Goal-Driven Learning

by Ram, Ashwin; Leake, David B.

• ISBN13: 9780262181655
• ISBN10: 0262181657
• Format: Hardcover
• Copyright: 1995-08-30
• Publisher: Bradford Books

Summary

In cognitive science, artificial intelligence, psychology, and education, a growing body of research supports the view that the learning process is strongly influenced by the learner's goals. The fundamental tenet of goal-driven learning is that learning is largely an active and strategic process in which the learner, human or machine, attempts to identify and satisfy its information needs in the context of its tasks and goals, its prior knowledge, its capabilities, and environmental opportunities for learning. This book brings together a diversity of research on goal-driven learning to establish a broad, interdisciplinary framework that describes the goal-driven learning process. It collects and solidifies existing results on this important issue in machine and human learning and presents a theoretical framework for future investigations. The book opens with an an overview of goal-driven learning research and computational and cognitive models of the goal-driven learning process. This introduction is followed by a collection of fourteen recent research articles addressing fundamental issues of the field, including psychological and functional arguments for modeling learning as a deliberative, planful process; experimental evaluation of the benefits of utility-based analysis to guide decisions about what to learn; case studies of computational models in which learning is driven by reasoning about learning goals; psychological evidence for human goal-driven learning; and the ramifications of goal-driven learning in educational contexts. The second part of the book presents six position papers reflecting ongoing research and current issues in goal-driven learning. Issues discussed include methods for pursuing psychological studies of goal-driven learning, frameworks for the design of active and multistrategy learning systems, and methods for selecting and balancing the goals that drive learning. A Bradford Book

Table of Contents

Foreword

Preface

Sources and Acknowledgments

Contributors

Learning, Goals, and Learning Goals

Why Goals?

An Everyday Example

Toward A Planful Model Of Learning

A Framework For Goal-Driven Learning

Major Issues In Goal-Driven Learning

What Is A Goal?

Types Of Goals

Task Goals

Learning Goals

Specifications, Policies, and Constraints

A Unifying View

Role Of Goals In Learning

Guiding the Performance Task

Guiding the Learning Task

Guiding Storage

Pragmatic Implications of Goal-Driven Learning

Summary

Acknowledgments

Notes

References

Current State of the Field

Planning to Learn

Introduction

Learning Requires Decision Making

Related Previous Research

Learning Goals

Planning To Learn

Learning Actions

Learning Resources

Conclusion: Decision Making In Learning

Note

References

Quantitative Results Concerning the Utility of Explanation-Based Learning

Introduction

Ebl And The Utility Problem

Initial Experiments With the Utility Problem

Will the Utility Problem Go Away?

Overview of Prodigy

The Prodigy EBL Component

Selecting What to Learn and Generating an Initial Explanation

Compression: Improving an Explanation

Evaluating the Utility of an Explanation

Performance Results

Evaluating the System's Components

Comparison with Macro-Operators

Critique of the Ebl Method

Multiple Types of Explanations?

Compression

Utility Evaluation

Critique Of The Experimental Methodology

Conclusion

Acknowledgment

Notes

References

The Use of Explicit Goals for Knowledge to Guide Inference and Learning

The Focus of Attention Problem

Inference and Desires for Knowledge

Do People Have Goals for Knowledge?

Computer Programs With Knowledge Goals: Two Case Studies

AQUA

IVY

A Theory of Knowledge Goals

Concept Specification

Task Specification

The Origins of Knowledge Goals

Using Knowledge Goals to Guide Processing

Theory of Inference Control

Mechanisms for Knowledge Goal Management

Retrieving Knowledge Goals

Knowledge Goals As A Theory Of Interestingness

Knowledge Planning: Learning Through the Satisfaction of Knowledge Goals

Comparison to Other Approaches

Conclusion: Automating Curiosity

Acknowledgments

Notes

References

Deriving Categories to Achieve Goals

Introduction

Overview

Exemplar Learning and Conceptual Combination as Modes of Category Acquisition

Structure of Goal-Derived Categories

Prototype Structure in Common Taxonomic Categories

Prototype Structure in Goal-Derived Categories

Stability of Prototype Structure in Common Taxonomic and Goal-Derived Categories

Determinants of Prototype Structure in Common Taxonomic and Goal-Derived Categories

