Uncertainty in Artificial Intelligence
First Name   Last Name   Password   Forgot Password   Log in!
    Proceedings   Proceeding details   Article details         Authors         Search    
A Unified Approach to Imprecision and Sensitivity of Beliefs in Expert Systems
David Spiegelhalter
Abstract:
Point probabilities for propositions in expert systems are often felt to be an inadequate expression of belief, since they not only may depend on imprecise assessments, but may also be very sensitive to further relevant information on the case in hand. We show that when a directed graph in used for representing dependent quantities, these sources of 'softness' are intimately related, and the total local 'variance' in a current belief can be decomposed into a number of components. This decomposition may be used for both explanation and planning purposes.
Keywords: Expert Systems, Sensitivity of Beliefs
Pages: 199-208
PS Link:
PDF Link: /papers/87/p199-spiegelhalter.pdf
BibTex:
@INPROCEEDINGS{Spiegelhalter87,
AUTHOR = "David Spiegelhalter ",
TITLE = "A Unified Approach to Imprecision and Sensitivity of Beliefs in Expert Systems",
BOOKTITLE = "Uncertainty in Artificial Intelligence 3 Annual Conference on Uncertainty in Artificial Intelligence (UAI-87)",
PUBLISHER = "Elsevier Science",
ADDRESS = "Amsterdam, NL",
YEAR = "1987",
PAGES = "199--208"
}


hosted by DSL   •   site info   •   help