# Completely Monotone Outer Approximations of Lower Probabilities on Finite Possibility Spaces

• Erik Quaeghebeur
Conference paper
Part of the Advances in Intelligent and Soft Computing book series (AINSC, volume 100)

## Abstract

Drawing inferences from general lower probabilities on finite possibility spaces usually involves solving linear programming problems. For some applications this may be too computationally demanding. Some special classes of lower probabilities allow for using computationally less demanding techniques. One such class is formed by the completely monotone lower probabilities, for which inferences can be drawn efficiently once their Möbius transform has been calculated. One option is therefore to draw approximate inferences by using a completely monotone approximation to a general lower probability; this must be an outer approximation to avoid drawing inferences that are not implied by the approximated lower probability. In this paper, we discuss existing and new algorithms for performing this approximation, discuss their relative strengths and weaknesses, and illustrate how each one works and performs.

## Keywords

lower probabilities Outer approximation Complete monotonicity Belief functions Möbius transform

## References

1. 1.
Baroni, P., Vicig, P.: An uncertainty interchange format with imprecise probabilities. Internat. J. Approx. Reason 40, 147–180 (2005), doi:10.1016/j.ijar.2005.03.001
2. 2.
de Campos, L.M., Huete, J.F., Moral, S.: Probability intervals: A tool for uncertain reasoning. Internat. J. Uncertain., Fuzziness Knowledge-Based Systems 2, 167–196 (1994), doi:10.1142/S0218488594000146
3. 3.
Chateauneuf, A., Jaffray, J.Y.: Some characterizations of lower probabilities and other monotone capacities through the use of Möbius inversion. Math. Social Sci. 17, 263–283 (1989), doi:10.1016/0165-4896(89)90056-5
4. 4.
Dempster, A.P.: Upper and lower probabilities induced by a multivalued mapping. Ann. Math. Stat. 38, 325–339 (1967), doi:10.1214/aoms/1177698950
5. 5.
Hall, J.W., Lawry, J.: Generation, combination and extension of random set approximations to coherent lower and upper probabilities. Reliab. Eng. Syst. Safety 85, 89–101 (2004), doi:10.1016/j.ress.2004.03.005
6. 6.
Quaeghebeur, E.: Learning from samples using coherent lower previsions. PhD thesis, Ghent University (2009), http://hdl.handle.net/1854/LU-495650
7. 7.
Quaeghebeur, E.: Characterizing the set of coherent lower previsions with a finite number of constraints or vertices. In: Spirtes, P., Grünwald, P. (eds.) Proceedings of UAI 2010, pp. 466–473. AUAI Press (2010), http://hdl.handle.net/1854/LU-984156
8. 8.
Quaeghebeur, E., De Cooman, G.: Extreme lower probabilities. Fuzzy Sets and Systems 159, 2163–2175 (2008), http://hdl.handle.net/1854/11713, doi:10.1016/j.fss.2007.11.020
9. 9.
Shafer, G.: A mathematical theory of evidence. Princeton University Press, Princeton (1976)
10. 10.
Troffaes, M., Quaeghebeur, E.: improb: A python module for working with imprecise probabilities (2011), https://github.com/equaeghe/improb, fork of https://github.com/mcmtroffaes/improb. Latest public release at http://packages.python.org/improb/
11. 11.
Walley, P.: Statistical Reasoning with Imprecise Probabilities. Chapman & Hall, Boca Raton (1991)