This paper presents a new classifier combination technique based on theDempster-Shafer theory of evidence. The Dempster-Shafer theory of evidence is apowerful method for combining measures of evidence from different classifiers.However, since each of the available methods that estimates the evidence ofclassifiers has its own limitations, we propose here a new implementation whichadapts to training data so that the overall mean square error is minimized. Theproposed technique is shown to outperform most available classifier combinationmethods when tested on three different classification problems.
展开▼