Kanal / Lemmer Uncertainty in Artificial Intelligence

1. Auflage 2014
ISBN: 978-1-4832-9652-4
Verlag: Elsevier Science & Techn.
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, Band Volume 4, 522 Seiten, Web PDF

Reihe: Machine Intelligence and Pattern Recognition

How to deal with uncertainty is a subject of much controversy in Artificial Intelligence. This volume brings together a wide range of perspectives on uncertainty, many of the contributors being the principal proponents in the controversy.Some of the notable issues which emerge from these papers revolve around an interval-based calculus of uncertainty, the Dempster-Shafer Theory, and probability as the best numeric model for uncertainty. There remain strong dissenting opinions not only about probability but even about the utility of any numeric method in this context.

Autoren/Hrsg.

Kanal, L. N.

Lemmer, J. F.

Inhaltsverzeichnis

1;Front Cover;1
2;Uncertainty in Artificial Intelligence;4
3;Copyright Page;5
4;PREFACE;6
5;CONTRIBUTORS;8
6;Table of Contents;10
7;PART
I: OVERVIEWS AND REVIEWS;14
7.1;Chapter 1.
Handling Uncertain Information : A Review of Numeric and Non-numeric Methods;16
7.1.1;1. Introduction;16
7.1.2;2. Representatiori Of Uncertain
Information;17
7.1.3;3. Probability;18
7.1.4;4. Evidence Theory;20
7.1.5;5. Possibility Theory;24
7.1.6;6. Non-numeric methods;25
7.1.7;7. Theory of Endorsements;27
7.1.8;8.
Combination Of Bodies Of Uncertain Information;29
7.1.9;9.
Drawing Inferences From Uncertain Information;32
7.1.10;10. Conclusions;37
7.1.11;References;38
7.2;CHAPTER
2. CONSENSUS RULES;40
7.2.1;1. INTRODUCTION;40
7.2.2;2. SOME CONSENSUS RULE CHARACTERISTICS;41
7.2.3;3. CHOICE OF WEIGHTS FOR LINEAR POOLS - UPDATING;43
7.2.4;4. CONCLUDING REMARKS;44
7.2.5;REFERENCES;45
8;PART
II: EXPLICATION OR CRITIQUE OF CURRENT APPROACHES TO UNCERTAINTY;46
8.1;Chapter 3.
Uncertainty Handling in Expert Systems: Uniform vs. Task-Specific Formalisms;48
8.1.1;1. Is There An "Uncertainty Handling" Problem?;48
8.1.2;2. Conflation of Different Aims;49
8.1.3;3. Uncertainty Handling in Classification Problem Solving;52
8.1.4;4. Concluding Remarks;57
8.1.5;References;58
8.2;CHAPTER
4. PROBABILISTIC REASONING IN PREDICTIVE EXPERT SYSTEMS;60
8.2.1;1. INTRODUCTION;60
8.2.2;2. AN EXAMPLE;61
8.2.3;3. 'EXCHANGEABILITY' AND DOUBT ABOUT PROBABILITIES;75
8.2.4;4. DISCUSSION: WHEN IS PROBABILITY APPROPRIATE ?;76
8.2.5;REFERENCES;78
8.3;CHAPTER
5. A FRAMEWORK FOR COMPARING UNCERTAIN INFERENCE SYSTEMS TO PROBABILITY;82
8.3.1;1. INTRODUCTION;82
8.3.2;2. FORMALISMS FOR REPRESENTING UNCERTAINTY;83
8.3.3;3. COMPARING UIS'S;85
8.3.4;4. EVALUATING DIFFERENCES IN RESULTS;91
8.3.5;5. COMPARING PERFORMANCE ON AN EXAMPLE RULE-SET;92
8.3.6;6. FINAL REMARKS;94
8.3.7;References;95
8.4;Chapter 6.
Probabilistic versus Fuzzy Reasoning;98
8.4.1;1 Introduction;98
8.4.2;2 The Probabilistic Approach;99
8.4.3;3 Combining Evidence;102
8.4.4;4 "Conflicting Evidence";106
8.4.5;5 Uncertain Evidence;107
8.4.6;6 The Fuzzy Approach;110
8.4.7;7 Conclusions;113
8.4.8;References;114
8.5;Chapter 7.
