Uncertainty in Artificial Intelligence 5

Description

This volume, like its predecessors, reflects the cutting edge of research on the automation of reasoning under uncertainty.

A more pragmatic emphasis is evident, for although some papers address fundamental issues, the majority address practical issues. Topics include the relations between alternative formalisms (including possibilistic reasoning), Dempster-Shafer belief functions, non-monotonic reasoning, Bayesian and decision theoretic schemes, and new inference techniques for belief nets. New techniques are applied to important problems in medicine, vision, robotics, and natural language understanding.

Front Cover 2
Uncertainty in Artificial Intelligence 5 5
Copyright Page 6
Preface 7
Table of Contents 9
Reviewers 13
Program Committee 13
Contributors 15
PART I: FUNDAMENTAL ISSUES 17
- Chapter 1. Lp—A Logic for Statistical Information 19
  - 1 Introduction 19
  - 2 Other Probability Logics 20
  - 3 Types of Statistical Knowledge 20
  - 4 Syntax and Semantics 21
  - 5 Syntax 22

Front Cover 2
Uncertainty in Artificial Intelligence 5 5
Copyright Page 6
Preface 7
Table of Contents 9
Reviewers 13
Program Committee 13
Contributors 15
PART I: FUNDAMENTAL ISSUES 17
- Chapter 1. Lp—A Logic for Statistical Information 19
  - 1 Introduction 19
  - 2 Other Probability Logics 20
  - 3 Types of Statistical Knowledge 20
  - 4 Syntax and Semantics 21
  - 5 Syntax 22
  - 6 Examples of Representation 24
  - 7 Deductive Proof Theory 25
  - 8 Degrees of Belief 28
  - Acknowledgments 29
  - References 29
- CHAPTER 2. REPRESENTING TIME IN CAUSAL PROBABILISTIC NETWORKS 31
  - 1 INTRODUCTION1 31
  - 2 THE DISTRIBUTION OF TIME 32
  - 3 MARKED POINT PROCESS REPRESENTATION 33
  - 4 NETWORKS OF "DATES" 35
  - References 44
  - Acknowledgements 44
- CHAPTER 3. CONSTRUCTING THE PIGNISTIC PROBABILITY FUNCTION IN A CONTEXT OF UNCERTAINTY 45
  - 1. Introduction 45
  - 2. The credibility function 46
  - 3. a-combined credibility spaces 47
  - 4. The pignistic probability function 50
  - 5. Co-credibility function 53
  - 6. The Moebius transformations of Cr 53
  - 7. Conclusions 54
  - Bibliography 54
  - Acknowledgements 55
- Chapter 4. Can Uncertainty Management Be Realized In A Finite Totally Ordered Probability Algebra? 57
  - 1 Introduction 57
  - 2 Finite totally ordered probability algebras 58
  - 3 Bayes theorem and reasoning by case 63
  - 4 Problems with legal finite totally ordered probability 64
  - 5 An experiment 67
  - References 69
  - 6 Conclusion 69
  - Acknowledgements 69
  - Appendix A: Derivation of 70
  - Appendix B: Examples of legal FTOPAs 71
  - Appendix C 72
- PART Il: DEFEASIBLE REASONING AND UNCERTAINTY 75
  - Chapter 5. Defeasible Reasoning and Uncertainty: Comments 77
    - 1 Overview 77
    - 3 Bonissone et al 78
    - 2 Goldszmidt & Pearl 78
    - 4 Loui 80
    - Acknowledgements 81
    - References 81
    - 5 Reference Classes: What They Didn't Talk About, But Somebody Should! 81
  - Chapter 6. Uncertainty and Incompleteness: Breaking the Symmetry of Defeasible Reasoning 83
    - 1 Introduction 83
    - 2 Plausible Reasoning Module 88
    - 3 Finding Admissible Labelings 90
    - 4 Algorithms and Heuristics 95
    - 5 Conclusions 97
    - References 98
  - Chapter 7. Deciding Consistency of Databases Containing Defeasible and Strict Information 103
    - 1 Introduction 103
    - 2 Notation and Preliminary Definitions 105
    - 3 Probabilistic Consistency and Entailment 106
    - 4 An Effective Procedure for Testing Consistency 109
