Nearest Neighbor Classifier based on Generalized Inter-point Distances for HDLSS Data
share
In high dimension, low sample size (HDLSS) settings, Euclidean distance based classifiers suffer from curse of dimensionality if the competing populations are similar in terms of their location and scale parameters. In this article, we propose a classifier which can discriminate between populations having different marginal distributions. A generalization of the proposed classifier has been developed to differentiate between populations having different joint distributions for the component variables. Performance of the proposed classifiers are demonstrated using a variety of simulated data sets.
READ FULL TEXT