The probability assessments of a Bayesian belief network generally include inaccuracies. These inaccuratcies influence the reliability of the network's ouput. an integral part of investigating the output's reliability is to study its robustness. Robustness pertains to the extent to which varying the probability assessments of the network influences the ouput. It is studied by subjecting the network to a senitivity analysis. In this paper, we address the issue of robustness of a belief network's output in view of the threshold model for decision amking. We present a method for sensitivity analysis that provides for the computation of bounds between which a network's assessments can be varied without inducing a change in recommended decision.
展开▼
