Some classification algorithms integrating Dempster-Shafer theory of evidence with the rank nearest neighbor rules

We propose five different ways of integrating Dempster-Shafer theory of evi dence and the rank nearest neighbor classification rules with a view to exp loiting the benefits of both. These algorithms have been tested on both rea l and synthetic data sets and compared with the k-NN, m-MRNN, and k-NNDST, which is an algorithm that also combines Dempster-Shafer theory with the le -NN rule. If different features have widely different variances then the di stance-based classifier, algorithms like Ic-NN and k-NNDST may not perform weil, but in this case the proposed algorithms are expected to perform bett er. Our simulation results indeed reveal this. Moreover, the proposed algor ithms are found to exhibit significant improvement over the m-MRNN rule.