Is Probability Theory Sufficient for Dealing with Uncertainty in AI: A Negative View;116
8.5.1;1. The Issue of Adequacy;116
8.5.2;2. Inference;119
8.5.3;REFERENCES AND RELATED PUBLICATIONS;126
8.6;CHAPTER
8. CONFIDENCE FACTORS, EMPIRICISM AND THE DEMPSTER-SHAFER THEORY OF EVIDENCE;130
8.6.1;REFERENCES;138
8.7;CHAPTER
9. PROBABILITY JUDGMENT IN ARTIFICIAL INTELLIGENCE;140
8.7.1;1. Two Probability Languages;140
8.7.2;2. Three Examples;142
8.7.3;3. Probability Judgment in Expert Systems;147
8.7.4;References;148
8.8;CHAPTER 10. THE INCONSISTENT USE OF MEASURES OF CERTAINTY IN ARTIFICIAL INTELLIGENCE
RESEARCH;150
8.8.1;1. INTRODUCTION;150
8.8.2;2. DISTINGUISHING BELIEF UPDATES FROM ABSOLUTE BELIEFS;151
8.8.3;3. HISTORICAL BLURRING OF BELIEF AND BELIEF UPDATE;151
8.8.4;4. INTUITIVE PROPERTIES OF MEASURES OF BELIEF;152
8.8.5;5. PROPERTIES OF BELIEF UPDATES;154
8.8.6;6. A PROBABILISTIC BELIEF UPDATE;155
8.8.7;7. INCONSISTENCY OF EQUATING ABSOLUTE BELIEFS WITH BELIEF UPDATES;155
8.8.8;8. EVIDENCE COMBINATION AND MODULARITY;156
8.8.9;9. THE MODULAR UPDATING PARADIGM;158
8.8.10;10. INCONSISTENT USE OF THE MODULAR UPDATING PARADIGM;158
8.8.11;11. MODULAR BELIEF UPDATING IN MYCIN;159
8.8.12;12. MODULAR BELIEF UPDATING IN INTERNIST-1;160
8.8.13;13. CONSEQUENCES OF THE INCONSISTENCY;161
8.8.14;14. RELEVANCE OF BIASES IN THE ELICITATION OF BELIEF;162
8.8.15;15. SUMMARY;162
8.8.16;REFERENCES;163
8.9;CHAPTER
11. EVIDENTIAL CONFIRMATION AS TRANSFORMED PROBABILITY: On The Duality of Priors and Updates;166
8.9.1;1. INTRODUCTION1: EVIDENTIAL CONFIRMATION VERSUS PROBABILITY;166
8.9.2;2. OVERVIEW;167
8.9.3;3. PROSPECTOR: BAYESIAN UPDATING WITH CONDITIONAL INDEPENDENCE;168
8.9.4;4. CFS ARE A TRANSFORM OF LIKELIHOOD RATIOS, WITH INDEPENDENCE ASSUMPTION;170
8.9.5;5. THE POINT-VALUED SPECIAL CASE OF DEMPSTER-SHAFER THEORY;171
8.9.6;6. COMBINING CFS IS MULTIPLYING LIKELIHOOD RATIOS IS DEMPSTER'S RULE;172
8.9.7;7. VIEWING EVIDENCE AS UPDATES VERSUS AS PRIORS;172
8.9.8;8. DISCUSSION;174
8.9.9;9. CONCLUSIONS;176
8.9.10;REFERENCES;177
8.10;CHAPTER
12. PROBABILISTIC INTERPRETATIONS FOR MYCINS CERTAINTY FACTORS;180
8.10.1;1. INTRODUCTION;180
8.10.2;2. MYCIN'S CERTAINTY FACTORS;181
8.10.3;3. OVERVIEW OF APPROACH;183
8.10.4;4. THE DESIDERATA OF CERTAINTY FACTORS;185
8.10.5;5. A CLOSER LOOK AT THE ORIGINAL DEFINITION;187
8.10.6;6. REQUIREMENT FOR A PROBABILISTIC INTERPRETATION;189
8.10.7;7. A PROBABILISTIC INTERPRETATION;189
8.10.8;8. OTHER PROBABILISTIC INTERPRETATIONS;192
8.10.9;9. THE ASSUMPTION OF CONDITIONAL INDEPENDENCE;193
8.10.10;10. SEQUENTIAL COMBINATION;196
8.10.11;11. NON-TREE NETWORKS;200
8.10.12;12. DISCUSSION;202
8.10.13;13. SUMMARY;203
8.10.14;APPENDIX;203
8.10.15;REFERENCES;208
8.11;CHAPTER
13. INDEPENDENCE AND BAYESIAN UPDATING METHODS;210
8.11.1;1.
