As the power of Bayesian techniques has become more fully realized, the field of artificial intelligence has embraced Bayesian methodology and integrated it to the point where an introduction to Bayesian techniques is now a core course in many computer science programs. Unlike other books on the subject, Bayesian Artificial Intelligence keeps mathematical detail to a minimum and covers a broad range of topics. The authors integrate all of Bayesian net technology and learning Bayesian net technology and apply them both to knowledge engineering. They emphasize understanding and intuition but also provide the algorithms and technical background needed for applications. Software, exercises, and solutions are available on the authors’ website.
Author(s): Kevin B. Korb Ann E. Nicholson
Edition: 1
Publisher: Chapman & Hall\/CRC
Year: 2004
Language: English
Pages: 365
Tags: Информатика и вычислительная техника;Искусственный интеллект;
Table of Contents......Page 0
Bayesian Artificial Intelligence......Page 1
Preface......Page 5
About the Authors......Page 7
Contents......Page 8
List of Figures......Page 15
List of Tables......Page 18
Notation......Page 19
Part I. PROBABILISTIC REASONING......Page 21
Reasoning under uncertainty......Page 22
Uncertainty in AI......Page 23
Probability calculus......Page 24
Conditional probability theorems......Page 27
Variables......Page 28
Interpretations of probability......Page 29
Bayes' theorem......Page 31
Betting and odds......Page 32
Expected utility......Page 34
Dutch books......Page 35
Bayesian reasoning examples......Page 36
The goal of Bayesian AI......Page 40
Are Bayesian networks Bayesian?......Page 41
Bibliographic notes......Page 42
Technical notes......Page 43
Probability Theory......Page 44
BayesÌ Theorem......Page 45
Applications......Page 46
Bayesian network basics......Page 48
Nodes and values......Page 49
Structure......Page 50
Conditional probabilities......Page 51
Reasoning with Bayesian networks......Page 52
Types of reasoning......Page 53
Types of evidence......Page 54
Reasoning with numbers......Page 55
Pearl's network construction algorithm......Page 56
Compactness and node ordering......Page 57
Conditional independence......Page 58
d-separation......Page 60
Asia......Page 62
Summary......Page 63
Bibliographic notes......Page 64
Modeling......Page 66
Conditional Independence......Page 68
d-separation......Page 69
Introduction......Page 72
Two node network......Page 73
Three node chain......Page 74
Exact inference in polytrees......Page 75
Kim and Pearl’s message passing algorithm......Page 76
Message passing example......Page 79
Algorithm features......Page 80
Inference with uncertain evidence......Page 81
Using a virtual node......Page 82
Virtual nodes in the message passing algorithm......Page 84
Clustering methods......Page 85
Junction trees......Page 87
Logic sampling......Page 91
Likelihood weighting......Page 93
Using virtual evidence......Page 94
Assessing approximate inference algorithms......Page 95
Belief revision......Page 96
Probability of evidence......Page 97
Causal inference......Page 98
Bibliographic notes......Page 100
Message passing......Page 101
Virtual / Likelihood evidence......Page 103
Clustering......Page 105
Causal reasoning......Page 106
Utilities......Page 107
Node types......Page 109
Football team example......Page 110
Evaluating decision networks......Page 111
Information links......Page 112
Types of actions......Page 114
Test-action combination......Page 116
Real estate investment example......Page 117
Evaluation using a decision tree model......Page 119
Value of information......Page 121
Dynamic Bayesian networks......Page 122
Nodes, structure and CPTs......Page 123
Reasoning......Page 125
Inference algorithms for DBNs......Page 127
Dynamic decision networks......Page 128
Mobile robot example......Page 129
Summary......Page 130
Bibliographic notes......Page 131
Modeling......Page 132
Dynamic Bayesian networks (DBNs)......Page 134
Introduction......Page 135
BN structures for medical problems......Page 136
Other medical applications......Page 138
Non-medical applications......Page 139
Bayesian poker......Page 140
