An Experimental Comparison of Numerical and Qualitative Probabilistic Reasoning
Abstract
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 - 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.
Cite
Text
Henrion et al. "An Experimental Comparison of Numerical and Qualitative Probabilistic Reasoning." Conference on Uncertainty in Artificial Intelligence, 1994. doi:10.1016/B978-1-55860-332-5.50045-6Markdown
[Henrion et al. "An Experimental Comparison of Numerical and Qualitative Probabilistic Reasoning." Conference on Uncertainty in Artificial Intelligence, 1994.](https://mlanthology.org/uai/1994/henrion1994uai-experimental/) doi:10.1016/B978-1-55860-332-5.50045-6BibTeX
@inproceedings{henrion1994uai-experimental,
title = {{An Experimental Comparison of Numerical and Qualitative Probabilistic Reasoning}},
author = {Henrion, Max and Provan, Gregory M. and Del Favero, Brendan and Sanders, Gillian},
booktitle = {Conference on Uncertainty in Artificial Intelligence},
year = {1994},
pages = {319-326},
doi = {10.1016/B978-1-55860-332-5.50045-6},
url = {https://mlanthology.org/uai/1994/henrion1994uai-experimental/}
}