    - 5 Examples 110
    - Acknowledgments 112
    - References 112
    - 6 Conclusions 112
  - CHAPTER 8. DEFEASIBLE DECISIONS: WHAT THE PROPOSAL IS AND ISN'T 115
    - 1 WHAT THE PROPOSAL IS 115
    - 2 WHAT THE PROPOSAL ISN'T 127
    - 3 AN OPEN CONVERSATION WITH RAIFFA 131
  - CHAPTER 9. CONDITIONING ON DISJUNCTIVE KNOWLEDGE: SIMPSON'S PARADOX IN DEFAULT LOGIC 133
    - 1. INTRODUCTION 133
    - 2. DOES AN EMU OR OSTRICH RUN? 135
    - 3. ARTS STUDENTS AND SCIENCE STUDENTS 136
    - 4. DISCUSSION OF THE PARADOX 137
    - 6. CONCLUSIONS 139
    - REFERENCES 140
    - ACKNOWLEDGEMENTS 140
  - PART Ill: ALGORITHMS FOR INFERENCE IN BELIEF NETS 143
    - Chapter 10. An Introduction to Algorithms for Inference in Belief Nets 145
      - 1. Introduction 145
      - 2. Qualitative, real, and interval-valued belief representations 145
      - 3. Early approaches 146
      - 4. Exact methods 147
      - 5. Two level belief networks 148
      - 6. Stochastic simulation and Monte Carlo schemes 149
      - 7. Final remarks 151
      - References 152
    - CHAPTER 11. d-SEPARATION: FROM THEOREMS TO ALGORITHMS 155
      - 1. INTRODUCTION 155
      - 2. SOUNDNESS AND COMPLETENESS OF d -SEPARATION 156
      - 3. THE MAIN RESULTS 158
      - ACKNOWLEDGEMENT 163
      - REFERENCES 164
    - Chapter 12. Interval Influence Diagrams 165
      - 1 Introduction 165
      - 2 Probabilistic Inference with Bounds on Probabilities 166
      - 3 Interval Influence Diagrams 167
      - 4 Transformations 168
      - 5 Example 172
      - 7 Conclusions 175
      - 6 Computational Characteristics 175
      - References 176
    - Chapter 13. A Tractable Inference Algorithm for Diagnosing Multiple Diseases 179
      - 1 Introduction 179
      - 2 The QMR model 180
      - 3 The Quickscore Algorithm 182
      - 4 Run-Time Performance of Quickscore 185
      - 5 Weaknesses of the Algorithm 185
      - 6 Conclusion 186
      - 7 Acknowledgments 186
      - References 186
    - CHAPTER 14. EVIDENCE ABSORPTION AND PROPAGATION THROUGH EVIDENCE REVERSALS 189
      - 1. Introduction 189
      - 2. Belief Diagrams 190
      - 3. Evidence Nodes and Evidence Propagation 191
      - 4. Probability Propagation 198
      - 5. Control of the Evidence Process 201
      - 6. Comparisons with the Pearl and the Lauritzen and Spiegelhalter Algorithms 203
      - 7. Conclusions 205
      - References 206
    - Chapter 15. An Empirical Evaluation of a Randomized Algorithm for Probabilistic Inference 207
      - 1. Introduction 207
      - 2. Methods and Procedures 209
      - 3. Results 215
      - 4. Discussion and Conclusions 219
      - 5. Acknowledgments 222
      - References 222
    - Chapter 16. Weighing and Integrating Evidence for Stochastic Simulation in Bayesian Networks 225
      - 1 Introduction 225
      - 2 The Evidence Weighting Technique 227
      - 3 Evidence Weighting With Evidential Integration 228
      - 4 Example 230
      - 5 Discussion 231
      - 6 Conclusions 234
      - References 235
    - CHAPTER 17. SIMULATION APPROACHES TO GENERAL PROBABILISTIC INFERENCE ON BELIEF NETWORKS 237
      - 1. Introduction 237
      - 2. The Algorithms 238
      - 3. Test Results 241
      - 4. Conclusions 243
      - 5. Acknowledgments 246
      - References 246
    - PART IV: SOFTWARE TOOLS FOR UNCERTAIN REASONING 249
      - Chapter 18. Software tools for uncertain reasoning: An Introduction 251
      - Chapter 19. Now that I Have a Good Theory of Uncertainty, What Else Do I Need? 253
        
