On the Estimation of Mass Functions Using Self Organizing Maps

In this paper, an innovative method for estimating mass functions using Kohonen’s Self Organizing Map is proposed. Our approach allows a smart mass belief assignment, not only for simple hypotheses, but also for disjunctions and conjunctions of hypotheses. This new method is of interest for solving estimation mass functions problems where a large quantity of multi-variate data is available. Indeed, the use of Kohonen map that allows to approximate the feature space dimension into a projected 2D space (so called map) simplifies the process of assigning mass functions. Experimentation on a benchmark database shows that our approach gives similar or better results than other methods presented in the literature so far, with an ability to handle large amount of data.