G. Iuculano, A. Lazzari, G. Pellegrini, A. Zanobini
THE EVALUATION OF MEASUREMENT UNCERTAINTY AND THE PRINCIPLE OF MINIMUM JOINT CROSS-ENTROPY
A measurement process represents a controlled learning process in which various aspects on uncertainty analysis are investigated.
A measurement process is performed if information supplied by it is likely to be considerably more accurate, stable and reliable than the pool of information already available.
The substantial amount of information, got with respect to the conditions prior to the result after the measurement process is performed, can be connected to the "Kullback's principle of minimum cross-entropy".
This, as it is known, is a correct method of inductive inference when no sufficient knowledge about the statistical distributions of the involved random variables is available before the measurement process is carried out except for the permitted ranges, the essential model relationships and some constraints, gained in past experience, valuable usually in terms of expectations of given functions or bounds on them.
In this paper the authors pointed out the connection between the evaluation of the uncertainty in a repeated measurements process and the "Kullback's principle of minimum cross-entropy".