An Introduction to Explanation-based Learning
Elsevier eBooksPublished 1 January 1988
Gerald DeJong
Citations56
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
This chapter presents an intuitive appreciation for EBL, discusses the various types of EBL generalization and presents several formalisms that have been advanced to handle some small fraction of them.
Abstract
This chapter presents an introduction to explanation-based learning (EBL). EBL is best viewed as a kind of learning from observation. It allows a system to acquire general knowledge through an analysis of a few specific episodes. Background knowledge plays a crucial role in the analysis process. In large part, the background knowledge substitutes for the massive training sets needed in traditional machine learning. It is convenient, though not necessary, to view EBL in the context of problem solving or, more precisely, learning about problem solving. The chapter presents an intuitive appreciation for EBL. It presents the comparison between EBL and similarity-based learning. The chapter discusses the various types of EBL generalization and presents several formalisms that have been advanced to handle some small fraction of them. It presents a historical account of EBL development and a discussion of a few of the important outstanding research issues.
Keywords
Computer Science
Learning Internal Representations by Error Propagation
16,235 Citations1985David E. Rumelhart, Geoffrey E. Hinton +1 more
Machine LearningInduction of Decision Trees
14,815 Citations1986J. R. Quinlan
This paper summarizes an approach to synthesizing decision trees that has been used in a variety of systems, and it describes one such system, ID3, in detail, which is described in detail.
Machine LearningInduction of decision trees
12,383 Citations1986J. R. Quinlan
This paper summarizes an approach to synthesizing decision trees that has been used in a variety of systems, and it describes one such system, ID3, in detail, which is described in detail.
Communications of the ACMMaintaining knowledge about temporal intervals
7,550 Citations1983James F. Allen
An interval-based temporal logic is introduced, together with a computationally effective reasoning algorithm based on constraint propagation, which is notable in offering a delicate balance between space and time.
Cognitive ScienceStructure‐Mapping: A Theoretical Framework for Analogy*
5,069 Citations1983Dedre Gentner
In the structure-mapplng theory, the interpretation rules ore characterized OS implicit rules for mapping knowledge about a base domain into a torget domain, defined by the existence of higher-order relations.
Elsevier eBooksA FRAMEWORK FOR REPRESENTING KNOWLEDGE
4,545 Citations1988Marvin Minsky
The enormous problem of the volume of background common sense knowledge required to understand even very simple natural language texts is discussed and it is suggested that networks of frames are a reasonable approach to represent such knowledge.
Artificial IntelligenceStrips: A new approach to the application of theorem proving to problem solving
4,073 Citations1971Richard Fikes, Nils J. Nilsson
A new problem solver called STRIPS is described that attempts to find a sequence of operators in a space of world models to transform a given initial world model into a model in which a given goal formula can be proven to be true.
Language<b>Scripts, plans, goals, and understanding</b> : an inquiry into human knowledge structures. By Roger C. Schank and Robert P. Abelson. Hillsdale, NJ: Lawrence Erlbaum Associates, 1977. Pp. 248.
3,083 Citations1978David B. Kronenfeld
For both people and machines, each in their own way, there is a serious problem in common of making sense out of what they hear, see, or are told about the world.
Psychological ReviewThe role of theories in conceptual coherence.
2,669 Citations1985Gregory L. Murphy, Douglas L. Medin
It is proposed that concepts are coherent to the extent that they fit people's background knowledge or naive theories about the world and to structure the attributes that are internal to a concept.
Artificial IntelligenceThe knowledge level
2,484 Citations1982Allen Newell
A theory of the nature of knowledge is proposed, namely, that there is another computer system level immediately above the symbol (or program) level and knowledge itself is the processing medium at this level and the principle of rationality plays a central role.
Cambridge University Press eBooksSimilarity and Analogical Reasoning
2,397 Citations1989Stella Vosniadou, Stella Vosniadou +22 more
Artificial IntelligenceQualitative process theory
2,023 Citations1984Kenneth D. Forbus
This paper describes the basic concepts of qualitative process theory, several different kinds of reasoning that can be performed with them, and discusses its implications for causal reasoning.
