Max-Ent in fast belief fusion
M. Pappalardo, Francesco Villecco · 2008
This paper deals with Jaynes' MaxEnt Principle and Yager's combination rule for weighting the masses associated with the focal ele- ments. The main goal is to manage uncertainty by suitably modelling the degrees of belief. The solution of this problem is classically approached by the Bayesian method based on probability functions. The DS theory also provides a useful framework for the representation of information con- tent in an uncertain variable. However in both approaches the weights, associated with the focal elements can be referred to probabilities, and must be know in advance. We show that, under Max-Entropy principle, the fusion of belief is faster. We present a new way of obtaining from two independent and equally reliable sources of evidence, a new set with the probability allotted on each focal element.