An Experimental Comparison of Numerical and Qualitative Probabilistic Reasoning
Max Henrion, Gregory Provan, Brendan del Favero, Gillian Sanders
Qualitative and infinitesimal probability schemes are consistent with the axioms of probability theory, but avoid the need for precise numerical probabilities. Using qualitative probabilities could substantially reduce the effort for knowledge engineering and improve the robustness of results. We examine experimentally how well infinitesimal probabilities (the kappa-calculus of Goldszmidt and Pearl) perform a diagnostic task Ñ troubleshooting a car that will not start by comparison with a conventional numerical belief network. We found the infinitesimal scheme to be as good as the numerical scheme in identifying the true fault. The performance of the infinitesimal scheme worsens significantly for prior fault probabilities greater than 0.03. These results suggest that infinitesimal probability methods may be of substantial practical value for machine diagnosis with small prior fault probabilities.
Keywords: Bayesian probabilities, kappa probabilities, diagnosis. networks, qualitative calculu
PDF Link: /papers/94/p319-henrion.pdf
AUTHOR = "Max Henrion
and Gregory Provan and Brendan del Favero and Gillian Sanders",
TITLE = "An Experimental Comparison of Numerical and Qualitative Probabilistic Reasoning",
BOOKTITLE = "Proceedings of the Tenth Conference Annual Conference on Uncertainty in Artificial Intelligence (UAI-94)",
PUBLISHER = "Morgan Kaufmann",
ADDRESS = "San Francisco, CA",
YEAR = "1994",
PAGES = "319--326"