# Basic Issues of Computing with Granular Probabilities

• George J. Klir
Part of the Studies in Fuzziness and Soft Computing book series (STUDFUZZ, volume 95)

## Abstract

It has increasingly been recognized that practical applicability of classical probability theory to some problems of current interest, particularly those dependent on human judgement, is severely restricted. This restriction is a result of the additivity axiom of classical probability theory. Due to this axiom, elementary events are required to be pairwise disjoint and the probability of each is required to be expressed precisely by a real number in the unit interval [0, 1]. The requirement of disjoint events makes mathematically good sense, but it becomes problematic whenever we leave the world of mathematics. As is well known, there is no method of scientific measurement that is free from error. As a consequence, observations in close neighborhoods of the sharp boundaries between events are unreliable and should be properly discounted. This results in a violation of the additivity axiom and, by implication, in a violation of the precision requirement. Hence, we need to deal with imprecise probabilities. This need is even more pronounced when instead of measurements we are dependent on assessments based on subjective human judgement. There are other convincing arguments for imprecise probabilities, including, for example, the following:
• Imprecision of probabilities is needed to reflect the amount of information on which they are based. The precision should increase with the amount of information available.

• Total ignorance can be properly modeled by vacuous probabilities, which are maximally imprecise (i.e., each covers the whole range [0, 1]), but not by any precise probabilities.

• Imprecise probabilities are generally easier to assess and elicit than precise ones.

• We may be unable to assess probabilities precisely in practice, even if that is possible in principle, because we lack the time and computational ability.

• A precise probability model that is defined on some class of events determines only imprecise probabilities for events outside the class.

• When several sources of information (sensors, experts, individuals in a group decision) are combined, the extent to which they are consistent can be reflected in the precision of the combined model.

## Keywords

Fuzzy Number Bayesian Inference Triangular Membership Function Additivity Axiom Fuzzy Interval
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

## References

1. [1]
Dubois, D. and H. Prade [ 1981 ] “Additions of interactive fuzzy numbers.” IEEE Trans. on Automatic Control, 26 (4), pp. 926–236
2. [2]
Giachetti, R. E. and R. E. Young [ 1997 ], “Analysis of the error in the standard approximation for multiplication of triangular and trapezoidal fuzzy numbers and the development of a new approximation.” Fuzzy Sets and Systems, 91 (1), pp. 1–13.
3. [3]
Giachetti, R.E. and R.E.Young [ 1997 ] “ A parametric representation of fuzzy numbers and their arithmetic operators.” Fuzzy Sets and Systems, 91 (2), pp. 185–202.
4. [4]
Kaufmann, A. and M.M.Gupta [1985], Introduction to Fuzzy Arithmetic. Van Nostrand, New York.Google Scholar
5. [5]
Klir, G. J. [ 1997 ], “The role of fuzzy arithmetic in engineering.” In: Ayyub, B. M., ed., Uncertainty Analysis in Engineering and the Science. Kluwer, Boston.Google Scholar
6. [6]
Klir, G. J. [ 1997 ], “Fuzzy arithmetic with requisite constraints.” Fuzzy Sets and Systems, 91 (2), pp. 165–175.
7. [7]
Klir, G.J. and Y.Pan [ 1998 ], Constrained fuzzy arithmetic: Basic questions and some answers.” Soft Computing, 2 (2), pp. 100–108.
8. [8]
Klir, G.J. and M.J.Wierman [ 1998 ], Uncertainty-Based Information: Elements of Generalized Information Theory. Physica -Verlag/Springer Verlag, Heidelberg and New York.
9. [9]
Klir, G. J. and B. Yuan [ 1995 ], Fuzzy Sets and Fuzzy Logic: Theory and Applications. Prentice Hall, Upper Saddle River, NJ.Google Scholar
10. [10]
Kyburg, H. E. [ 1987 ], “Bayesian and non-Bayesian evidential updating.” Artifical Intelligence, 31, pp. 271–293.
11. [11]
Pan, Y. [ 1997a ], “Revised hierarchical analysis method based on crisp and fuzzy entries and its application to assessment of prior probability distributions.” Inter. J. of General Systems, 26 (1–2), pp. 1110–131.Google Scholar
12. [12]
Pan, Y. [ 1997b ], Calculus of Fuzzy Probabilities and Its Applications. PhD Dissertation in Systems Science, Binghamton University-SUNY.Google Scholar
13. [13]
Pan, Y. and G. J. Klir [1997], “Bayesian inference based on interval-valued prior distributions and likelihoods.” J. of Intelligent & Fuzzy Systems,5(3).Google Scholar
14. [14]
Pan, Y. and B. Yuan [ 1997 ], “Bayesian inference of fuzzy probabilities.” Inter. J. of General Systems, 26 (1–2), pp. 73–90.
15. [15]
Walley, P. [ 1991 ], Statistical Reasoning With Imprecise Probabilities. Chapman and Hall, London.
16. [16]
Wang, Z. and G. J. Klir [ 1992 ], Fuzzy Measure Theory. Plenum Press, New York.
17. [17]
Weichselberger, K. and Pohlmann [ 1990 ], A Methodology for Uncertainty in Knowledge-Based Systems. Springer-Verlag, New York.
18. [18]
Zadeh, L. A. [ 1968 ], “Probability measures of fuzzy events.” J. of Math. Analysis and Applications, 23, pp. 421–427.
19. [19]
Zadeh, L. A. [1975], “The concept of a linguistic variable and its application to approximate reasoning I, II, III.” Information Sciences, 8, pp. 199–251, 301–357; 9, pp. 43–80.Google Scholar
20. [20]
Zadeh, L. A. [ 1996 ], “Fuzzy logic = computing with words.” IEEE Trans. on Fuzzy Systems, 4 (2), pp. 103–111.
21. [21]
Zadeh, L. A. [ 1997 ], “Toward a theory of fuzzy information granulation and its centrality in human reasoning and fuzzy logic.” Fuzzy Sets and Systems, 90 (2), pp. 111–127.