Pattern Recognition and Neural Networks

✍ Scribed by Brian D. Ripley

• Publisher: Cambridge University Press
• Year: 2008
• Tongue: English
• Leaves: 415
• Edition: 1
• Category: Library

✦ Synopsis

Ripley brings together two crucial ideas in pattern recognition: statistical methods and machine learning via neural networks. He brings unifying principles to the fore, and reviews the state of the subject. Ripley also includes many examples to illustrate real problems in pattern recognition and how to overcome them.

✦ Table of Contents

Cover
......Page 1
Back Cover
......Page 3
Contents
......Page 5
Preface......Page 9
Notation
......Page 12
1
Introduction and Examples......Page 13
1.1 How do neural methods differ?......Page 16
1.2 The pattern recognition task......Page 17
1.3 Overview of the remaining chapters......Page 21
1.4 Examples......Page 22
1.5 Literature......Page 27
2
Statistical Decision Theory......Page 29
2.1 Bayes rules for known distributions......Page 30
2.2 Parametric models......Page 38
2.3 Logistic discrimination......Page 55
2.4 Predictive classification......Page 57
2.5 Alternative estimation procedures......Page 67
2.6 How complex a model do we need?......Page 71
2. 7 Performance assessment......Page 78
2.8 Computational learning approaches......Page 89
3
Linear Discriminant Analysis......Page 103
3.1 Classical linear discrimination......Page 104
3.2 Linear discriminants via regression......Page 113
3.3 Robustness......Page 117
3.4 Shrinkage methods......Page 118
3.5 Logistic discrimination......Page 121
3.6 Linear separation and perceptrons......Page 128
4
Flexible Discriminants......Page 133
4.1 Fitting smooth parametric functions......Page 134
4.2 Radial basis functions......Page 143
4.3 Regularization......Page 148
5
Feed-forward Neural Networks......Page 155
5.1 Biological motivation......Page 157
5.2 Theory......Page 159
5.3 Learning algorithms......Page 160
5.4 Examples......Page 172
5.5 Bayesian perspectives......Page 175
5.6 Network complexity......Page 180
5.7 Approximation results......Page 185
6.1 Non-parametric estimation of class densities......Page 193
6.2 Nearest neighbour methods......Page 203
6.3 Learning vector quantization......Page 213
6.4 Mixture representations......Page 219
7
Tree-structured Classifiers......Page 225
7.1 Splitting rules......Page 228
7.2 Pruning rules......Page 233
7.3 Missing values......Page 243
7.4 Earlier approaches......Page 247
7.5 Refinements......Page 249
7.6 Relationships to neural networks......Page 252
7.7 Bayesian trees......Page 253
8
Belief Networks......Page 255
8.1 Graphical models and networks......Page 258
8.2 Causal networks......Page 274
8.3 Learning the network structure......Page 287
8.4 Boltzmann machines......Page 291
8.5 Hierarchical mixtures of experts......Page 295
9
Unsupervised Methods......Page 299
9.1 Projection methods......Page 300
9.2 Multidimensional scaling......Page 317
9.3 Clustering algorithms......Page 323
10 Finding Good Pattern
Features......Page 339
10.1 Bounds for the Bayes error......Page 340
10.2 Normal class distributions......Page 341
10.3 Branch-and-bound techniques......Page 342
10.4 Feature extraction......Page 343
A.1 Maximum likelihood and MAP estimation......Page 345
A.2 The EM algorithm......Page 346
A.3 Markov chain Monte Carlo......Page 349
A.4 Axioms for conditional independence......Page 351
A.5 Optimization......Page 354
Glossary......Page 359
References......Page 367
Author Index......Page 403
Subject Index......Page 411
Blank Page......Page 2