Stochastic identification of the "object-attribute" table based on the modified Rabiner's method

Abstract

This article is about solving the problem of stochastic identification of the "Object-attribute"table based on subsets of stochastic ergodic matrices. The table has N rows and m columns. The identification is based on the implementation of the modified Rabiner's method. We assume that the elements of m columns of the table are a discrete Markov chain of length N. The identification of each column is based on calculating the maximum probability that the Markov chain is generated based on the distribution law represented by one ergodic stochastic matrix from a given subset. An algorithm for solving this problem is proposed. Estimates of the time and hardware complexity of this algorithm, which are executed in parallel on a distributed computing system, are obtained. The dependence of the obtained estimates on the number of rows and columns of the identified table is determined. The value of N has a linear effect on the time complexity of an algorithm that implements MMR and is executed in parallel. A promising direction for future research is the distributed implementation of the proposed Algorithm.