page 1  (20 pages)
2to next section

The Functional Transfer of Knowledge for

Coronary Artery Disease Diagnosis

Daniel L. Silver (1,2) and Robert E. Mercer (1) and Gilbert A. Hurwitz (2,3) (1) Department of Computer Science, University of Western Ontario,
(2) Department of Nuclear Medicine, Victoria Hospital, and
(3) Departments of Diagnostic Radiology and Nuclear Medicine,
University of Western Ontario,
London, Ontario, Canada N6A 3K7
email: dsilver@csd.uwo.ca

January 15, 1997

Abstract

A distinction between two forms of task knowledge transfer, representational and functional, is reviewed followed by a discussion of ?MTL, a modified version of the multiple task learning (MTL) neural network method of functional transfer. The ?MTL method employs a separate learning rate, ?k, for each task output node k. ?k varies as a function of a measure of relatedness, Rk, between the kth task and the primary task of interest. An ?MTL network is applied to a diagnostic domain of four levels of coronary artery disease. Results of experiments demonstrate the ability of ?MTL to develop a predictive model for one level of disease which has superior diagnostic ability over models produced by either single task learning or standard multiple task learning.

Keywords: task knowledge transfer, artificial neural networks, knowledge based inductive bias, task relatedness, medical decision making