Lapko O. Mixed methods for calculating of uncertainty and their application in artificial intelligence systems

The dissertation focuses on different ways of the uncertainty calculating in fuzzy decision systems. In the paper presents a new approach to the calculation of uncertainties, that can be applied to problems of optimal evaluation, decision making , analysis and interpretation of the experiment etc. Also were viewed the existing ways of describing the uncertain quantities such as probability, through the random variable, and were presented new ways ofpossibility, fuzzy variable and mixed through fuzzy random variables. Also discussed various ways of defining conditional of uncertain events, probabilityandsome possibilities, ie approaches: Zadeh, Hisdal, Dubois, Prada, Dempster and Koeman. For considerate calculation methods proposed methods to assess uncertainty calculation results, that allow to compare the efficiency of probability and possibilities approaches to the uncertainty calculation. Special attention is given to research opportunity of using fuzzy specifications for diagnostic problem solving in a variety of formulations and probability - possibilistic approaches to estimates obtain of the fuzzy inference system. In addition, the new examples are given to investigate their capabilities on existing and proposed methods of calculating uncertainties.