification

Evolutionary Computation Overview

EC Paradigm Attributes

Implementation

Genetic Algorithms

Overview of Genetic Algorithms

A Sample GA Problem

Review of GA Operations in the Simple Example

Schemata and the Schema Theorem

Comments on Genetic Algorithms

Evolutionary Programming

Evolutionary Programming Procedure

Finite State Machine Evolution for Prediction

Function Optimization

Comments on Evolutionary Programming

Evolution Strategies

Selection

Key Issues in Evolution Strategies

Genetic Programming

Particle Swarm Optimization

Developments

Resources

Summary

Exercises

chapter four Evolutionary Computation Implementations

Implementation Issues

Homogeneous versus Heterogeneous Representation

Population Adaptation versus Individual Adaptation

Static versus Dynamic Adaptation

Flowcharts versus Finite State Machines

Handling Multiple Similar Cases

Allocating and Freeing Memory Space

Error Checking

Genetic Algorithm Implementation

Programming Genetic Algorithms

Running the GA Implementation

Particle Swarm Optimization Implementation

Programming the PSO Implementation

Programming the Co-evolutionary PSO

Running the PSO Implementation

Summary

Exercises

chapter five Neural Network Concepts and Paradigms

Neural Network History

Where Did Neural Networks Get Their Name?

The Age of Camelot

The Dark Age

The Renaissance

The Age of Neoconnectionism

The Age of Computational Intelligence

What Neural Networks Are andWhy They Are Useful

Neural Network Components and Terminology

Terminology

Input and Output Patterns

NetworkWeights

Processing Elements

Processing Element Activation Functions

Neural Network Topologies

Terminology

Two-layer Networks

Multilayer Networks

Neural Network Adaptation

Terminology

Hebbian Adaptation

Competitive Adaptation

Multilayer Error Correction Adaptation

Summary of Adaptation Procedures

ComparingNeuralNetworks and Other Information ProcessingMethods

Stochastic Approximation

Kalman Filters

Linear and Nonlinear Regression

Correlation

Bayes Classification

Vector Quantization

Radial Basis Functions

Computational Intelligence

Preprocessing

Selecting Training, Test, and Validation Datasets

Preparing Data

Postprocessing

Denormalization of Output Data

Summary

Exercises

chapter six Neural Network Implementations

Implementation Issues

Topology

Back-propagation Network Initialization and Normalization

LearningVector QuantizerNetwork Initialization andNormalization

Feedforward Calculations for the Back-propagation Network

Feedforward Calculations for the LVQ-I Net

Back-propagation SupervisedAdaptation by Error Back-propagation

LVQ Unsupervised Adaptation Calculations

The LVQ Supervised Adaptation Algorithm

Issues in Evolving Neural Networks

Advantages and Disadvantages of Previous EvolutionaryApproaches

Evolving Neural Networks with Particle Swarm Optimization

Back-propagation Implementation

Programming a Back-propagation Neural Network

Running the Back-propagation Implementation

The Kohonen Network Implementations

Programming the Learning Vector Quantizer

Running the LVQ Implementation

Programming the Self-organizing Feature Map

Running the SOFM Implementation

Evolutionary Back-propagation Network Implementation

Programming the Evolutionary Back-propagation Network

Running the Evolutionary Back-propagation Network

Summary

Exercises

chapter seven Fuzzy Systems Concepts and Paradigms

History

Fuzzy Sets and Fuzzy Logic

Logic, Fuzzy and Otherwise

Fuzziness Is Not Probability

The Theory of Fuzzy Sets

Fuzzy Set Membership Functions

Linguistic Variables

Linguistic Hedges

Approximate Reasoning

Paradoxes in Fuzzy Logic

Equality of Fuzzy Sets

Containment

NOT: The Complement of a Fuzzy Set

AND: The Intersection of Fuzzy Sets

OR: The Union of Fuzzy Sets

Compensatory Operators

Fuzzy Rules

Fuzzification

Fuzzy Rules Fire in Parallel

Defuzzification

Other Defuzzification Methods

Measures of Fuzziness

Developing a Fuzzy Controller

Why Fuzzy Control

A Fuzzy Controller

Building a Mamdani-type Fuzzy Controller

Fuzzy Controller Operation

Takagi-Sugeno-Kang Method

Summary

Exercises

chapter eight Fuzzy Systems Implementations

Implementation Issues

Fuzzy Rule Representation

Evolutionary Design of a Fuzzy Rule System

An Object-oriented Language: C++

Fuzzy Rule System Implementation

Programming Fuzzy Rule Systems

Running the Fuzzy Rule System

Iris Dataset Application

Evolving Fuzzy Rule Systems

Programming the Evolutionary Fuzzy Rule System

Running the Evolutionary Fuzzy Rule System

Summary

Exercises

chapter nine Computational Intelligence Implementations

Implementation Issues

Adaptation of Genetic Algorithms

Fuzzy Adaptation

Knowledge Elicitation

Fuzzy Evolutionary Fuzzy Rule System Implementation

Programming the Fuzzy Evolutionary Fuzzy Rule System

Running the Fuzzy Evolutionary Fuzzy Rule System

Choosing the Best Tools

Strengths andWeaknesses

Modeling and Optimization

Practical Issues

Applying Computational Intelligence to Data Mining

An Example Data Mining System

Summary

Exercises

chapter ten Performance Metrics

General Issues

Selecting Gold Standards

Partitioning the Patterns for Training, Testing, and Validation

Cross Validation

Fitness and Fitness Functions

Parametric and Nonparametric Statistics

Percent Correct

Average Sum-squared Error

Absolute Error

Normalized Error

Evolutionary Algorithm Effectiveness Metrics

Mann-Whitney U Test

Receiver Operating Characteristic Curves

Recall and Precision

Other ROC-related Measures

Confusion Matrices

Chi-square Test

Summary

Exercises

chapter eleven Analysis and Explanation

Sensitivity Analysis

Relation Factors

Zurada Sensitivity Analysis

Evolutionary Computation Sensitivity Analysis

Hinton Diagrams

Computational Intelligence Tools for Explanation Facilities

Explanation Facility Requirements

Neural Network Explanation

Fuzzy Expert System Explanation

Evolutionary Computation Tools for Explanation

An Example Neural Network Explanation Facility

Summary

Exercises

Bibliography

Index

About the Authors