ON THE USE OF MINIMUM CROSS ENTROPY PRINCIPLE AND BAYES’ THEOREM FOR THE UNCERTAINTY EVALUATION IN A MEASUREMENT PROCESS

G. Iuculano, G. Pellegrini, A. Zanobini

Abstract

In this paper the evaluation of measurement uncertainty in a multivariate model is carried out by applying the principle of minimum cross entropy (MINCENT) and Bayes’ theorem.<BR />
In particular the MINCENT optimization procedure is used to translate the information contained in the known form of likelihood into a prior distribution for Bayesian inference. The methodology is adapted and tested on a recalibration model. Some basic ideas and general remarks on the Bayesian probability theory and entropy optimization principles are reported too.

Keywords

Measurement Uncertainty, Bayesian Inference, Minimum cross Entropy

Download

PWC-2006-TC21-008u.pdf

DOI

-

IMEKO TC

TC21 - Mathematical Tools for Measurements

Back to the proceedings