Digestive database evidential clustering based on possibility theory
Anas Dahabiah, John Puentes, Basel Solaiman · 2008
Abstract:- A new method that aims to automatically classify a set of objects in spite of the imperfection and the uncertainty of their heterogeneously-assigned data is proposed in this paper. This method is based essentially on possibility theory to estimate the similarity among the objects, and on belief theory and multidimensional scaling methods to construct the compatible evidential class partition. This method is applied to a medical database and robust results have been obtained without knowing the key attributes of the concerned pathologies and without taking into account any a priori medical knowledge.