THE STRUCTURE OF A NEURO-FUZZY NETWORK BASED ON BAYESIAN LOGICAL-PROBABILISTIC MODEL

UDC 519.226.3; 510.644.4; 510.647; 004.032.26
DOI: 10.36871/2618-9976.2022.12.004

Authors

Gulnara I. Kozhomberdieva
Candidate of Technical Sciences, Associate Professor, Emperor Alexander I St. Petersburg State Transport University, Saint Petersburg, Russia
Dmitry P. Burakov
Candidate of Technical Sciences, Associate Professor, Emperor Alexander I St. Petersburg State Transport University, Saint Petersburg, Russia
Georgii A. Khamchichev
Postgraduate Student, Emperor Alexander I St. Petersburg State Transport University, Saint Petersburg, Russia

Abstract

The article presents a multilayer structure of a neurofuzzy network based on the Bayesian logicalprobabilistic model of fuzzy inference, previously proposed, researched and implemented by the authors. A brief description of the Bayesian logicalprobabilistic model is given, an example of setting up a neurofuzzy network for solving a fuzzy inference problem is presented. The example shows which network parameters can be used for its training. According to the authors, the proposed network structure with three parametric layers is comparable to the wellknown Takagi– Sugeno–Kang and Wang–Mendel fuzzy neural networks.

Keywords

Artificial neural network Multilayer neural network, Fuzzy neural network, Neurofuzzy network, Fuzzy inference, Fuzzy logic, Bayesian logicalprobabilistic model of fuzzy inference, Bayesian approach, Probabilistic logic, Bayes’ theorem