Review

Abstract

The NGE learning model represents mainstream AI work on machine learning; that is, it is an instance of symbol processing computational models which learn to perform a task or categorize a set of objects. However, work in other fields has tackled similar problems, usually with very different techniques. Sometimes other fields use formal models quite similar to those used by researchers within AI, but generally they are very different. In this chapter, I overview work in AI, psychology, and statistics that attempts to solve the problem of learning categories or concepts. The purposes of this review are (1) to show how the exemplar-based learning theory fits into AI research, and (2) to highlight areas in which this research, other AI work, and work in related fields are complementary. Obviously, each field needs to be aware of the successes (and failures) of the others, in order to avoid re-inventing algorithms or theories that have been explored elsewhere.