The main contributions of this thesis are a Bayesian theory of learning classification rules, the unification and comparison of this theory with some previous theories of learning, and two extensive applications of the theory to the problems of learning class probability trees and bounding error when learning logical rules. The thesis is motivated by considering some current research issues in machine learning such as bias, overfitting and search, and considering the requirements placed on a learning system when it is used for knowledge acquisition. Basic Bayesian decision theory relevant to the problem of learning classification rules is reviewed, then a Bayesian framework for such learning is presented. The framework has three components: the hypothesis space, the learning protocol, and criteria for successful learning. Several learning protocols are analysed in detail: queries, logical, noisy, uncertain and positive-only examples. The analysis is done by interpreting a protocol as a...

Abstract not found

Multistrategy task-adaptive learning (MTL) comprises a class of methods in which the learner determines by itself which strategy or combination of strategies is most appropriate for a given learning task defined by the learner's goal, the leamer's background knowledge (BK) and the input to the learning process. The paper presents a MTL method which is based on building a plausible justification that the learner's input is a consequence of its BK. The method assumes a general learning goal of deriving any useful knowledge from a given input and integrates dynamically a whole range of learning sategies. It also behaves as a singlestrategy method when the relationship between the input and the BK satisfies the requirements of the single-strategy method, and the general learning goal of the MTL method is specialized to the goal of the single-strategy method.

The paper presents an approach to the automation of knowledge acquisition for expert systems. The approach is based on several general principles emerging from the field of machine learning: expert system building as a three phase process, understanding-based knowledge extension, knowledge acquisition through multistrategy learning, consistency-driven concept formation and refinement, closed-loop learning. and synergistic cooperation between the human expert and the learning system. In this approach, an expert system is built by a human expert and a learning system. The human expert defmes the framework for the expert system and provides an incomplete and partially incorrect knowledge base. The learning system incrementally extends, updates, and improves the knowledge base through learning from the human expert. This approach is illustrated by the learning system shell NeoDISCIPLE.

This paper presents the results of a multi-faceted research and development effort that synergistically integrates artificial intelligence research with military strategy research and practical deployment of agents into education. It describes recent advances in the Disciple approach to agent development by subject matter experts with limited assistance from knowledge engineers, the innovative application of Disciple to the development of agents for strategic center of gravity analysis, and the deployment and evaluation of these agents in several courses at the US Army War College

As knowledge-based systems are addressing increasingly complex domains, their roles are shifting from classical expert systems to interactive assistants. To develop and maintain such systems, an integration of thorough knowledge acquisition procedures and sustained learning from experience is called for. A knowledge level modeling perspective has shown to be useful for analyzing the various types of knowledge related to a particular domain and set of tasks, and for constructing the models of knowledge contents needed in an intelligent system. To be able to meet the requirements of future systems with respect to robust competence and adaptive learning behavior, particularly in open and weak theory domains, a stronger emphasis should be put on the combined utilization of casespecific and general domain knowledge. In this chapter we present a framework for integrating KA and ML methods within a total knowledge modeling cycle, favoring an iterative rather than a top down approac...

This paper presents an approach to domain modeling and knowledge acquisition that consists of a gradual and goal-driven improvement of an incomplete domain model provided by a human expen. Our approach is based on a multistrategy learning method that allows a system with incomplete knowledge to learn general inference or problem solving rules from specific facts or problem solving episodes received from the human expen. The system will learn the general knowledge pieces by considering all their possible instances in the current domain model. trying to learn complete and consistent descriptions. Because of the incompleteness of the domain model the learned rules will have exceptions that are eliminated by refining the definitions of the existing concepts or by defining new concepts.

The paper explores several general issues in developing a multistrategy task-adaptive learning (MTL) system. The system aims at integrating a whole range of learning strategies, such as explanation-based learning, empirical generalization, abduction, constn3. ctive induction, learning by analogy and abstraction. The integration is dynamic, i.e. the way different strategies are evoked depends on the learning task at hand. The key idea of the learning method is that the learner tries to "understand" the put in terms of its current knowledge, and then uses this understandihg to improve the knowledge. This process may involve both certain and plausible reasoning. The paper extends and generalizes the previous work on this topic.

Abstract. This paper presents an experiment of parallel knowledge base development by subject matter experts, performed as part of the DARPA’s Rapid Knowledge Formation Program. It introduces the Disciple-RKF development environment used in this experiment and proposes design guidelines for systems that support authoring of problem solving knowledge by subject matter experts. Finally, it compares Disciple-RKF with the other development environments from the same DARPA program, providing further support for the proposed guidelines. 1

This paper presents a learning-based approach to the automation of knowledge acquisition for expert systems. An expert system is viewed as an explicit mooel of a human expert's competence and perfonnance. We distinguish three phases in the development of such a model. The fIrst one consists of defIning a framework for the mooel, in terms of a knowledge representation formalism and an associated problem solving methoo. The second phase consists of defIning a preliminary mooel that describes the basic concepts of the expertise domain. The last phase consists of incrementally extending and improving the domain model through learning from the human expert. The paper describes the learning system NeoDISCIPLE which illustrates the usefulness of six principles for automating the knowledge acquisition process: expert system building as a threephase mooeling of human expertise, understanding-based knowledge extension, knowledge acquisition through multistrategy learning, consistency-driven concept fonnation and refinement, closed-loop learning, and cooperation between the human expert and the learning system.