Title :
Parameter estimation with missing data via equalization-maximization
Author :
Stoica, Petre ; Xu, Luzhou ; Li, Jian
Author_Institution :
Dept. of Inf. Technol., Uppsala Univ., Sweden
Abstract :
The expectation-maximization (EM) algorithm is often used in maximum likelihood (ML) estimation problems with missing data. However, EM can be rather slow to converge. In this paper, we introduce a new algorithm for parameter estimation problems with missing data, which we call equalization-maximization (EqM) (for reasons to be explained later). We derive the EqM algorithm in a general context and illustrate its use in the specific case of a Gaussian autoregressive time series with a varying amount of missing observations. In the presented examples, EqM outperforms EM in terms of computational speed, at a comparable estimation performance.
Keywords :
Gaussian distribution; autoregressive processes; maximum likelihood estimation; optimisation; time series; EqM; Gaussian autoregressive time series; equalization-maximization; maximum likelihood estimation; missing data; missing observations; parameter estimation; Convergence; Councils; Information technology; Iterative algorithms; Maximum likelihood estimation; Parameter estimation; Probability density function; Virtual reality;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 2005. Proceedings. (ICASSP '05). IEEE International Conference on
Print_ISBN :
0-7803-8874-7
DOI :
10.1109/ICASSP.2005.1415944