UNCERTAINTY AND A NEW MEASURE FOR CLASSIFICATION UNCERTAINTY
Haluk Damgacioglu, Cem İyigün · World Scientific proceedings series on computer engineering and information science · 2012
Ben-Israel and Iyigun ([1] and [2]) presents a new clustering method which is probabilistic distance clustering (P-D Clustering). In this method, the probability of assignment to cluster for each point is inversely proportional to distances between data point and centers of clusters according to given number of clusters and their centers. In this paper, we study on new uncertainty measure for classification using the assignment probabilities of P-D Clustering. Moreover, the relationship of the new measure with Kullback - Liebner divergence is discussed.