BACKGROUND;210
8.11.2;2. COUNTEREXAMPLES;211
8.11.3;3. IMPOSSIBILITY OF MULTIPLE UPDATING;212
8.11.4;4. DISCUSSION;213
8.11.5;REFERENCES;214
8.12;Chapter 14.
Uncertain Reasoning using Maximum Entropy Inference;216
8.12.1;1. Introduction;216
8.12.2;2. The Information-Theoretic Justification;216
8.12.3;3. The Axiomatic Justification;218
8.12.4;4. Other Methods of Reasoning with Uncertainty;220
8.12.5;5. Conclusion;222
8.12.6;References;222
8.13;CHAPTER
15. RELATIVE ENTROPY, PROBABILISTIC INFERENCE, AND AI;224
8.13.1;I. INTRODUCTION;224
8.13.2;II. INFORMATION, ENTROPY, AND RELATIVE-ENTROPY;224
8.13.3;III. THE PRINCIPLE OF MINIMUM RELATIVE ENTROPY;226
8.13.4;IV. RELATIVE ENTROPY AND AI;227
8.13.5;REFERENCES;227
8.14;CHAPTER
16. SELECTING UNCERTAINTY CALCULI AND GRANULARITY: AN EXPERIMENT IN TRADING-OFF PRECISION AND COMPLEXITY;230
8.14.1;1. INTRODUCTION;230
8.14.2;2. AGGREGATION OPERATORS;232
8.14.3;3. LINGUISTIC VARIABLES DEFINED ON THE INTERVAL
[0,1];237
8.14.4;4. DESCRIPTION OF THE EXPERIMENTS AND REQUIRED TECHNIQUES;239
8.14.5;5. EXPERIMENT RESULTS AND ANALYSIS;245
8.14.6;6. CONCLUSIONS;253
8.14.7;FOOTNOTES;255
8.14.8;REFERENCES;257
8.14.9;APPENDIX: PROPERTIES OF T-NORM OPERATORS;259
8.15;Chapter 17.
What Uncertainty Judgments Can Tell About the Underlying Subjective Probabilities;262
8.15.1;References;270
8.16;CHAPTER
18. AN INEQUALITY PARADIGM FOR PROBABILISTIC KNOWLEDGE;272
8.16.1;1. OVERVIEW;272
8.16.2;2. UNCONDITIONAL TYPE-1 PROBABILISTIC THEORIES;275
8.16.3;3. CONDITIONAL TYPE-1 PROBABILISTIC THEORIES;276
8.16.4;4. ADDITIONAL NON-TYPE-1 ASSUMPTIONS AS CONSTRAINTS;278
8.16.5;5. TYPE-2 PROBABILISTIC LOGIC;279
8.16.6;6. EVIDENCE AND CONFIRMATION;280
8.16.7;7. APPLICATION OF THE PARADIGM: ENTAILMENT;281
8.16.8;8. APPLICATION OF THE PARADIGM: INFERENCE;284
8.16.9;9. DISCUSSION AND CONCLUSIONS;285
8.16.10;REFERENCES;288
9;PART
Ill: SYNTHESIS OF CURRENT APPROACHES TO UNCERTAINTY;290
9.1;CHAPTER
19. AN EXPERT SYSTEM FRAMEWORK FOR NON-MONOTONIC REASONING ABOUT PROBABILISTIC ASSUMPTIONS;292
9.1.1;The Problem;292
9.1.2;A New Approach: Conceptual Outline;293
9.1.3;Theory of Belief Functions;294
9.1.4;Conflict of Evidence;296
9.1.5;Alternative Approaches: Interdependence Versus Modularity;298
9.1.6;Basic
Features of NMP;299
9.1.7;Culprits and Denials;301
9.1.8;Conditions for Belief Revision;302
9.1.9;Conflict as Control Over Revision;302
9.1.10;When specific revisions are not justified;303
9.1.11;A role for qualitative reasoning;304
9.1.12;Conclusion;304
9.1.13;Footnote;305
9.1.14;References;305
9.2;Chapter 20.
Metaprobability and Dempster-Shafer In Evidential Reasoning;308
9.2.1;1. INTRODUCTION;308
9.2.2;2. METAPROBABILITY THEORY;308
9.2.3;3. DEMPSTER-SHAFER THEORY;310
9.2.4;4. EXPERIMENT;311
9.2.5;5. ANALYSIS OF EXPERIMENTAL RESULTS;314
9.2.6;6. CONCLUSIONS;315
9.2.7;REFERENCES;315
9.3;Chapter 21.
