Technical Reports - Query Results

ÖFAI-TR-92-06 ( 412kB g-zipped PostScript file)

Learning with a Qualitative Domain Theory by Means of Plausible Explanations

Gerhard Widmer

This chapter describes an approach to learning on the basis of a qualitative domain theory. The theory consists of a mixture of strict rules and general dependency statements. The domain theory supports plausible explanations of training instances. These explanations are used to create initial concepts via a kind of `plausible EBG', and also to guide subsequent empirical generalization of learned concepts. The method has been implemented in a system that learns to solve complex problems in the domain of tonal music. This chapter presents the application domain, describes the learning method (with special emphasis on the plausible inference strategies used), presents empirical results, and shows how this approach naturally leads to a framework for multistrategy learning.

Citation: Widmer G.: Learning with a Qualitative Domain Theory by Means of Plausible Explanations, In Machine Learning: A Multistrategy Approach, Vol. IV, (R.S. Michalski, G. Tecuci, Eds.), San Mateo, CA, Morgan Kaufmann, 1994.