On set-valued fuzzy measures. (Q1429853): Difference between revisions

Set-valued fuzzy measures (i.e., monotone real-subset-values set functions \(\pi\) assigning \(\{0\}\) to \(\pi(\emptyset)\)) are genuine generalization and extension of the concept of fuzzy measures (i.e., monotone real-valued set functions vanishing at the empty set). For example, to any system \((\mu_i)_{i\in I}\) of fuzzy measures on the same space a set-valued fuzzy measure \(\pi\) given by \[ \pi(A)= [\inf\mu_i(A),\sup\mu_i(A)] \] can be assigned (compare the concept of lower and upper probabilities, see also [\textit{B. De Finetti}, ``Theory of probability. A critical introductory treatment'', Vol. I (1974; Zbl 0328.60002)]). On the other hand, set-valued fuzzy measures extends the Artstein concept of set-valued measures [\textit{Z. Artstein}, Trans. Am. Math. Soc. 165, 103--125 (1972; Zbl 0237.28008)]. The paper deals mainly with set-valued fuzzy measures defined by set-valued fuzzy integrals (generalizations of the Aumann integral). Moreover, set-valued pseudo-additive measures are introduced and discussed. Though the main aim of the paper is the introduction of the set-valued fuzzy measures; it brings also some variants of classical results from the measure theory. For example, a version of the Radon-Nikodým theorem is given.