Knowledge Structures and Evidential Reasoning in Decision Analysis;316
9.3.1;1. Introduction;316
9.3.2;2. Factors and Evidence;317
9.3.3;3. A View from Dempster-Shafer Theory;326
9.3.4;4. Concluding Remark;327
9.3.5;Reference;328
9.4;Chapter 22.
A General Approach to Decision Making with Evidential Knowledge;330
9.4.1;1. Introduction;330
9.4.2;2. A Decision Making Paradigm;330
9.4.3;3. Representation of State Knowledge;332
9.4.4;4. Decision Making with D-S Granules;334
9.4.5;5. Conclusion;339
9.4.6;References;340
10;PART
IV: INCORPORATING UNCERTAINTY IN SYSTEMS;342
10.1;Chapter 23.
Implementing Probabilistic Reasoning;344
10.1.1;1. Introduction;344
10.1.2;2. An overview of MRS;344
10.1.3;3. Probabilistic databases;345
10.1.4;4. Forward chaining;347
10.1.5;5. Backward chaining;349
10.1.6;6. Resolution;350
10.1.7;7. Conclusion;350
10.1.8;References;351
10.2;CHAPTER
24. EXACT REASONING ABOUT UNCERTAINTY: ON THE DESIGN OF EXPERT SYSTEMS FOR DECISION SUPPORT;352
10.2.1;1. INTRODUCTION: MAKING DIFFICULT DECISIONS;352
10.2.2;2. EXPERT SYSTEMS AS DECISION AIDS;353
10.2.3;3. USING EXPERT SYSTEMS FOR DEVELOPING REPRESENTATIONS OF SPECIFIC DECISIONS;354
10.2.4;4. REPRESENTING DECISIONS AS INFLUENCE DIAGRAMS;356
10.2.5;5. RACHEL: A PILOT-LEVEL EXPERT DECISION ANALYST;356
10.2.6;6. CONCLUSION;357
10.2.7;REFERENCES;357
10.3;CHAPTER 25. MODEL-BASED PROBABILISTIC SITUATION IN FERENCE IN HIERARCHICAL HYPOTHESIS
SPACES;360
10.3.1;1. INTRODUCTION;360
10.3.2;2. NUMERICAL SUPPORT FOR HYPOTHESIS HIERARCHIES;363
10.3.3;3. PROBABILISTIC CONFLICT RESOLUTION;366
10.3.4;4. SUMMARY;368
10.3.5;REFERENCES;368
10.3.6;APPENDIX A;368
10.4;CHAPTER
26. A CONSTRAINT - PROPAGATION APPROACH TO PROBABILISTIC REASONING;370
10.4.1;1. INTRODUCTION: BAYES NETWORKS AND CONSTRAINTS PROPAGATION;370
10.4.2;2. PROPAGATION IN SINGLY-CONNECTED NETWORKS;373
10.4.3;3. PROPAGATION IN MULTIPLY-CONNECTED NETWORKS;377
10.4.4;CONCLUSIONS;380
10.4.5;REFERENCES;381
10.5;CHAPTER
27. INTELLIGENT PROBABILISTIC INFERENCE;384
10.5.1;I. INTRODUCTION;384
10.5.2;2. BASIC FRAMEWORK;385
10.5.3;3. PROBABILISTIC INFERENCE;387
10.5.4;4. TRANSFORMATIONS;388
10.5.5;5. SOLUTION PROCEDURE;390
10.5.6;6. INFORMATION REQUIRED;391
10.5.7;7. CONCLUSIONS;394
10.5.8;REFERENCES;395
10.6;CHAPTER
28. AN ODDS RATIO BASED INFERENCE ENGINE;396
10.6.1;1. INTRODUCTION;396
10.6.2;2. UNCERTAIN INFERENCE METHOD;397
10.6.3;3. COMPUTING METHODS;400
10.6.4;4. SUMMARY;401
10.6.5;REFERENCES;401
11;PART
V: TECHNIQUES FOR INDUCING AND PROCESSING UNCERTAIN INFORMATION;404
11.1;CHAPTER
29. INDUCTIVE INFERENCE AND THE REPRESENTATION OF UNCERTAINTY;406
11.1.1;Introduction;406
11.1.2;Knowledge Sets;406
11.1.3;Inductive Inference;407
11.1.4;Updating;408
11.1.5;Min-score Inference with Information Updating;409
11.1.6;References;410
11.2;CHAPTER
30. REPRESENTING, COMBINING AND USING UNCERTAIN ESTIMATES;412
11.2.1;1. INTRODUCTION: OVERVIEW AND MOTIVATION;412
11.2.2;2.
