Bayesian networks : a practical guide to applications

Bayesian Networks, the result of the convergence of artificial intelligence with statistics, are growing in popularity. Their versatility and modelling power is now employed across a variety of fields for the purposes of analysis, simulation, prediction and diagnosis.

This book provides a general introduction to Bayesian networks, defining and illustrating the basic concepts with pedagogical examples and twenty real-life case studies drawn from a range of fields including medicine, computing, natural sciences and engineering.

Designed to help analysts, engineers, scientists and professionals taking part in complex decision processes to successfully implement Bayesian networks, this book equips readers with proven methods to generate, calibrate, evaluate and validate Bayesian networks.

The book:

Provides the tools to overcome common practical challenges such as the treatment of missing input data, interaction with experts and decision makers, determination of the optimal granularity and size of the model. Highlights the strengths of Bayesian networks whilst also presenting a discussion of their limitations. Compares Bayesian networks with other modelling techniques such as neural networks, fuzzy logic and fault trees. Describes, for ease of comparison, the main features of the major Bayesian network software packages: Netica, Hugin, Elvira and Discoverer, from the point of view of the user. Offers a historical perspective on the subject and analyses future directions for research.

Written by leading experts with practical experience of applying Bayesian networks in finance, banking, medicine, robotics, civil engineering, geology, geography, genetics, forensic science, ecology, and industry, the book has much to offer both practitioners and researchers involved in statistical analysis or modelling in any of these fields.