Machine LearningExplanation-Based Generalization: A Unifying View
1,175 Citations1986Tom M. Mitchell, Richard M. Keller +1 more
This paper proposes a general, domain-independent mechanism, called EBG, that unifies previous approaches to explanation-based generalization, and is illustrated in the context of several example problems, and used to contrast several existing systems for explanation- based generalization.
Artificial IntelligencePlanning for conjunctive goals
1,110 Citations1987David Chapman
Theorems that suggest that efficient general purpose planning with more expressive action representations is impossible are presented, and ways to avoid this problem are suggested.
Graphical ModelsLearning and Relearning in Boltzmann Machines
1,052 Citations2001Geoffrey E. Hinton, T. J. Sejnowski
This chapter contains sections titled: Relaxation Searches, Easy and Hard learning, The Boltzmann Machine Learning Algorithm, An Example of Hard Learning, Achieving Reliable Computation with Unreliable Hardware, and an Example of the Effects of Damage.
Artificial IntelligenceLearning and executing generalized robot plans
1,035 Citations1972Richard Fikes, Peter E. Hart +1 more
Some major new additions to the STRIPS robot problem-solving system are described, including a process for generalizing a plan produced by STriPS so that problem-specific constants appearing in the plan are replaced by problem-independent parameters.
DSpace@MIT (Massachusetts Institute of Technology)Learning Structural Descriptions From Examples
1,007 Citations1970Patrick Henry Winston
Machine LearningExplanation-Based Learning: An Alternative View
835 Citations1986Gerald DeJong, Raymond J. Mooney
Six specific problems with the previously proposed framework for the explanation-based approach to machine learning are outlined and an alternative generalization method to perform explanation- based learning of new concepts is presented.
Computer CompactsA theory and methodology of inductive learning
778 Citations1983Ryszard S. Michalski
National Conference on Artificial IntelligenceThe multi-purpose incremental learning system AQ15 and its testing application to three medical domains
761 Citations1986Ryszard S. Michalski, Igor Mozetič +2 more
The demonstration that by applying the proposed method of cover truncation and analogical matching, called TRUNC, one may drastically decrease the complexity of the knowledge base without affecting its performance accuracy is demonstrated.
Machine LearningExplanation-based generalization: A unifying view
718 Citations1986Tom M. Mitchell, Richard M. Keller +1 more
This paper proposes a general, domain-independent mechanism, called EBG, that unifies previous approaches to explanation-based generalization, and is illustrated in the context of several example problems, and used to contrast several existing systems for explanation- based generalization.
Artificial IntelligenceCommonsense reasoning about causality: Deriving behavior from structure
520 Citations1984Benjamin Kuipers
This paper presents a qualitative-reasoning method for predicting the behavior of mechanisms characterized by continuous, time-varying parameters that can detect a previously unsuspected landmark point at which the system is in stable equilibrium.
Machine LearningChunking in Soar: The Anatomy of a General Learning Mechanism
510 Citations1986John E. Laird, Paul S. Rosenbloom +1 more
The anatomy of chunking in Soar is examined and a new demonstration of its learning capabilities involving the acquisition and use of macro-operators is provided.
Machine LearningExplanation-based learning: An alternative view
488 Citations1986Gerald DeJong, Raymond J. Mooney
Six specific problems with the previously proposed framework for the explanation-based approach to machine learning are outlined and an alternative generalization method to perform explanation- based learning of new concepts is presented.
FigshareDerivational analogy : a theory of reconstructive problem solving and expertise acquisition
477 Citations2018Jaime Carbonell
Derivational analogy, a method of solving problems based on the transfer of past experience to new problem situations, is discussed in the context of other general approaches to problem solving.
DSpace@MIT (Massachusetts Institute of Technology)A Computational Model of Skill Acquisition
341 Citations1973Gerald Jay Sussman
Machine LearningLearning by Experimentation: Acquiring and Refining Problem-Solving Heuristics
339 Citations1983Tom M. Mitchell, Paul E. Utgoff +1 more
This chapter focuses on the issue of learning heuristics to guide a forward-search problem solver, and describes a computer program called lex, which acquires problem-solving Heuristics in the domain of symbolic integration.
