Compiegne ISIPTA'11 home
Latest info: Pictures now online.

Kevin Loquin


Significance of a decision making problem under uncertainty

Abstract

In this paper, we work on the interval dominance based extension of the Savage EUM approach. While usual probabilities only handle variability due uncertainty, imprecise probabilities handle in a unique framework uncertainty due to variability and uncertainty due to lack of knowledge (epistemic uncertainty). This second side of uncertainty generates ambiguity in the decision process which does not appear in the usual probabilistic Savage approach. Ambiguity here means that the preference relation used to perform the decision is not complete, i.e. that some decisions are not comparable. This comparability (or its opposite ambiguity) is linked to the informativity of the assessed imprecise probabilities. Our proposal, in this paper, is that significance is the informativity degree of the imprecise probability model used in the imprecise SEUM approach which makes the problem change from ambiguous to comparable. We discuss a theoretical definition as well as a pragmatical one.

Keywords

Savage EUM, decision theory, imprecise probability, interval dominance, significance


Download area

The paper is available in the following formats:

E-mail addresses

Kevin Loquin   kevin.loquin@gmail.com

Send any remarks to isipta13@hds.utc.fr.