Author Notes

Editors

OLIVIER POURRET , Electricité de France

PATRICK NAÏM , ELSEWARE, France

BRUCE MARCOT , USDA Forest Service, Oregon, USA

Foreword	p. ix
Preface	p. xi
1 Introduction to Bayesian networks	p. 1
1.1 Models	p. 1
1.2 Probabilistic vs. deterministic models	p. 5
1.3 Unconditional and conditional independence	p. 9
1.4 Bayesian networks	p. 11
2 Medical diagnosis	p. 15
2.1 Bayesian networks in medicine	p. 15
2.2 Context and history	p. 17
2.3 Model construction	p. 19
2.4 Inference	p. 26
2.5 Model validation	p. 28
2.6 Model use	p. 30
2.7 Comparison to other approaches	p. 31
2.8 Conclusions and perspectives	p. 32
3 Clinical decision support	p. 33
3.1 Introduction	p. 33
3.2 Models and methodology	p. 34
3.3 The Busselton network	p. 35
3.4 The PROCAM network	p. 40
3.5 The PROCAM Busselton network	p. 44
3.6 Evaluation	p. 46
3.7 The clinical support tool: TakeHeartII	p. 47
3.8 Conclusion	p. 51
4 Complex genetic models	p. 53
4.1 Introduction	p. 53
4.2 Historical perspectives	p. 54
4.3 Complex traits	p. 56
4.4 Bayesian networks to dissect complex traits	p. 59
4.5 Applications	p. 64
4.6 Future challenges	p. 71
5 Crime risk factors analysis	p. 73
5.1 Introduction	p. 73
5.2 Analysis of the factors affecting crime risk	p. 74
5.3 Expert probabilities elicitation	p. 75
5.4 Data preprocessing	p. 76
5.5 A Bayesian network model	p. 78
5.6 Results	p. 80
5.7 Accuracy assessment	p. 83
5.8 Conclusions	p. 84
6 Spatial dynamics in France	p. 87
6.1 Introduction	p. 87
6.2 An indicator-based analysis	p. 89
6.3 The Bayesian network model	p. 97
6.4 Conclusions	p. 109
7 Inference problems in forensic science	p. 113
7.1 Introduction	p. 113
7.2 Building Bayesian networks for inference	p. 116
7.3 Applications of Bayesian networks in forensic science	p. 120
7.4 Conclusions	p. 126
8 Conservation of marbled murrelets in British Columbia	p. 127
8.1 Context/history	p. 127
8.2 Model construction	p. 129
8.3 Model calibration, validation and use	p. 136
8.4 Conclusions/perspectives	p. 147
9 Classifiers for modeling of mineral potential	p. 149
9.1 Mineral potential mapping	p. 149
9.2 Classifiers for mineral potential mapping	p. 151
9.3 Bayesian network mapping of base metal deposit	p. 157
9.4 Discussion	p. 166
9.5 Conclusions	p. 171
10 Student modeling	p. 173
10.1 Introduction	p. 173
10.2 Probabilistic relational models	p. 175
10.3 Probabilistic relational student model	p. 176
10.4 Case study	p. 180
10.5 Experimental evaluation	p. 182
10.6 Conclusions and future directions	p. 185
11 Sensor validation	p. 187
11.1 Introduction	p. 187
11.2 The problem of sensor validation	p. 188
11.3 Sensor validation algorithm	p. 191
11.4 Gas turbines	p. 197
11.5 Models learned and experimentation	p. 198
11.6 Discussion and conclusion	p. 202
12 An information retrieval system	p. 203
12.1 Introduction	p. 203
12.2 Overview	p. 205
12.3 Bayesian networks and information retrieval	p. 206
12.4 Theoretical foundations	p. 207
12.5 Building the information retrieval system	p. 215
12.6 Conclusion	p. 223
13 Reliability analysis of systems	p. 225
13.1 Introduction	p. 225
13.2 Dynamic fault trees	p. 227
13.3 Dynamic Bayesian networks	p. 228
13.4 A case study: The Hypothetical Sprinkler System	p. 230
13.5 Conclusions	p. 237
14 Terrorism risk management	p. 239
14.1 Introduction	p. 240
14.2 The Risk Influence Network	p. 250
14.3 Software implementation	p. 254
14.4 Site Profiler deployment	p. 259
14.5 Conclusion	p. 261
15 Credit-rating of companies	p. 263
15.1 Introduction	p. 263
15.2 Naive Bayesian classifiers	p. 264
15.3 Example of actual credit-ratings systems	p. 264
15.4 Credit-rating data of Japanese companies	p. 266
15.5 Numerical experiments	p. 267
15.6 Performance comparison of classifiers	p. 273
15.7 Conclusion	p. 276
16 Classification of Chilean wines	p. 279
16.1 Introduction	p. 279
16.2 Experimental setup	p. 281
16.3 Feature extraction methods	p. 285
16.4 Classification results	p. 288
16.5 Conclusions	p. 298
17 Pavement and bridge management	p. 301
17.1 Introduction	p. 301
17.2 Pavement management decisions	p. 302
17.3 Bridge management	p. 307
17.4 Bridge approach embankment - case study	p. 308
17.5 Conclusion	p. 312
18 Complex industrial process operation	p. 313
18.1 Introduction	p. 313
18.2 A methodology for Root Cause Analysis	p. 314
18.3 Pulp and paper application	p. 321
18.4 The ABB Industrial IT platform	p. 325
18.5 Conclusion	p. 326
19 Probability of default for large corporates	p. 329
19.1 Introduction	p. 329
19.2 Model construction	p. 332
19.3 BayesCredit	p. 335
19.4 Model benchmarking	p. 341
19.5 Benefits from technology and software	p. 342
19.6 Conclusion	p. 343
20 Risk management in robotics	p. 345
20.1 Introduction	p. 345
20.2 DeepC	p. 346
20.3 The ADVOCATE II architecture	p. 352
20.4 Model development	p. 354
20.5 Model usage and examples	p. 360
20.6 Benefits from using probabilistic graphical models	p. 361
20.7 Conclusion	p. 362
21 Enhancing Human Cognition	p. 365
21.1 Introduction	p. 365
21.2 Human foreknowledge in everyday settings	p. 366
21.3 Machine foreknowledge	p. 369
21.4 Current application and future research needs	p. 373
21.5 Conclusion	p. 375
22 Conclusion	p. 377
22.1 An artificial intelligence perspective	p. 377
22.2 A rational approach of knowledge	p. 379
22.3 Future challenges	p. 384
Bibliography	p. 385
Index	p. 427

Available:*

On Order

Summary

Summary

Author Notes

Table of Contents