Machine LearningChunking in Soar: The anatomy of a general learning mechanism
298 Citations1986John E. Laird, Paul S. Rosenbloom +1 more
The anatomy of chunking in Soar is examined and a new demonstration of its learning capabilities involving the acquisition and use of macro-operators is provided.
Shift of bias for inductive concept learning
288 Citations1984Paul E. Utgoff
It is shown that search for appropriate bias is itself a major part of the learning task, and that mechanical procedures for conducting a well-directed search for an appropriate bias can be created.
Elsevier eBooksAchieving Several Goals Simultaneously**The research reported herein was sponsored by the National Science Foundation under Grant GJ-36146.
263 Citations1981Richard Waldinger
It is argued that skeleton modelling is valuable as a planning technique apart from its use in plan modification, particularly because it facilitates the representation of “influential actions” whose effects may be far reaching.
International Joint Conference on Artificial IntelligenceLEAP: a learning apprentice for VLSI design
249 Citations1985Tom M. Mitchell, Sridhar Mahadevan +1 more
Munich Personal RePEc Archive (Ludwig Maximilian University of Munich)Understanding goal-based stories
242 Citations1980Robert Wilensky
PAM (Plan Applier Mechanism) is a computer program that understands stories by reasoning about the situations they reference, and produces representations for the stories that include the inferences needed to connect each story's events.
Defense Technical Information Center (DTIC)The structure-mapping engine
242 Citations1986Brian Falkenhainer, Kenneth D. Forbus +1 more
The SME is based on Gentner's Structure-Mapping theory of analogy, and provides a "tool kit" for constructing matching algorithms consistent with this theory, making it a candidate component for machine learning systems as well.
Machine LearningLearning at the Knowledge Level
202 Citations1986Thomas G. Dietterich
The paper analyzes two classes of learning programs, called symbol level learning and nondeductive knowledge level learning, and speculates on the possibility of developing coherent theories of each.
Artificial IntelligenceGeneralization learning techniques for automating the learning of heuristics
202 Citations1970D. A. Waterman
Procedures are developed which permit a problem-solving program employing heuristics in production rule form to learn to improve its performance by evaluating and modifying existingHeuristics and hypothesizing new ones, either during an explicit training process or during normal program operation.
Elsevier eBooksStrategy Learning with Multilayer Connectionist Representations
198 Citations1987Charles W. Anderson
A two-layer connectionist system is presented that develops its search from a weak to a task-specific strategy and fine-tunes its performance, applied to a simulated, real-time, balance-control task.
DSpace@MIT (Massachusetts Institute of Technology)Learning physical descriptions from functional definitions, examples, and precedents
198 Citations1983Patrick Henry Winston, Thomas O. Binford +2 more
The theory is a synthesis of two sets of ideas: ideas about learning from precedents and exercises developed at MIT and ideas about physical description developed at Stanford, and is illustrated by way of representative experiments.
Munich Personal RePEc Archive (Ludwig Maximilian University of Munich)Generalization and memory in an integrated understanding system
183 Citations1980Michael Lebowitz
An integrated system for language understanding IPP (Integrated Partial Parser), that encompasses the ability to generalize and record information in long-term memory as well as conceptual analysis is described.
Conceptual Clustering: Inventing Goal-Oriented Classifications of Structured Objects
181 Citations1986Robert E. Stepp, Ryszard S. Michalski
The AQ15 Inductive Learning System: An Overview and Experiments
171 Citations1986Ryszard S. Michalski, Igor Mozetič +2 more
International Joint Conference on Artificial IntelligenceSelectively generalizing plans for problem-solving
160 Citations1985Steven Minton
MORRIS is described, a heuristic problem solver that measures the utility of plan fragments to determine whether they are worth learning and comparing the performance of MORRIS to a less selective learning system.
Machine LearningThe Role of Heuristics in Learning by Discovery: Three Case Studies
158 Citations1983Douglas B. Lenat
The first case study presented, am, demonstrates that new domains of knowledge can be developed mechanically by using heuristics, and hypothesizes a more powerful plausible generation scheme that does function in that manner.
International Joint Conference on Artificial IntelligenceLearning and Problem Solving.
