oalogo2  

AUTHOR(S): 

I. Gorynin, E. Azeraf, W. Sabbagh, E. Monfrini, W. Pieczynski

 

TITLE

Optimal Filtering in Hidden and Pairwise Gaussian Markov Systems

pdf PDF

ABSTRACT

In a hidden Markov model (HMM), the system goes through a hidden Markovian sequence of states (X) and produces a sequence of emissions (Y). We define the hidden Gaussian Markov model (HGMM) as the HMM where the hidden process is Gaussian and is affected by a normal white noise. The Kalman filter (KF) is a fast optimal statistical estimation method for the HGMMs and is very popular among the practitioners. However, the classic HGMM formulation is too restrictive. It extends to recent pairwise Gaussian Markov model (PGMM) where we assume that the pair (X, Y) is Gaussian Markovian. Moreover, there exists a KF version for the PGMM. The PGMM is more general than HGMM and in particular, the PGMM hidden process is not necessarily Markovian. The authors share their findings on about enhancing the KF when improving HGMM to PGMM. We discover singular cases where HGMM is at least ten times less accurate than the PGMM. On average, PGMM outperforms HGMM by twenty percent.

KEYWORDS

Bayesian estimation, Hidden Markov models, Pairwise Markov models, Kalman filter

REFERENCES

[1] O. Cappé, E. Moulines, and T. Ryden, Inference in hidden Markov models, Springer, 2005. [1] O. Cappé, E. Moulines, and T. Ryden, Inference in hidden Markov models, Springer, 2005. 

[2] M. S. Grewal and A. P. Andrews, Kalman Filtering: Theory and Practice Using MATLAB, Wiley, 2008. 

[3] V. Nemesin and S. Derrode, Robust blind pairwise Kalman algorithms using QR decompositions, IEEE Trans. on Signal Processing, Vol. 61, No. 1, 2013, pp. 5-9. 

[4] V. Nemesin and S. Derrode, Robust partial learning in linear Gaussian systems, IEEE Trans. on Automatic Control, Vol. 60, No. 9, 2015, pp. 2518- 2523. 

[5] A. Genz, Methods for generating random orthogonal matrices. Monte Carlo and Quasi-Monte Carlo Methods. 1998, pp. 199—213. 

[6] W. Pieczynski and F. Desbouvries, On triplet Markov chains, International Symposium on Applied Stochastic Models and Data Analysis, Brest, France, May 2005. 

[7] M. V. Kulikova, Gradient-based Parameter Estimation in Pairwise Linear Gaussian System, IEEE Transactions on Automatic Control, accepted, 2016. 

[8] R. S. Liptser and A. N. Shiryaev, Statistics of Random Processes: I. General theory, Springer, 2001. 

[9] P. Lanchantin, J. Lapuyade-Lahorgue, and W. Pieczynski, Unsupervised segmentation of randomly switching data hidden with non-Gaussian correlated noise, Signal Processing, Vol. 91, No. 2, 2011, pp. 163-175.

Cite this paper

I. Gorynin, E. Azeraf, W. Sabbagh, E. Monfrini, W. Pieczynski. (2016) Optimal Filtering in Hidden and Pairwise Gaussian Markov Systems. Mathematical and Computational Methods, 1, 259-263

 

cc.png
Copyright © 2016 Author(s) retain the copyright of this article.
This article is published under the terms of the Creative Commons Attribution License 4.0