Five-card stud poker......Page 141
A decision network for poker......Page 142
Betting with randomization......Page 145
Bluffing......Page 146
The domain......Page 147
The DBN model......Page 148
Case-based evaluation......Page 151
An extended sensor model......Page 152
A Nice Argument Generator (NAG)......Page 154
NAG architecture......Page 155
Example: An asteroid strike......Page 156
The psychology of inference......Page 157
Example: The asteroid strike continues......Page 159
Summary......Page 160
Bibliographic notes......Page 161
Problems using the example applications......Page 162
New Applications......Page 163
References......Page 165
Introduction......Page 168
Path models......Page 170
WrightÌs first decomposition rule......Page 172
Learning linear models is complex......Page 176
Conditional independence learners......Page 178
Markov equivalence......Page 181
PC algorithm......Page 184
Causal discovery versus regression......Page 186
Bibliographic notes......Page 187
Technical notes......Page 188
Problems......Page 189
Introduction......Page 192
Parameterizing a binomial model......Page 193
Parameterizing a multinomial model......Page 196
Incomplete data......Page 198
Approximate solutions......Page 199
Causal interaction......Page 204
Noisy-or connections......Page 205
Classification trees and graphs......Page 206
Dual model discovery......Page 208
Bibliographic notes......Page 209
Technical notes......Page 210
Experimental Problem......Page 211
Programming Problems......Page 212
Introduction......Page 213
Cooper & Herskovits’ K2......Page 214
Learning variable order......Page 216
MDL causal discovery......Page 217
Lam and Bacchus’sMDL code for causal models......Page 219
Metric pattern discovery......Page 221
An MML code for causal structures......Page 223
An MML metric for linear models......Page 226
Metropolis search......Page 227
Prior constraints......Page 229
MML models......Page 230
Experimental evaluation......Page 231
Quantitative evaluation......Page 232
Summary......Page 233
Technical notes......Page 234
Evaluation Problem......Page 235
Part III: Knowledge Engineering......Page 236
Bayesian network modeling tasks......Page 238
KEBN lifecycle model......Page 239
Prototyping and spiral KEBN......Page 240
Are BNs suitable for the domain problem?......Page 241
Process management......Page 242
Variables and values......Page 243
Graphical structure......Page 246
Probabilities......Page 254
Local structure......Page 260
Variants of Bayesian networks......Page 263
Modeling example: missing car......Page 264
Decision networks......Page 267
Adaptation......Page 270
Adapting parameters......Page 271
Structural adaptation......Page 272
Bibliographic notes......Page 273
Problems......Page 274
References......Page 275
Sensitivity to evidence......Page 276
Sensitivity to changes in parameters......Page 283
Case-based evaluation......Page 284
Explanation methods......Page 285
Validation methods......Page 286
Predictive accuracy......Page 288
Expected value......Page 289
Kullback-Leibler divergence......Page 290
Information reward......Page 292
Bayesian information reward......Page 293
Summary......Page 295
Technical notes......Page 296
Problems......Page 298
The initial prototype......Page 299
Subsequent developments......Page 300
Ongoing Bayesian poker......Page 301
An intelligent tutoring system for decimal understanding......Page 302
The ITS domain......Page 303
ITS system architecture......Page 305
Expert elicitation......Page 306
Automated methods......Page 313
Field trial evaluation......Page 315
KEBN aspects......Page 316
The seabreeze prediction problem......Page 317
The data......Page 318
Bayesian network modeling......Page 319
Experimental evaluation......Page 320
KEBN aspects......Page 324
Summary......Page 325
Appendix A. Notation......Page 326
Introduction......Page 328
Murphy’s Software Package Survey......Page 329
Analytica......Page 334
BayesiaLab......Page 335
Bayes Net Toolbox (BNT)......Page 336
GeNIe......Page 337
Hugin......Page 338
JavaBayes......Page 339
Netica......Page 340
BUGS......Page 341
CaMML......Page 342
WinMine......Page 343
References......Page 344