Goal-Derived Categories in Planning

Frame Instantiation

Role of Goal-Derived Categories in Frame Instantiation

Factors That Guide the Selection of Instantiations

Deriving Ad Hoc Categories

Fields of Goal-Derived Categories

Roles of Common Taxonomic and Goal-Derived Categories in the Cognitive System

Time Course of Categorization: Primary vs. Secondary Categorizations

Lexicalization and the Time Course of Categorization

Goal-Derived Categories as the Interface between Event Frames and World Models

Conclusion

Acknowledgments

Notes

References

Harpoons and Long Sticks: The Interaction of Theory and Similarity in Rule Induction

Introduction

Rule Induction Paradigms

Simultaneous and Sequential Induction

Using Theories to Determine the Feature Space

Supervised Versus Unsupervised Learning

Study 1: Simultaneous Rule Induction

Goals of the Study

Method

Results

Discussion

Study 2: Sequential Rule Induction

Method

Results

Discussion

Models Of Theoretical And Empirical Learning

Previous Models

A Tightly Coupled Approach

Summary

Acknowledgments

Notes

References

Introspective Reasoning Using Meta-Explanations for Multistrategy Learning

Introduction

Overview of the Approach

Example: The Drug Bust

The Explanation-Based Understanding Task

Representation of Trace Meta-XPs

Representation of Introspective Meta-XPs

Successful Prediction

Expectation Failure

Retrieval Failure

Incorporation Failure

Associating Knowledge Goals and Learning Strategies with Introspective Meta-XPs

Mis-indexed Knowledge Structure

Novel Situation

Incorrect Background Knowledge

Combinations and Extensions

Related Work

Discussion and Future Research

Conclusions

Acknowledgments

Notes

References

Goal-Directed Learning: A Decision-Theoretic Model for Deciding What to Learn Next

Introduction

Utility Of Learning Goals

Performance Element

Utility Of Plans

An Example in the Ralph World

Related Work

Conclusions

References

Goal-Based Explanation Evaluation

Introduction

Overview

Previous Perspectives

Psychological Approaches

Philosophical Views

AI Approaches

A Theory of Goal-Based Evaluation

A Computer Model

ACCEPTER's Basic Algorithm

Representation of Explanations

Explanation Evaluation For Routine Understanding

Evaluation for Showing an Event's Reasonableness

Evaluation for Identifying Flaws in Prior Reasoning

Beyond Routine Understanding: Evaluation for More Specific Goals

Evaluation Dimensions for Prediction

Evaluation Dimensions for Repair

Evaluation Dimensions for Control

Evaluation Dimensions for Actors' Contributions to an Outcome

Summary of Evaluation Dimensions

The Value of Goal-Based Evaluation

Focusing Explanation toward Knowledge Gaps

Guiding Explanation Construction in Case-Based Explanation

Dealing with the Imperfect Theory Problem

Conclusion

Notes

References

Planning to Perceive

Why Gather Information?

How to Gather Information

Deciding to Gather Information

Using Heuristics

Related Work

Discussion

Acknowledgments

Notes

References

Planning and Learning in PRODIGY: Overview of an Integrated Architecture

Introduction

The Prodigy Architecture

The Problem Solver

Knowledge Representation

Problem Definition and Problem Solving

Control Rules

The Learning Modules

Dimensions Of The Architecture

Comparison With Other Architectures

Acknowledgments

References

A Learning Model for the Selection of Problem-Solving Strategies in Continuous Physical Systems

