Knowledge representation and acquision

Comparative analysis of the inference methods based on the fuzzy truth value for the MISO-structure systems

V. G. Sinuk, S. V. Kulabukhov

Belgorod Shukhov State Technological University, Belgorod, Russia

Abstract: Fuzzy inference method based on fuzzy truth value enables us to perform fuzzy inference with polynomial computational complexity in case of multiple fuzzy inputs. This method also allows us to compare fuzzy implications against approximate reasoning of human intelligence. In this paper, several criteria for such comparison are proposed. First, the abovementioned fuzzy inference method is considered, the article provides its formal definition. Then the comparison criteria are defined in an appropriate form. The paper illustrates the principle of how different fuzzy implications can be tested against these criteria. As an example, Mamdani’s model with different t-norms has been examined. Finally, logical model with Godel and Rescher has been considered. In each case, fulfilled criteria are enumerated.

Keywords: fuzzy truth value, performance criteria, fuzzy inference methods.

DOI: 10.14357/20718594200208

 English version: , 2021, 48:6, 362–365

Bibliographic databases:

•