The advent of automatic data classification can be traced back to the early sixties, namely to a software tool for discriminating abnormal from normal electrocardiograms which was actually in routine use for some time . Two important resources were then available, a theoretical one — Bayesian decision theory —, and a practical one — computational facilities, albeit extremely modest by present day standards.
KeywordsLoss Function Class Label Multilayer Perceptrons Bayesian Decision Theory Risk Functional
Unable to display preview. Download preview PDF.