Muutke küpsiste eelistusi

E-raamat: Statistical Approach to Neural Networks for Pattern Recognition [Wiley Online]

(CSIRO, Australia)
  • Formaat: 288 pages, Charts: 1 B&W, 0 Color; Photos: 2 B&W, 0 Color; Drawings: 23 B&W, 0 Color; Maps: 7 B&W, 0 Color; Graphs: 64 B&W, 0 Color
  • Sari: Wiley Series in Computational Statistics
  • Ilmumisaeg: 07-Aug-2007
  • Kirjastus: Wiley-Interscience
  • ISBN-10: 470148152
  • ISBN-13: 9780470148150
Teised raamatud teemal:
  • Wiley Online
  • Hind: 141,68 €*
  • * hind, mis tagab piiramatu üheaegsete kasutajate arvuga ligipääsu piiramatuks ajaks
Statistical Approach to Neural Networks for Pattern RecognitionSuurem pilt
  • Formaat: 288 pages, Charts: 1 B&W, 0 Color; Photos: 2 B&W, 0 Color; Drawings: 23 B&W, 0 Color; Maps: 7 B&W, 0 Color; Graphs: 64 B&W, 0 Color
  • Sari: Wiley Series in Computational Statistics
  • Ilmumisaeg: 07-Aug-2007
  • Kirjastus: Wiley-Interscience
  • ISBN-10: 470148152
  • ISBN-13: 9780470148150
Teised raamatud teemal:
Wiley Online e-raamatudRohkem infot Wiley Online kohta
Raamatu kodulehekülg: https://onlinelibrary.wiley.com/doi/book/10.1002/9780470148150
An accessible and up-to-date treatment featuring the connection between neural networks and statistics A Statistical Approach to Neural Networks for Pattern Recognition presents a statistical treatment of the Multilayer Perceptron (MLP), which is the most widely used of the neural network models. This book aims to answer questions that arise when statisticians are first confronted with this type of model, such as:

How robust is the model to outliers?

Could the model be made more robust?

Which points will have a high leverage?

What are good starting values for the fitting algorithm?

Thorough answers to these questions and many more are included, as well as worked examples and selected problems for the reader. Discussions on the use of MLP models with spatial and spectral data are also included. Further treatment of highly important principal aspects of the MLP are provided, such as the robustness of the model in the event of outlying or atypical data; the influence and sensitivity curves of the MLP; why the MLP is a fairly robust model; and modifications to make the MLP more robust. The author also provides clarification of several misconceptions that are prevalent in existing neural network literature.

Throughout the book, the MLP model is extended in several directions to show that a statistical modeling approach can make valuable contributions, and further exploration for fitting MLP models is made possible via the R and S-PLUS® codes that are available on the book's related Web site. A Statistical Approach to Neural Networks for Pattern Recognition successfully connects logistic regression and linear discriminant analysis, thus making it a critical reference and self-study guide for students and professionals alike in the fields of mathematics, statistics, computer science, and electrical engineering.
Notation and Code Examples xi
Preface xiii
Acknowledgments xvii
1 Introduction
1
1.1 The perceptron
4
2 The Multi—Layer Perceptron Model
9
2.1 The multi—layer perceptron (MLP)
9
2.2 The first and second derivatives
12
2.3 Additional hidden layers
14
2.4 Classifiers
15
2.5 Complements and exercises
16
3 Linear Discriminant Analysis
19
3.1 An alternative method
21
3.2 Example
22
3.3 Flexible and penalized LDA
23
3.4 Relationship of MLP models to LDA
26
3.5 Linear classifiers
27
3.6 Complements and exercises
30
4 Activation and Penalty Functions
35
4.1 Introduction
35
4.2 Interpreting outputs as probabilities
35
4.3 The "universal approximator" and consistency
37
4.4 Variance and bias
38
4.5 Binary variables and logistic regression
39
4.6 MLP models and cross-entropy
40
4.7 A derivation of the softmax activation function
43
4.8 The "natural" pairing and Δq
45
4.9 A comparison of least squares and cross-entropy
47
4.10 Conclusion
48
4.11 Complements and exercises
48
5 Model Fitting and Evaluation
53
5.1 Introduction
53
5.2 Error rate estimation
54
5.3 Model selection for MLP models
57
5.4 Penalized training
62
5.5 Complements and exercises
65
6 The Task-based MLP
69
6.1 Introduction
69
6.2 The task-based MLP
70
6.3 Pruning algorithms
71
6.4 Interpreting and evaluating task-based MLP models
76
6.5 Evaluating the models
87
6.6 Conclusion
88
6.7 Complements and exercises
89
7 Incorporating Spatial Information into an MLP Classifier
93
7.1 Allocation and neighbor information
93
7.2 Markov random fields
98
7.3 Hopfield networks
100
7.4 MLP neighbor models
101
7.5 Sequential updating
107
7.6 Example - Martin's farm
109
7.7 Conclusion
111
7.8 Complements and exercises
114
8 Influence Curves for the Multi–layer Perceptron Classifier
121
8.1 Introduction
121
8.2 Estimators
122
8.3 Influence curves
123
8.4 M–estimators
124
8.5 The MLP
128
8.6 Influence curves for pc
136
8.7 Summary and Conclusion
139
9 The Sensitivity Curves of the MLP Classifier
143
9.1 Introduction
143
9.2 The sensitivity curve
144
9.3 Some experiments
145
9.4 Discussion
151
9.5 Conclusion
157
10 A Robust Fitting Procedure for MLP Models 159
10.1 Introduction
159
10.2 The effect of a hidden layer
160
10.3 Comparison of MLP with robust logistic regression
162
10.4 A robust MLP model
166
10.5 Diagnostics
172
10.6 Conclusion
175
10.7 Complements and exercises
176
11 Smoothed Weights 179
11.1 Introduction
179
11.2 MLP models
184
11.3 Examples
187
11.4 Conclusion
198
11.5 Complements and exercises
200
12 Translation Invariance 203
12.1 Introduction
203
12.2 Example 1
205
12.3 Example 2
208
12.4 Example 3
209
12.5 Conclusion
214
13 Fixed-slope Training 219
13.1 Introduction
219
13.2 Strategies
221
13.3 Fixing Y or Ω
222
13.4 Example 1
222
13.5 Example 2
223
13.6 Discussion
223
Bibliography 227
Appendix A: Function Minimization 245
A.1 Introduction
245
A.2 Back-propagation
246
A.3 Newton-Raphson
247
A.4 The method of scoring
249
A.5 Quasi-Newton
250
A.6 Conjugate gradients
250
A.7 Scaled conjugate gradients
252
A.8 Variants on vanilla "back-propagation"
253
A.9 Line search
254
A.10 The simplex algorithm
254
A.11 Implementation
255
A.12 Examples
255
A.13 Discussion and Conclusion
256
Appendix B: Maximum Values of the Influence Curve 261
Topic Index 265


Robert A. Dunne, PhD, is Research Scientist in the Mathematical and Information Sciences Division of the Commonwealth Scientific and Industrial Research Organization (CSIRO) in North Ryde, Australia. Dr. Dunne received his PhD from Murdoch University, and his research interests include remote sensing and bioinformatics.
Püsilink: https://www.kriso.ee/db/9780470148150_pe.html
Märksõnad: