Modeling and Parameter Estimation of a Class of General Hidden Markov Model
Received date: 2004-05-13
Revised date: 2004-07-09
Online published: 2005-03-15
胡可 , 张大力 . 一类广义隐马尔科夫模型的建模与参数估计(英文)[J]. 中国科学院大学学报, 2005 , 22(2) : 210 -217 . DOI: 10.7523/j.issn.2095-6134.2005.2.014
It is wel-l known that HMM has been widely used in many fields. In this paper we will discuss a more general model, which is similar to Pairwise Markov Model ( PMM) proposed by Wojciech Pieczynski. Compared to HMM, the state process here is not necessarily aMarkov chain. So it has more general applications in image segmentation, speech signal processing, and etc. We will give a complete mathemat ical description for this model with discrete states and discrete observations, including modeling,state estimation and parameter estimat ion, which haven. t been studied before. Based on the method proposed here, we will get a recursive algorithm for the estimation of the state and the parameters.
[1] L R Rabiner. A tut orial on Hidden Markov Models and select ed appl ications in speech recognit ion. Proceedings of the IEEE, 1989, 77( 2) : 257~286
[2] Wojciech Pieczynski. Pairwise Markov chains. IEEE Transanctions on Pattern Analysis and Machine Intelligence, 2003, 25(5) : 634~ 639
[3] Robert J Ell iott, L Aggoun, J B Moore. Hidden Markov Model est imation and cont rol.New York: Springer-Verlag, 1995
[4] Francois Desbouvries, Wojciech Picczynski. Parti cle f ilt ering with pairwise Markov processes, IEEE. 2003
/
| 〈 |
|
〉 |