Introduction

IOPS Model

Adaptation and Learning in IOPS Model

System Characterization

Adaptation within IOPS Model

Learning within the IOPS Model

An IOPS-Based Implementation

Automated Robotic Assembly: An Experimental Domain

An IOPS-Based Scheduler for Assembly Robots

Scheduling as Selecting Planned Actions

Scheduling as Patching up Subplans for Subassemblies

Adaptation and Learning in the lOPS-Scheduler

Discussion and Conclusion

Notes

References

Explicitly Biased Generalization

Introduction

The Select-Inhibit Method of Concept Learning

Generalization Heuristics For Biasing Induction

Axioms

General Definitions

The Irrelevance Generalization Heuristic

The Cohesion Generalization Heuristic

The Independence Generalization Heuristic

Cautious and Uncautious Heuristic Modes

Bias and Active Learning, Analytical Learning, Error Resolution

Active Learning

Analytical Learning

Error Resolution

Partial Program Trace

Results

Related Work On Bias

Future Work

Summary

Consistency Requirement And Definition Of Consistency-Preserving

Irrelevance Heuristic

Irrelevance Heuristic Condition

Irrelevance Generalization Operator

Consistency Theorem for Irrelevance Heuristic

Independence Heuristic

Indepencence Condition

Independence Generalization Operator

Consistency Theorem for Independence Heuristic

Acknowledgments

Notes

References

Three Levels of Goal Orientation in Learning

Method

Participants

Materials

Procedure

Protocol Analysis

Results

Differences in Levels of Goal Orientation

Differences among Goal Orientation Groups: Goal Cue Selections and Performance

Differences among Goal Orientation Groups: Approaches to Learning and Sensitivity to Evaluat...

Background Factors and Goal Orientation

Discussion

Acknowledgments

Note

References

Characterizing the Application of Computer Simulations in Education: Instructional Criteria

Introduction

General Classification Of Models

Introduction

Quantitative Versus Qualitative Models

Quantitative Models

Qualitative Models

Interaction and Scenarios

Learning Goals for Simulations

Introduction

The Concept of a "Learning Goal"

A Classification of Simulation Learning Goals

Learning about Knowledge Acquisition

Learning Processes In Exploratory Learning Environments

Learner Activity and Simulations

Notes

References

CURRENT RESEARCH AND RECENT DIRECTIONS

Goal-Driven Learning: Fundamental Issues and Symposium Report

Introduction

What Is Goal-Driven Learning?

Effects of Goal-Driven Learning on Human and Machine Learning

Issues In Goal-Driven Learning

What Are the Types of Learning Goals?

How Do Learning Goals Arise?

How Do Learning Goals Affect the Learning Process?

How Do Different Types of Learning Goals Relate to Each Other?

How Are Learning Goals Represented?

The Properties Of Goal-Driven Learners

The Relationship Between Goal-Driven And Non-Goal-Driven Learning

Conclusions

Acknowledgments

Notes

References

Storage Side Effects: Studying Processing to Understand Learning

Intentions And Information Storage

The Extent of Storage Side Effects

The Ubiquity of Goal-Driven Storage

Understanding Goal-Driven Storage Through Analysis Of Goal-Driven Processing

Processing and Storage in Classic Problem Solving

Processing and Storage in Implicit Orientation

Conclusion

Acknowledgments

References

Goal-Driven Learning in Multistrategy Reasoning and Learning Systems

Introduction

The Nature Of Goal-Driven Learning

An Architecture for Introspective Multistrategy Learning

A Taxonomy of Reasoning Failures

Two Case Studies

Meta-TS

Meta-AQUA

Conclusions

Acknowledgments

Notes

References

Inference to the Best Plan: A Coherence Theory of Decision

Howard's Dilemma

Principles Of Deliberative Coherence

Computational Implementation

Comparison With Classical Decision Theory

Conclusion: Goals And Learning

Acknowledgments

References

Toward Goal-Driven Integration of Explanation and Action

Introduction

Isolated versus Interactive Explanation

Chapter Overview

Real World Explanation: An Everyday Example

Explanations and Explanation Generation

The Need for Interaction with the Environment to Support Explanation

The Need for Incremental Interaction between the Explanatory Process and Explainer Goals

Issues in Building A Model of Goal-Driven Interactive Explanation

Content Theories for GDIE

A Simple Illustration

A Broader Perspective

Progress Toward The Model

ACCEPTER

Deciding When to Explain

Formulating Knowledge Goals to Guide Search for Explanations

Extending the Model

Relationship to Other Approaches

Conclusions

Acknowledgments

Notes

References

Learning as Goal-Driven Inference

Inferential Theory of Learning

Learning As Goal-Guided Inference

Learning Goals

Goal Dependency Networks

Knowledge Transmutations: Inferential Primitives for Learning

Toward A Computational Model of Goal-Driven Machine Learning

Index

Table of Contents provided by Publisher. All Rights Reserved.