Orthogonally-Constrained Extraction of Independent Non-Gaussian Component from Non-Gaussian Background Without ICA

0492879 - ÚTIA 2019 RIV GB eng C - Conference Paper (international conference)
Koldovský, Z. - Tichavský, Petr - Ono, N.
Orthogonally-Constrained Extraction of Independent Non-Gaussian Component from Non-Gaussian Background Without ICA.
Latent Variable Analysis and Signal Separation. Cham: Springer, 2018 - (Deville, Y.; Gannot, S.; Mason, R.; Plumbley, M.; Ward, D.), s. 161-170. Lecture Notes in Computer Science, 10891. ISBN 978-3-319-93763-2. ISSN 0302-9743. E-ISSN 1611-3349.
[Latent Variable Analysis and Signal Separation. Guilford (GB), 02.07.2018-05.07.2018]
R&D Projects: GA ČR GA17-00902S
Institutional support: RVO:67985556
Keywords : Independent Component Analysis * Blind source separation * blind source extraction
OECD category: Statistics and probability
http://library.utia.cas.cz/separaty/2018/SI/tichavsky-0492879.pdf

We propose a new algorithm for Independent Component Extraction that extracts one non-Gaussian component and is capable to exploit the non-Gaussianity of background signals without decomposing them into independent components. The algorithm is suitable for situations when the signal to be extracted is determined through initialization, it shows an extra stable convergence when the target component is dominant. In simulations, the proposed method is compared with Natural Gradient and One-unit FastICA, and it yields improved results in terms of the Signal-to-Interference ratio and the number of successful extractions.
Permanent Link: http://hdl.handle.net/11104/0286552