On the Maximum of Conditional Entropy for Upper/Lower Probabilities Generated by Random Sets
- 365 Downloads
Imprecision on probabilities is expressed through upper/lower probability intervals. The lower probability is always assumed to be a convex capacity and sometimes to be an m—monotone capacity. A justification of this latter assumption based on the existence of underlying random sets is given. Standard uncertainty measures of information theory, such as the Shannon entropy and other indices consistent with the Lorenz ordering, are extended to imprecise probability situations by taking their worst-case evaluation, which is shown to be achieved for a common probability. In the same spirit, the information brought by a question is evaluated by the maximum value of the associated conditional entropy. Computational aspects, which involve the resolution of decomposable convex programs, are discussed.
Key wordsAmbiguity Capacities Conditioning Entropy Random Sets Uncertainty Upper/Lower Probabilities
Unable to display preview. Download preview PDF.
- G. Hardy, J. littlewood, and G. PolyaInequalitiesCambridge Univ. Press, 1934.Google Scholar
- J.Y. JaffrayDynamic decision making and belief functionsAdvances in the Dempster-Shafer Theory of Evidence, (R. Yager, M. Fedrizzi, and J. Kacprzyk, eds.), Wiley, 1994, pp. 331–351.Google Scholar
- O.L. MangasarianNonlinear ProgrammingMc Graw-Hill, 1969.Google Scholar
- M. MinouxProgrammation MathématiqueDunod, 1983.Google Scholar
- H.T. Nguyen and E.A. WalkerOn Decision-making using belief functionsAdvances in the Dempster-Shafer Theory of Evidence, (R. Yager, R. Fedrizzi, and J. Kacprzyk, eds.), Wiley, New-York, 1994, pp. 331–330.Google Scholar
- C. PicardGraphes et QuestionnairesGauthier-Villars, 1972.Google Scholar
- G. ShaferA Mathematical Theory of EvidencePrinceton University Press, Princeton, New Jersey, 1976.Google Scholar
- H.M. ThomaBelief functions computationsConditional logic in expert systems, (I.R. Goodman, M. Gupta, H.T. Nguyen, and G.S. Rogers, eds.), North Holland, New York, 1991, pp. 269–308.Google Scholar