        1 Normative vs. Prescriptive Theories of Uncertainty 253
        
        2 Dynamic Classification Problems: Situation Assessment 254
        
        3 RUM's Theory and Constraints 256
        
        4 The Integrated RUM/RUMrunner Technology 260
        
        5 Addressing the DCP's Reasoning Requirements 265
        
        6 RUM/RUMrunner Applications 267
        
        7 Conclusions 268
        
        References 268
      - Chapter 20. Knowledge Acquisition Techniques for Intelligent Decision Systems: Integrating Axotl and Aquinas in DDUCKS 271
        
        1. INTRODUCTION 271
        
        2. APPROACH 275
        
        3. DISCUSSION 282
        
        ACKNOWLEDGEMENTS 283
        
        REFERENCES 284
      - Chapter 21. BaRT: A Bayesian Reasoning Tool for Knowledge Based Systems 287
        
        1. Introduction 287
        
        2. Classificatory Problem Solving 289
        
        3. A Bayesian Reasoning Tool 292
        
        Acknowledgements 296
        
        References 296
        
        4. Conclusion 296
      - PART V: KNOWLEDGE ACQUISITION, MODELLING, AND EXPLANATION 299
        
        Chapter 22. Assessment, criticism and improvement of imprecise subjective probabilities for a medical expert system 301
        
        1. Introduction 301
        
        2. Background, Assessments and Data 302
        
        3. Criticising the probability assessments 303
        
        4. Discrimination and Reliability 305
        
        5. Learning from experience 307
        
        6. Discussion 308
        
        References 309
        
        Acknowledgments 309
        
        Chapter 23. Automated construction of sparse Bayesian networks from unstructured probabilistic models and domain information 311
        
        1 Introduction 312
        
        2 Bayesian networks 313
        
        3 The construction algorithm 315
        
        4 Results 318
        
        5 Discussion and further work 319
        
        6 Acknowledgements 322
        
        References 324
        
        Chapter 24. A Decision-Analytic Model for Using Scientific Data 325
        
        I THE PROPLEM 325
        
        2 A MODEL FOR THE USE OF REPORTED SCIENTIFIC DATA 327
        
        3 USING BIASES TO PARAMETERIZE THE SPACE OF CLINICAL STUDIES 330
        
        4 PREVIOUS ATTEMPTS TO MODEL THE USE OF SCIENTIFIC DATA 332
        
        5 USES OF THE MODEL 332
        
        REFERENCES 333
        
        ACKNOWLEDGMENTS 333
        
        Chapter 25. Verbal expressions for probability updates How much more probable is "much more probable"? 335
        
        1. Introduction 335
        
        2. Hypotheses about Phrase Selection Functions 336
        
        3. Experimental Design 338
        
        4. Analysis 339
        
        5. Conclusions 342
        
        Footnotes 343
        
        References 343
        
        Acknowledgements 343
        
        PART VI: APPLICATIONS TO VISION AND RECOGNITION 345
        
        Chapter 26. Map Learning with Indistinguishable Locations 347
        
        1 Introduction 347
        
        2 Spatial Modeling 348
        
        3 Map Learning 351
        
        4 Discussion 355
        
        5 Related Work 356
        
        References 357
        
        Chapter 27. Plan Recognition in Stories and in Life 359
        
        1 Introduction 359
        
        2 Preliminaries 360
        
        3 The "Knob" Theory 362
        
        4 The "Mention" Theory 365
        
        5 Conclusion 366
        
        Appendix: Work in Progress 367
        
        References 367
        
        Chapter 28. HIERARCHICAL EVIDENCE ACCUMULATION IN THE PSEIKI SYSTEM and EXPERIMENTS IN MODEL-DRIVEN MOBILE ROBOT NAVIGATION 369
        
