Selection of perturbation experiments for model discrimination
Proceedings of Fourteenth European Conference on Artificial Intelligence, ECAI 2000, W. Horn (ed.), IOS Press, Amsterdam, 191-195, 2000
I. Vatcheva, H. de Jong, N. Mars
In many situations a system is described by several competing models. In order to distinguish among the proposed models, further information about the behavior of the system is required. One way to obtain such information is to perform suitably chosen perturbation experiments. This paper introduces a method for the selection of optimal perturbation experiments for discriminating among a set of competing dynamical models. The models are assumed to have the form of semi-quantitative differential equations. The method employs an optimization criterion based on the entropy measure of information.
