Automatic discounting rate computation using a dissent criterion

John Klein (1) , Olivier Colot (1)

1 LAGIS-SI

An iterative method for automatic discounting rate computation is introduced in this paper. As part of a conjunctive combination of sources, discounting reduces the impact of a source on the result of the fusion. The key problems related to discounting are: what sources must be discounted and up to what degree. We propose a method that jointly tackles these two issues by computing rates using a dissent measure. The dissension of the sources is evaluated by comparing them to the average of their mass functions. Sources very different from others (like sources resulting from outliers) are thus heavily discounted. This discounting process is iterated so as to obtain a minimum distance between the conjunctive combination of the discounted sources and a categorical mass function. The efficiency of the method is evaluated through several simulations.

Mots clés: belief functions; discounting

Domaines: Analyse de données, Statistiques et Probabilités [physics.data-an]

John Klein, Olivier Colot. Automatic discounting rate computation using a dissent criterion. Workshop on the theory of belief functions, Mar 2010, Brest, France. pp.1-6. ⟨hal-00668110⟩