Correcting Imperfect Domain Theories: A Knowledge-Level Analysis

Explanation-Based Learning (Mitchell et al., 1986; DeJong and Mooney, 1986) has shown promise as a powerful analytical learning technique. However, EBL is severely hampered by the requirement of a complete and correct domain theory for successful learning to occur. Clearly, in non-trivial domains, developing such a domain theory is a nearly impossible task. Therefore, much research has been devoted to understanding how an imperfect domain theory can be corrected and extended during system performance. In this paper, we present a characterization of this problem, and use it to analyze past research in the area. Past characterizations of the problem (e.g, (Mitchell et al., 1986; Rajamoney and DeJong, 1987)) have viewed the types of performance errors caused by a faulty domain theory as primary. In contrast, we focus primarily on the types of knowledge deficiencies present in the theory, and from these derive the types of performance errors that can result. Correcting the theory can be viewed as a search through the space of possible domain theories, with a variety of knowledge sources that can be used to guide the search.

[1]  Jack Mostow,et al.  Adaptive Search by Explanation-Based Learning of Heuristic Censors , 1990, AAAI.

[2]  Tom M. Mitchell,et al.  Becoming Increasingly Reactive , 1990, AAAI.

[3]  Richard J. Doyle,et al.  Constructing and Refining Causal Explanations from an Inconsistent Domain Theory , 1986, AAAI.

[4]  John E. Laird,et al.  Integrating, Execution, Planning, and Learning in Soar for External Environments , 1990, AAAI.

[5]  Yolanda Gil,et al.  Acquiring domain knowledge for planning by experimentation , 1992 .

[6]  Andrea Pohoreckyj Danyluk,et al.  The Use of Explanations for Similarity-based Learning , 1987, IJCAI.

[7]  Jude Shavlik,et al.  Refinement ofApproximate Domain Theories by Knowledge-Based Neural Networks , 1990, AAAI.

[8]  Oren Etzioni,et al.  Explanation-Based Learning: A Problem Solving Perspective , 1989, Artif. Intell..

[9]  Kamal M. Ali,et al.  Augmenting Domain Theory for Explanation-Based Generalization , 1989, ML.

[10]  Michael Redmond.  Combining Case-Based Reasoning, Explanation-Based Learning, and Learning form Instruction , 1989, ML.

[11]  Michael R. Lowry,et al.  Learning Physical Descriptions From Functional Definitions, Examples, and Precedents , 1983, AAAI.

[12]  R. James Firby,et al.  An Investigation into Reactive Planning in Complex Domains , 1987, AAAI.

[13]  Raymond J. Mooney,et al.  Learning Plan Schemata From Observation: Explanation-Based Learning for Plan Recognition , 1990, Cogn. Sci..

[14]  Robert Dale,et al.  Proceedings of the Thirteenth Annual Conference of the Cognitive Science Society , 1991 .

[15]  Jack Mostow,et al.  Parsing to Learn Fine Gralned Rules , 1988, AAAI.

[16]  David Jechiel Mostow,et al.  Mechanical transformation of task heuristics into operational procedures , 1981 .

[17]  Jaime G. Carbonell,et al.  Learning by Experimentation , 1987 .

[18]  Michael G. Dyer,et al.  Using Prior Learning to Facilitate the Learning of New Causal Theories , 1987, IJCAI.

[19]  Earl D. Sacerdoti,et al.  Planning in a Hierarchy of Abstraction Spaces , 1974, IJCAI.

[20]  Richard S. Sutton,et al.  Integrated Architectures for Learning, Planning, and Reacting Based on Approximating Dynamic Programming , 1990, ML.

[21]  David C. Wilkins,et al.  Knowledge Base Refinement Using Apprenticeship Learning Techniques , 1988, AAAI.

[22]  Craig A. Knoblock.  Learning Abstraction Hierarchies for Problem Solving , 1990, AAAI.

[23]  Stan Matwin,et al.  Explanation-Based Learning with Incomplete Theories: A Three-step Approach , 1990, ML.

[24]  Sridhar Mahadevan.  Using Determinations in EBL: A Solution to the incomplete Theory Problem , 1989, ML.

