Accuracy of Sequential Bayesian Information Fusion
Jan R. J. Nunnink, Gregor Pavlin, M.H. Hamza · UvA-DARE (University of Amsterdam) · 2005
Sequential Bayesian information fusion is a process in which streams of observations from multiple sources are fused in order to form a more complete and accurate situation assessment. Allthough the information sources are potentially unreliable, and we generally have little control over them, it is possible to directly influence the accuracy of the fusion through the parameters of the Bayesian network (BN). In this paper we analyse the expected accuracy of decision making based on fusion with semi-dynamic BNs, in terms of the model parameters and decision thresholds.