Pattern Recognition, Second Edition

\*Approaches pattern recognition from the designer's point of view \*New edition highlights latest developments in this growing field, including independent components and support vector machines, not available elsewhere \*Supplemented by computer examples selected from applications of interest Pattern recognition is a scientific discipline that is becoming increasingly important in the age of automation and information handling and retrieval. This volume's unifying treatment covers the entire spectrum of pattern recognition applications, from image analysis to speech recognition and communications. This book presents cutting-edge material on neural networks, - a set of linked microprocessors that can form associations and uses pattern recognition to "learn". A direct result of more than 10 years of teaching experience, the text was developed by the authors through use in their own classrooms. \*Approaches pattern recognition from the designer's point of view \*New edition highlights latest developments in this growing field, including independent components and support vector machines, not available elsewhere \*Supplemented by computer examples selected from applications of interest Cover......Page 1 6.3 The Karhunen-Love Transform......Page 0 Title Page......Page 2 Copyright......Page 3 Table of Contents......Page 4 Preface......Page 12 1.1 Is Pattern Recognition Important?......Page 14 1.2 Features, Feature Vectors, and Classifiers......Page 16 1.3 Supervised versus Unsupervised Pattern Recognition......Page 19 1.4 Outline of the Book......Page 21 2.2 Bayes Decision Theory......Page 26 2.3 Discriminant Functions and Decision Surfaces......Page 32 2.4 Bayesian Classification for Normal Distributions......Page 33 2.5 Estimation of Unknown Probability Density Functions......Page 40 2.6 The Nearest Neighbor Rule......Page 57 3.2 Linear Discriminant Functions and Decision Hyperplanes......Page 68 3.3 The Perceptron Algorithm......Page 70 3.4 Least Squares Methods......Page 78 3.5 Mean Square Estimation Revisited......Page 85 3.6 Support Vector Machines......Page 90 4.2 The XOR Problem......Page 106 4.3 The Two-Layer Perceptron......Page 107 4.4 Three-Layer Perceptrons......Page 114 4.5 Algorithms Based on Exact Classification of the Training Set......Page 115 4.6 The Backpropagation Algorithm......Page 117 4.7 Variations on the Backpropagation Theme......Page 125 4.8 The Cost Function Choice......Page 128 4.9 Choice of the Network Size......Page 131 4.10 A Simulation Example......Page 137 4.11 Networks with Weight Sharing......Page 139 4.12 Generalized Linear Classifiers......Page 140 4.13 Capacity of the l-Dimensional Space in Linear Dichotomies......Page 142 4.14 Polynomial Classifiers......Page 144 4.15 Radial Basis Function Networks......Page 146 4.16 Universal Approximators......Page 150 4.17 Support Vector Machines: The Nonlinear Case......Page 152 4.18 Decision Trees......Page 156 4.19 Discussion......Page 163 5.1 Introduction......Page 176 5.2 Preprocessing......Page 177 5.3 Feature Selection Based on Statistical Hypothesis Testing......Page 179 5.4 The Receiver Operating Characteristics CROC Curve......Page 186 5.5 Class Separability Measures......Page 187 5.6 Feature Subset Selection......Page 194 5.7 Optimal Feature Generation......Page 200 5.8 Neural Networks and Feature Generation/Selection......Page 204 5.9 A Hint on the Vapnik-Chernovenkis Learning Theory......Page 206 6.1 Introduction......Page 220 6.2 Basis Vectors and Images......Page 221 6.4 The Singular Value Decomposition......Page 228 6.5 Independent Component Analysis......Page 232 6.6 The Discrete Fourier Transform (DFT)......Page 239 6.7 The Discrete Cosine and Sine Transforms......Page 243 6.8 The Hadamard Transform......Page 244 6.9 The Haar Transform......Page 246 6.10 The Haar Expansion Revisited......Page 248 6.11 Discrete Time Wavelet Transform (DTWT)......Page 252 6.12 The Multiresolution Interpretation......Page 262 6.14 A Look at Two-Dimensional Generalizations......Page 265 6.15 Applications......Page 268 7.1 Introduction......Page 282 7.2 Regional Features......Page 283 7.3 Features for Shape and Size Characterization......Page 307 7.4 A Glimpse at Fractals......Page 316 8.1 Introduction......Page 334 8.2 Measures Based on Optimal Path Searching Techniques......Page 335 8.3 Measures Based on Correlations......Page 350 8.4 Deformable Template Models......Page 356 9.2 The Bayes