[25]  Prasad Tadepalli,et al.  Lazy ExplanationBased Learning: A Solution to the Intractable Theory Problem , 1989, IJCAI.

[26]  David J. Mostow,et al.  Machine Transformation of Advice Into a Heuristic Search Procedure , 1983 .

[27]  Paul S. Rosenbloom,et al.  Abstraction in Problem Solving and Learning , 1989, IJCAI.

[28]  Scott W. Bennett.  Reducing Real-world Failures of Approximate Explanation-based Rules , 1990, ML.

[29]  Patrick Henry Winston,et al.  Learning physical descriptions from functional definitions , 1983, AAAI 1983.

[30]  Tom M. Mitchell,et al.  Generalization as Search , 2002 .

[31]  Michael Hucka,et al.  Correcting and Extending Domain Knowledge using Outside Guidance , 1990, ML.

[32]  Gerald DeJong,et al.  The Classification, Detection and Handling of Imperfect Theory Problems , 1987, IJCAI.

[33]  Ajay Gupta,et al.  Explanation-based Failure Recovery , 1987, AAAI.

[34]  Allen Newell,et al.  A Universal Weak Method: Summary of Results , 1983, IJCAI.

[35]  Andrea Pohoreckyj Danyluk,et al.  Finding New Rules for Incomplete Theories: Explicit Biases for Induction with Contextual Information , 1989, ML.

[36]  Stuart J. Russell,et al.  A Logical Approach to Reasoning by Analogy , 1987, IJCAI.

[37]  Yolanda Gil,et al.  A Domain-Independent Framework for Effective Experimentation in Planning , 1991, ML.

[38]  Gheorghe Tecuci,et al.  DISCIPLE-1: Interactive Apprentice System in Weak Theory Fields , 1987, IJCAI.

[39]  Scott Bennett,et al.  Approximation in Mathematical Domains , 1987, IJCAI.

[40]  G. DeJong,et al.  An Alternative To Deduction , 1991 .

[41]  Allen Newell,et al.  SOAR: An Architecture for General Intelligence , 1987, Artif. Intell..

[42]  Michael Hucka,et al.  Learning in Tele-autonomous Systems using Soar , 1989 .

[43]  Michael J. Pazzani,et al.  Integrated Learning with Incorrect and Incomplete Theories , 1988, ML.

[44]  Craig A. Knoblock.  Search Reduction in Hierarchical Problem Solving , 1991, AAAI.

[45]  Gerald DeJong,et al.  Active Explanation Reduction: An Approach to the Multiple Explanations Problem , 1988, ML.

[46]  Raymond J. Mooney,et al.  Changing the Rules: A Comprehensive Approach to Theory Refinement , 1990, AAAI.

[47]  Michael J. Pazzani.  Detecting and Correcting Errors of Omission After Explanation-Based Learning , 1989, IJCAI.

[48]  Robert J. Hall.  Learning by Failing to Explain , 1986, AAAI.

[49]  Francesco Bergadano,et al.  A Knowledge Intensive Approach to Concept Induction , 1988, ML Workshop.

[50]  Kristian J. Hammond.  Learning to Anticipate and Avoid Planning Problems through the Explanation of Failures , 1986, AAAI.

[51]  Jack Mostow,et al.  Failsafe - A Floor Planner that Uses EBG to Learn from Its Failures , 1987, IJCAI.

[52]  Kurt VanLehn,et al.  Learning one Subprocedure per Lesson , 1987, Artif. Intell..

[53]  Gerhard Widmer,et al.  A tight integration of deductive and inductive learning , 1989, ICML 1989.

[54]  Steve A. Chien.  Using and Refining Simplifications: Explanation-Based Learning of Plans in Intractable Domains , 1989, IJCAI.

[55]  John H. Holland,et al.  Escaping brittleness: the possibilities of general-purpose learning algorithms applied to parallel rule-based systems , 1995 .

[56]  David E. Smith,et al.  Reasoning About Action I: A Possible Worlds Approach , 1987, Artif. Intell..

[57]  Nicholas S. Flann.  Applying Abstraction and Simplification to Learn in Intractable Domains , 1990, ML.