Identification of automata Markov models using a modified “forward-backward” algorithm
Alsu Rafailovna Nurutdinova · Journal of Physics Conference Series · 2019
In 1989, Rabiner L.R proposed the solution of the problem of identifying hidden Markov models by induction using the modified "forward-backward" algorithm. In this paper, we propose the application of a modification of this method for solving the recognition of automaton Markov models following sequences of finite simple homogeneous Markov chains generated on the basis of stochastic ergodic matrices, which in turn determine the class of the automata Markov model.