Learning comprehensible theories from structured data

This tutorial discusses some knowledge representation issues in machine learning. The focus is on machine learning applications for which the individuals that are the subject of learning have complex structure. To represent such individuals, a rich knowledge representation language based on higher-order logic is introduced. The logic is also employed to construct comprehensible hypotheses that one might want to learn about the individuals. The tutorial introduces the main ideas of this approach to knowledge representation in a mostly informal way and gives a number of illustrations. The application of the ideas to decision-tree learning is also illustrated with an example.