Title :
A Monte Carlo Algorithm for Optimal Quantization in Hidden Markov Models
Author :
Tadic, V.B. ; Doucet, A.
Author_Institution :
Univ. of Bristol, Bristol
Abstract :
In this paper, the problem of the optimal quantization of a signal generated by a hidden Markov model is considered. For this problem, an efficient algorithm based on Monte Carlo sampling, gradient estimation techniques and stochastic approximation is proposed. The properties of the proposed algorithm are analyzed both theoretically and through simulations.
Keywords :
Monte Carlo methods; gradient methods; hidden Markov models; quantisation (signal); signal sampling; stochastic processes; Monte Carlo algorithm; Monte Carlo sampling; gradient estimation techniques; hidden Markov models; optimal quantization; stochastic approximation; Algorithm design and analysis; Analytical models; Approximation algorithms; Hidden Markov models; Monte Carlo methods; Quantization; Signal generators; Statistics; Stochastic processes; Zinc;
Conference_Titel :
Information Theory, 2007. ISIT 2007. IEEE International Symposium on
Conference_Location :
Nice
Print_ISBN :
978-1-4244-1397-3
DOI :
10.1109/ISIT.2007.4557374