DESIRABLE PROPERTIES OF A SOLUTION;416
11.2.3;3.
SOME CANDIDATE SOLUTIONS;421
11.2.4;4. A PROPOSAL: VIRTUAL SAMPLING;423
11.2.5;REFERENCES;427
11.3;Chapter 31.
Machine Learning, Clustering and Polymorphy;428
11.3.1;Introduction;428
11.3.2;Rationale;429
11.3.3;WITT Structure;431
11.3.4;Implications for Applications in AI and Information
Retrieval;432
11.3.5;Some Results: WITT Studies;434
11.3.6;REFERENCES;440
11.4;CHAPTER
32. INDUCTION, OF AND BY PROBABILITY;442
11.4.1;1. INTRODUCTION;442
11.4.2;2. REPRESENTATIONS FOR INDUCTIVE LEARNING;444
11.4.3;3. PROBABILITY AS PRODUCT OF LEARNING;447
11.4.4;4. PROBABILITY AS INDUCTIVE CRITERION;453
11.4.5;5. SUMMARY;454
11.4.6;REFERENCES;455
12;PART
VI: ALTERNATE PERSPECTIVES;458
12.1;CHAPTER
33. THREE ARGUMENTS FOR EXTENDING THE FRAMEWORK OF PROBABILITY;460
12.1.1;1. INTRODUCTION;460
12.1.2;2. UNCERTAINTY AS A TYPE OF KNOWLEDGE;461
12.1.3;3. SOME LOGICAL TYPES OF UNCERTAINTY;462
12.1.4;4. WEAK METHODS FOR MANAGING UNCERTAINTY;464
12.1.5;5. WEAK METHODS AND THE COMBINATION OF INFORMATION;466
12.1.6;6. META-LEVEL REASONING ABOUT UNCERTAINTY METHODS;467
12.1.7;REFERENCES;471
12.1.8;DOCUMENTATION NOTES;471
12.2;Chapter 34.
Interval-Based Decisions for Reasoning Systems;472
12.2.1;1. Interval Measures;472
12.2.2;2. Estimation and Decision;473
12.2.3;3. Secondary Criterion Solutions;475
12.2.4;5. Examples and Contrasts;478
12.2.5;6. Epistemological Considerations;481
12.2.6;7. Conclusion;483
12.2.7;Notes and References;483
12.3;CHAPTER
35. THE APPLICATION OF ALGORITHMIC PROBABILITY TO PROBLEMS IN ARTIFICIAL INTELLIGENCE;486
12.3.1;1. INTRODUCTION;486
12.3.2;2. ALGORITHMIC COMPLEXITY;489
12.3.3;3. A GENERAL SYSTEM FOR SOLVING PROBLEMS;491
12.3.4;4. USING PROBABILITY DISTRIBUTIONS TO REPRESENT KNOWLEDGE;498
12.3.5;5. RELATION OF ALGORITHMIC PROBABILITY TO OTHER METHODS OF DEALING WITH UNCERTAINTY;502
12.3.6;6. PRESENT STATE OF DEVELOPMENT OF THE SYSTEM;502
12.3.7;References;503
13;PART
VII: ALTERNATIVES TO MINIMAX IN GAME PLAYING;506
13.1;CHAPTER
36. AN EXPLANATION OF AND CURE FOR MINIMAX PATHOLOGY;508
13.1.1;1. Introduction;508
13.1.2;2. A Pathological Game: Board Splitting;509
13.1.3;3. A Nonpathological Evaluation Function for Board Splitting;511
13.1.4;4. The Effect of F-wins;512
13.1.5;5. Curing Pathology by Recognizing F-wins;514
13.1.6;6. Conclusions and Work in Progress;516
13.1.7;References;517
13.2;CHAPTER 37. AN EVALUATION OF TWO ALTERNATIVES TO MINIMAX;518
13.2.1;1. Introduction;518
13.2.2;2. Results and Data Analysis;519
13.2.3;3. Conclusion;521
13.2.4;REFERENCES;521