Dreyfus Neural Networks

Methodology and Applications
1. Auflage 2005
ISBN: 978-3-540-28847-3
Verlag: Springer
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, 498 Seiten, eBook

Neural networks represent a powerful data processing technique that has reached maturity and broad application. When clearly understood and appropriately used, they are a mandatory component in the toolbox of any engineer who wants make the best use of the available data, in order to build models, make predictions, mine data, recognize shapes or signals, etc. Ranging from theoretical foundations to real-life applications, this book is intended to provide engineers and researchers with clear methodologies for taking advantage of neural networks in industrial, financial or banking applications, many instances of which are presented in the book. For the benefit of readers wishing to gain deeper knowledge of the topics, the book features appendices that provide theoretical details for greater insight, and algorithmic details for efficient programming and implementation. The chapters have been written by experts and edited to present a coherent and comprehensive, yet not redundant, practically oriented introduction.

Scientists

Zielgruppe

Research

Autoren/Hrsg.

Dreyfus, Gérard

Inhaltsverzeichnis

1;Preface;5
1.1;Reading Guide;6
2;Contents;8
3;List of Contributors;16
4;1 Neural Networks: An Overview;18
4.1;1.1 Neural Networks: Definitions and Properties;19
4.2;1.2 When and How to Use Neural Networks with Supervised Training;41
4.3;1.3 Feedforward Neural Networks and Discrimination ( Classification);49
4.4;1.4 Some Applications of Neural Networks to Various Areas of Engineering;67
4.5;1.5 Conclusion;93
4.6;1.6 Additional Material;94
4.7;References;97
5;2 Modeling with Neural Networks: Principles and Model Design Methodology;101
5.1;2.1 What Is a Model?;101
5.2;2.2 Elementary Concepts and Vocabulary of Statistics;103
5.3;2.3 Static Black-Box Modeling;108
5.4;2.4 Input Selection for a Static Black-Box Model;111
5.5;2.5 Estimation of the Parameters (Training) of a Static Model;119
5.6;2.6 Model Selection;147
5.7;2.7 Dynamic Black-Box Modeling;165
5.8;2.8 Dynamic Semiphysical (Gray Box) Modeling;191
5.9;2.9 Conclusion: What Tools?;202
5.10;2.10 Additional Material;203
5.11;References;215
6;3 Modeling Methodology: Dimension Reduction and Resampling Methods;218
6.1;3.1 Introduction;218
6.2;3.2 Preprocessing;219
6.3;3.3 Input Dimension Reduction;222
6.4;3.4 Principal Component Analysis;222
6.5;3.5 Curvilinear Component Analysis;226
6.6;3.6 The Bootstrap and Neural Networks;235
6.7;References;245
7;4 Neural Identification of Controlled Dynamical Systems and Recurrent Networks;246
7.1;4.1 Formal Definition and Examples of Discrete-Time Controlled Dynamical Systems;247
7.2;4.2 Regression Modeling of Controlled Dynamical Systems;257
7.3;4.3 On-Line Adaptive Identification and Recursive Prediction Error Method;265
7.4;4.4 Innovation Filtering in a State Model;273
7.5;4.5 Recurrent Neural Networks;285
7.6;4.6 Learning for Recurrent Networks;291
7.7;4.7 Appendix (Algorithms and Theoretical Developments);298
7.8;References;302
8;5 Closed-Loop Control Learning;303
8.1;5.1 Generic Issues in Closed-Loop Control of Nonlinear Systems;304
8.2;5.2 Design of a Neural Control with an Inverse Model;308
8.3;5.3 Dynamic Programming and Optimal Control;317
8.4;5.4 Reinforcement Learning and Neuro-Dynamic Programming;328
8.5;References;339
9;6 Discrimination;342
9.1;6.1 Training for Pattern Discrimination;343
9.2;6.2 Linear Separation: The Perceptron;347
9.3;6.3 The Geometry of Classification;349
9.4;6.4 Training Algorithms for the Perceptron;352
9.5;6.5 Beyond Linear Separation;368
9.6;6.6 Problems with More than two Classes;375
9.7;6.7 Theoretical Questions;377
9.8;6.8 Additional Theoretical Material;387
9.9;References;389
10;7 Self-Organizing Maps and Unsupervised Classification;391
10.1;7.1 Notations and Definitions;393
10.2;7.2 The k-Means Algorithm;395
10.3;7.3 Self-Organizing Topological Maps;404
10.4;7.4 Classification and Topological Maps;427
10.5;7.5 Applications;433
10.6;References;453
11;8 Neural Networks without Training for Optimization;455
11.1;8.1 Modelling an Optimisation Problem;455
11.2;8.2 Complexity of an Optimization Problem;458
11.3;8.3 Classical Approaches to Combinatorial Problems;459
11.4;8.4 Introduction to Metaheuristics;460
11.5;8.5 Techniques Derived from Statistical Physics;461
11.6;8.6 Neural Approaches;475
11.7;8.7 Tabu Search;496
11.8;8.8 Genetic Algorithms;496
11.9;8.9 Towards Hybrid Approaches;497
11.10;8.10 Conclusion;497
11.11;References;498
12;About the Authors;503
13;Index;505

Neural Networks: An Overview.- Modeling with Neural Networks: Principles and Model Design Methodology.- Modeling Metholodgy: Dimension Reduction and Resampling Methods.- Neural Identification of Controlled Dynamical Systems and Recurrent Networks.- Closed-Loop Control Learning.- Discrimination.- Self-Organizing Maps and Unsupervised Classification.- Neural Networks without Training for Optimization.

Über Autor(innen)

Scientists