        1. APPLICATION 369
        
        2. REPRESENTATION AND FLOW OF CONTROL 371
        
        3. ACCUMULATION OF EVIDENCE 377
        
        4. EDGE-BASED vs. REGION-BASED OPERATION 381
        
        6. ROBOT SELF-LOCATION USING PSEIKI 382
        
        5. ARE INDEPENDENCE CONDITIONS SATISFIED? 382
        
        7. CONCLUDING REMARKS 384
        
        8. REFERENCES 385
        
        Chapter 29. Model-Based Influence Diagrams For Machine Vision 387
        
        1 Introduction 387
        
        2 Model-Based Reasoning for Machine Vision 389
        
        3 Sequential Control for Machine Vision Inference 390
        
        4 Model Guided Influence Diagram Construction 391
        
        5 Dynamic Instantiation for Sequential Control 394
        
        6 Conclusions 400
        
        Acknowledgments 403
        
        References 403
        
        CHAPTER 30. THE APPLICATION OF DEMPSTER SHAFER THEORY TO A LOGIC-BASED VISUAL RECOGNITION SYSTEM 405
        
        1 INTRODUCTION 405
        
        2 DEMPSTER SHAFER THEORY REVIEW 406
        
        4 DEMPSTER SHAFER THEORY FORMULATION IN LOGIC-BASED TERMS 407
        
        3 PROPOSITIONAL LOGIC REVIEW 407
        
        5 ATMS-BASED IMPLEMENTATION OF DEMPSTER SHAFER THEORY 410
        
        6 MODEL-BASED VISUAL RECOGNITION USING AN EXTENDED ATMS 413
        
        References 420
        
        7 DISCUSSION 420
        
        ACKNOWLEDGEMENTS 420
        
        Chapter 31. Efficient Parallel Estimation for Markov Random Fields 423
        
        1 Introduction 423
        
        2 Generating Most Probable Labelings 424
        
        3 Markov Random Fields 425
        
        4 HCF 426
        
        5 Local HCF 427
        
        6 Test Results 427
        
        7 Conclusions and Future Work 431
        
        A Proof of Convergence for Local HCF 433
        
        References 434
        
        B Comparing Local HCF and HCF 434
        
        PART VII: COMPARING APPROACHES TO UNCERTAIN REASONING 437
        
        Chapter 32. Comparing Approaches to Uncertain Reasoning: Discussion System Condemnation Pays Off 439
        
        CHAPTER 33. A PROBABILITY ANALYSIS OF THE USEFULNESS OF DECISION AIDS1 443
        
        1.0 INTRODUCTION 443
        
        2.0 FALLIBLE VS. INFALLIBLE ADVICE 445
        
        3.0 OVERCOMING THE COST OF FALLIBILITY 448
        
        4.0 DISCUSSION 451
        
        REFERENCES 452
        
        CHAPTER 34. INFERENCE POLICIES1, 2 453
        
        1.0 SATISFYING REQUIREMENTS 453
        
        2.0 STANDARD INFERENCE POLICIES 454
        
        3.0 NONSTANDARD INFERENCE POLICIES 456
        
        4.0 SUMMARY AND DISCUSSION 460
        
        REFERENCES 460
        
        CHAPTER 35. COMPARING EXPERT SYSTEMS BUILT USING DIFFERENT UNCERTAIN INFERENCE SYSTEMS 461
        
        1. INTRODUCTION 461
        
        2. METHOD 463
        
        3. RESULTS AND DISCUSSION 466
        
        4. CONCLUSIONS 469
        
        5. REFERENCES 470
        
        Chapter 36. Shootout-89, An Evaluation of Knowledge-based Weather Forecasting Systems 473
        
        Author index 475

Description

TOC

Reviews (0)