UDC 519.21
USING HIDDEN MARKOV MODELS IN ESTIMATING
THE PARAMETERS OF HIERARCHICAL SYSTEMS
Abstract. The method of parametric estimation for hierarchical stochastic models under incomplete observations is considered.
The method is based on the features of the correlation structure of hierarchical models. The main attention is paid to the practical implementation
of the method. In particular, an approach is proposed that combines analytical studies and empirical verification of the solutions.
Specific examples of constructing consistent estimates of the vector parameters of the deformation function are provided and illustrated
by direct calculations with numerical data of the simulation model.
Keywords: hidden Markov model, queueing system, statistical estimation, the deformation function.
FULL TEXT
REFERENCES
- Voina A.A. Statistical estimation in the hidden Markov model of the hierarchical structure. Kibernetika i sistemnyj analiz. 2014. Vol. 50, N 6. P. 87–104.
- Gikhman I.I., Skorokhod A.V. Introduction to the theory of stochastic processes [in Russian]. Moscow: Nauka, 1977. 568 p.
- Ivchenko G.I., Kashtanov V.A., Kovalenko I.N. Queuing theory [in Russian]. Moscow: Vysshaya shkola, 1982. 256 p.
- Rabiner L.R., Juang B.H. An introduction to hidden Markov models. IEEE ASSP Mag. 1986. Vol. 3, N 1. P. 4–16.
- Voina O.A., Czapla E. An application of the correlation structure of a Markov chain for the estimation of shift parameters in queuing systems. Theor. Prob. and Math. Statist. 2005. N 71. P. 53–61.
- Voina A.A. Asymptotic optimization for stochastic models based on a complex Poisson process. Kibernetika i sistemnyj analiz. 2011. Vol. 47, N 4. P. 165–175.
- Voina O.A. Nonparametric estimation for a compound Poisson process governed by a Markov chain. Theor. Probability and Math. Statist. 2012. № 85. P. 41–52.