Conditional Independence in Possibility Theory
Artificial Intelligence
2013-02-28 v1
Abstract
Possibilistic conditional independence is investigated: we propose a definition of this notion similar to the one used in probability theory. The links between independence and non-interactivity are investigated, and properties of these relations are given. The influence of the conjunction used to define a conditional measure of possibility is also highlighted: we examine three types of conjunctions: Lukasiewicz - like T-norms, product-like T-norms and the minimum operator.
Keywords
Cite
@article{arxiv.1302.6806,
title = {Conditional Independence in Possibility Theory},
author = {Pascale Fonck},
journal= {arXiv preprint arXiv:1302.6806},
year = {2013}
}
Comments
Appears in Proceedings of the Tenth Conference on Uncertainty in Artificial Intelligence (UAI1994)