论文信息 - A tight integration of deductive and inductive learning

A tight integration of deductive and inductive learning

The paper presents a novel approach to combining deductive and inductive learning strategies in a flexible way. Our method is best described under the general heading ‘Explain and Compile’: by loosening our notion of an explanation, we arrive at a uniform learning framework which provides for a flexible combination of inductive and deductive tendencies. The algorithm was developed in the context of an interactive Learning Apprentice System for two-voice counterpoint composition.

Gerhard Widmer | G. Widmer

[1] Michael J. Pazzani,et al. Integrating Explanation-Based and Empirical Learning Methods in OCCAM , 1988, EWSL.

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

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

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

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

[6] Michael Lebowitz,et al. Integrated Learning: Controlling Explanation , 1986, Cogn. Sci..

[7] Thomas G. Dietterich,et al. Selecting Appropriate Representations for Learning from Examples , 1986, AAAI.

[8] Stuart J. Russell. Analogical and inductive reasoning , 1987 .