Conference paper
The EM Algorithm in Independent Component Analysis
We investigate two techniques for independent component analysis which use the expectation-maximization algorithm. Analysis and simulations show that convergence becomes extraordinarily slow for almost all cases, compared to other optimization techniques. The two alternatives considered are "adaptive overrelaxed EM" and Ucminf (a BFGS with soft line search), which both improves the convergence dramatically with little or no extra analytical work.
We discuss the generality and perspectives of the findings.
Language: | English |
---|---|
Year: | 2005 |
Pages: | v/169,v/170,v/171,v/172 |
Proceedings: | 2005 IEEE International Conference on Acoustics, Speech, and Signal Processing |
ISBN: | 0780388747 and 9780780388741 |
ISSN: | 2379190x and 15206149 |
Types: | Conference paper |
DOI: | 10.1109/ICASSP.2005.1416267 |
ORCIDs: | Winther, Ole |
Acceleration Algorithm design and analysis Analytical models Convergence Cost function Covariance matrix Expectation-maximization algorithms Independent component analysis Maximum likelihood estimation Proposals adaptive overrelaxed EM convergence of numerical methods expectation-maximization algorithm independent component analysis optimisation optimization random noise signal processing soft line search