153 Citations1983Tom M. Mitchell
Cognitive ScienceIntegrated Learning: Controlling Explanation
150 Citations1986Michael Lebowitz
A combination of the two methods—applying explanation-based techniques during the course of similiarity-based learning can achieve the power of explanationbased learning without some of the computational problems that can otherwise arise in domains lacking detailed explanatory rules.
Cognitive ScienceData‐Driven Discovery of Physical Laws
147 Citations1981Pat Langley
BACON.3 has shown its generality by rediscovering versions of the ideal gas law, Kepler's third law of planetary motion, Coulomb's law, Ohm'slaw, and Galileo's laws for the pendulum and constant acceleration.
Munich Personal RePEc Archive (Ludwig Maximilian University of Munich)Causal and Teleological Reasoning In Circuit Recognition
143 Citations1979Johan de Kleer
A theory of human-like reasoning in the general domain of designed physical systems, and in particular, electronic circuits, is presented, which can construct by qualitative causal analysis a mechanism graph describing the functional topology of the system.
Machine LearningLearning at the knowledge level
135 Citations1986Thomas G. Dietterich
The paper analyzes two classes of learning programs, called symbol level learning and nondeductive knowledge level learning, and speculates on the possibility of developing coherent theories of each.
Cognitive ScienceConstraints and Preferences in Inductive Learning: An Experimental Study of Human and Machine Performance
112 Citations1987Douglas L. Medin, William D. Wattenmaker +1 more
The results show that human inductive generalizations tend toward greater specificity than would be expected if conceptual simplicity were the key constraint on inductions.
Cambridge University Press eBooksUse of analogy in a production system architecture
108 Citations1989John Anderson, Ross Thompson
Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign)A domain independent explanation-based generalizer
108 Citations1986Raymond J. Mooney, Scott Bennett
This method is compared and contrasted with other approaches to generalizing explanations, including an abstract version of the algorithm used in the STRIPS system and the EBG technique recently developed by Mitchell, Keller, and Kedar-Cabelli.
Artificial IntelligenceA new basis for state-space learning systems and a successful implementation
99 Citations1983Larry Rendell
A new basis for state-space learning systems is described which centers on a performance measure localized in feature space, and despite the absence of any objective function the parameter vector is locally optimal.
Machine LearningMachine Transformation of Advice Into a Heuristic Search Procedure
98 Citations1983David J. Mostow
This chapter is a case study in the mechanical mapping of domain-specific problems onto general methods, using as a detailed example the derivation of a heuristic search procedure for the advice “avoid taking points.”
Generalizations based on explanations
94 Citations1981Gerald DeJong
The system described here is designed to grasp some p r i n c i p l e under ly ing a na tu ra l language i n p u t in a new schema in the system.
Elsevier eBooksExtending Problem Solver Capabilities Through Case-Based Inference
90 Citations1987Janet L. Kolodner
Decisions that previously took several reasoning steps to make may be possible through application of a generalized schema, if abstract schemata can be derived from cases that have been seen previously, generalized knowledge can be augmented.
Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign)A general explanation-based learning mechanism and its application to narrative understanding
63 Citations1990Raymond J. Mooney
This thesis describes a general EBL mechanism, EGGS, which can make use of declarative knowledge stored in the form of horn clauses, rewrite rules, or STRIPS operators and reviews several recent psychological experiments which demonstrate that GENESIS can be productively interpreted as a cognitive model of certain types of human learning.
International Joint Conference on Artificial IntelligenceSubstantial constructive induction using layered information compression: tractable feature formation in search
61 Citations1985Larry Rendell
This paper addresses a problem of induction (generalization learning) which is more difficult than any comparable work in AI and achieves considerable generality with superior noise management and low computational complexity.
International Joint Conference on Artificial IntelligenceExplanation-based generalization in a logic-programming environment
58 Citations1987Haym Hirsh
A domain-independent implementation of explanation-based generalization (EBG) within a logic-programming environment where explanation is interleaved with generalization, so that as the training instance is proven to be a positive example of the goal concept, the generalization is simultaneously created.
National Conference on Artificial IntelligenceSelecting appropriate representations for learning from examples
56 Citations1986Nicholas S. Flann, Thomas G. Dietterich
A learning system that employs two different representations: one for learning and one for performance, and many fewer training instances are required to learn the concept, the biases of the learning program are very simple, and the learning system requires virtually no "vocabulary engineering" to learn concepts in a new domain.
