Credal networks for military identification problems
Authors: Alessandro Antonucci, Brühlmann, R., Alberto Piatti and Marco Zaffalon
Abstract: Credal networks are imprecise probabilistic graphical models generalizing Bayesian networks to convex sets of probability mass functions. This makes credal networks particularly suited to capture and model expert knowledge under very general conditions, including states of qualitative and incomplete knowledge. In this paper, we present a credal network for risk evaluation in case of intrusion of civil aircrafts into a no-fly zone. The different factors relevant for this evaluation, together with an independence structure over them, are initially identified. These factors are observed by sensors, whose reliabilities can be affected by variable external factors, and even by the behavior of the intruder. A model of these observation mechanisms, and the necessary fusion scheme for the information returned by the sensors measuring the same factor, are both completely embedded into the structure of the credal network. A pool of experts, facilitated in their task by specific techniques to convert qualitative judgments into imprecise probabilistic assessments, has made possible the quantification of the network. We show the capabilities of the proposed network by means of some preliminary tests referred to simulated scenarios. Overall, we can regard this application as an useful tool to support military experts in their decision, but also as a quite general imprecise-probability paradigm for information fusion.
Details: In de Cooman, G, Vejnarová, J., Zaffalon, M. (Eds.), ISIPTA '07: Proceedings of the fifth International Symposium on on Imprecise Probability: Theories and Applications. Action M Agency, pp. 1-10.
A version similar to the published paper can be downloaded.