The University of Sussex

A computational model for the data compression metaphor

Christopher Thornton

Although it is often noted that the process of data compression forms an illuminating metaphor for concept learning mechanisms, the computational basis of the metaphor is not usually made explicit. In this paper, an algorithm is described which implements "climbing-tree" inductive generalisation as a form of clustering process. The algorithm is correctly classified as a generalised data compression algorithm; it therefore provides a plausible model for the data compression metaphor.


This paper is not available online