Classifier......Page 364 9.3 Markov Chain Models......Page 365 9.4 The Viterbi Algorithm......Page 366 9.5 Channel Equalization......Page 369 9.6 Hidden Markov Models......Page 374 9.7 Training Markov Models via Neural Networks......Page 386 9.8 A Discussion of Markov Random Fields......Page 388 10.2 Error Counting Approach......Page 398 10.3 Exploiting the Finite Size of the Data Set......Page 400 10.4 A Case Study from Medical Imaging......Page 403 11.1 Introduction......Page 410 11.2 Proximity Measures......Page 417 12.1 Introduction......Page 442 12.2 Categories of Clustering Algorithms......Page 444 12.3 Sequential Clustering Algorithms......Page 446 12.4 A Modification of BSAS......Page 450 12.5 A Two-Threshold Sequential Scheme......Page 451 12.6 Refinement Stages......Page 454 12.7 Neural Network Implementation......Page 456 13.1 Introduction......Page 462 13.2 Agglomerative Algorithms......Page 463 13.3 The Cophenetic Matrix......Page 489 13.4 Divisive Algorithms......Page 491 13.5 Choice of the Best Number of Clusters......Page 493 14.1 Introduction......Page 502 14.2 Mixture Decomposition Schemes......Page 504 14.3 Fuzzy Clustering Algorithms......Page 513 14.4 Possibilistic Clustering......Page 535 14.5 Hard Clustering Algorithms......Page 542 14.6 Vector Quantization......Page 546 15.2 Clustering Algorithms Based on Graph Theory......Page 558 15.3 Competitive Learning Algorithms......Page 565 15.4 Branch and Bound Clustering Algorithms......Page 574 15.5 Binary Morphology Clustering Algorithms (BMCAs)......Page 577 15.6 Boundary Detection Algorithms......Page 586 15.7 Valley-Seeking Clustering Algorithms......Page 589 15.8 Clustering via Cost Optimization (Revisited)......Page 591 15.9 Clustering Using Genetic Algorithms......Page 595 15.10 Other Clustering Algorithms......Page 596 16.1 Introduction......Page 604 16.2 Hypothesis Testing Revisited......Page 605 16.3 Hypothesis Testing in Cluster Validity......Page 607 16.4 Relative Criteria......Page 618 16.5 Validity of Individual Clusters......Page 634 16.6 Clustering Tendency......Page 637 Appendix A: Hints from Probability and Statistics......Page 656 Appendix B: Linear Algebra Basics......Page 668 Appendix C: Cost Function Optimization......Page 672 Appendix D: Basic Definitions from Linear Systems Theory......Page 690 Index......Page 694 Back Cover......Page 706 Pattern recognition is a scientific discipline that is becoming increasingly important in the age of automation and information handling and retrieval. Patter Recognition, 2e covers the entire spectrum of pattern recognition applications, from image analysis to speech recognition and communications. This book presents cutting-edge material on neural networks, - a set of linked microprocessors that can form associations and uses pattern recognition to "learn" -and enhances student motivation by approaching pattern recognition from the designer's point of view. A direct result of more than 10 years of teaching experience, the text was developed by the authors through use in their own classrooms.

*Approaches pattern recognition from the designer's point of view
*New edition highlights latest developments in this growing field, including independent components and support vector machines, not available elsewhere
*Supplemented by computer examples selected from applications of interest Pattern recognition is a scientific discipline that is becoming increasingly important in the age of automation and information handling and retrieval. Patter Recognition, 2e covers the entire spectrum of pattern recognition applications, from image analysis to speech recognition and communications. This book presents cutting-edge material on neural networks, - a set of linked microprocessors that can form associations and uses pattern recognition to "learn" -and enhances student motivation by approaching pattern recognition from the designer's point of view. A direct result of more than 10 years of teaching experience, the text was developed by the authors through use in their own classrooms. *Approaches pattern recognition from the designer's point of view *New edition highlights latest developments in this growing field, including independent components and support vector machines, not available elsewhere *Supplemented by computer examples selected from applications of interest Pattern recognition is the scientific discipline whose goal is the classification of objects into a number of categories or classes.