Fuzzy neural network for Markovian arrival processes’ intensity classification

Abstract: The paper studies the applications of fuzzy neural networks that use membership functions as activation functions to solve the problems of classification and estimation of Markov arrival processes’ intensity. Computer simulation tools in MATLAB are used for this research. Markov arrival processes are represented by two successive simple (Poisson) processes, each of which is characterized by the intensity of event arrival. Using neural networks, we can obtain the solutions to dichotomy problems, including the classification and estimation of the intensity of the two Poisson processes. Computer modeling has confirmed the effectiveness of the classification problem for the intensity of Markovian arrival processes based on neural networks.

Keywords: fuzzy activation functions, fuzzy neural net- works, markovian arrival processes, intensity classification