Upper and lower probabilities induced by a fuzzy random variable (Q533130)

This is an interesting paper on fuzzy random variables (frv's), their theoretical background and their interpretation. Two different approaches on frv's can be found in the literature: the approach by Puri and Ralescu where a frv is viewed as a linguistic variable, and the approach by Kruse and Meyer, where a frv is viewed as an incompletely known standard random variable. The authors show that Puri-Ralescu leads to a classical probability model, and that Kruse-Meyer leads to a second order imprecise probability model. They introduce a third kind interpretation of frv's as families of conditional possibility measures. Futheron, they describe the upper-lower probability model associated to it, and they check that the upper and lower probabilities are infinity-order capacities. Multivalued mappings associated to these capacities are discussed. Moreover, the authors show relationships between upper and lower models, between classical and second order imprecise models. These results can be applied in future works to make inference from fuzzy sample data.