International Joint Conference on Artificial IntelligenceThe use of explanations for similarity-based learning
51 Citations1987Andrea Danyluk
An integrated learning approach that combines Explanation-Based and Similarity-Based Learning methods to make learning in an inherently complex domain feasible is proposed.
BAC0N.5: the discovery of conservation laws
51 Citations1981Pat Langley, Gary L. Bradshaw +1 more
BACON.5 is a program that discovers empirical laws by applying a small set of data-driven heuristics to detect regularities in numeric and nominal data, leading BACONS to formulate hypotheses, define theoretical terms, and postulate intrinsic properties.
eScholarship (California Digital Library)Purpose-Directed Analogy
38 Citations1985Kedar-Cabelli, Smadar
Mechanical transformation of task heuristics into operational procedures
37 Citations1981David Jechiel Mostow
This dissertation characterizes the operationalization of advice as a series of problem transformations leading from the advice statement to a procedure for achieving a specified goal (avoid taking points) or evaluating a specified quantity (decide whether an opponent is void).
Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign)Schema Acquisition from One Example: Psychological Evidence for Explanation-Based Learning.
31 Citations1987Woo‐kyoung Ahn, Raymond J. Mooney +2 more
Experimental evidence is presented that people can learn a plan schema from a single narrative and that the learned schema agrees with that predicted by EBL, suggesting that GENESIS, originally constructed as a machine learning system, can be interpreted as a psychological model of learning a complex schemaFrom a single example.
International Joint Conference on Artificial IntelligenceMs. maloprop, a language comprehension program
31 Citations1977Eugene Charniak
International Joint Conference on Artificial IntelligenceDISCIPLE-1: interactive apprentice system in weak theory fields
29 Citations1987Yves Kodratoff, Gheorghe Tecuci
A first prototypical version of DISCIPLE and its use to the design of technologies for the manufacturing of loudspeakers is presented and demonstrates not only that over-generalization is harmless but also useful and necessary, when interacting with a user.
International Joint Conference on Artificial IntelligenceUsing prior learning to facilitate the learning of new causal theories
24 Citations1987Michael J. Pazzani, Michael Oyer +1 more
An approach to learning causal knowledge which lies in between two extremely different approaches to learning, which can create causal theories which are not implications of existing causal theories and which snares a goal with explanation-based learning.
Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign)Generalizing the structure of explanations in explanation-based learning
21 Citations1988Jude Shavlik
This research brings explanation-based learning closer to its goal of being able to acquire the full concept inherent in the solution to a specific problem.
International Joint Conference on Artificial IntelligenceApproximation in mathematical domains
21 Citations1987Scott Bennett
The approximation method strongly supports the use of a mix of quantitative and qualitative reasoning over either a purely quantitative or qualitative approach when dealing with the intractable domain problem in mathematical domains.
Some thoughts on schemata
13 Citations1975Wallace Chafe
Some ideas and questions that have arisen from efforts over the past few years to develop a model of what I call verbalization: the set of processes by which a person converts nonverbal knowledge into a verbal output are summarized.
Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign)Simplifications in Temporal Persistence: An Approach to the Intractable Domain Theory Problem in Explanation-Based Learning
12 Citations1987Steve Chien
This thesis describes an approach to problem-solving and learning designed to deal with computationally ill-behaved domains and first implementation of this approach, which learns by constructing and generalizing causal explanations of observed plans.
eScholarship (California Digital Library)Explanation and Generalization Based Memory
10 Citations1985Pazzani, Michael J.
Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign)Explanation-based learning via constraint posting and propagation
9 Citations1987Paul O’Rorke
This thesis presents an idealized model of explanation-based learning centered on the notion of constraint posting and propagation, where problems are solved by posting constraints and solutions are generalized by eliminating unnecessary constraints.
Elsevier eBooksLearning and Reusing Explanations
9 Citations1987Kristian J. Hammond
This paper suggests an alternative approach in which the connections between expectation failures and their causes are saved and reused to explain similar failures that occur in the future.
…