This paper is concerned with the use of Dempster-Shafer theory in 'fusion'
classifiers. We argue that the use of predictive accuracy for basic probabi
lity assignments can improve the overall system performance when compared t
o 'traditional' mass assignment techniques. We demonstrate the effectivenes
s of this approach in a case study involving the detection of static thermo
static valve faults in a diesel engine cooling system. (C) 2001 Elsevier Sc
ience B.V. All rights reserved.