Efficient Variant of Algorithm FastICA for Independent Component Analysis Attaining the Cramér-Rao Lower Bound

0411390 - UTIA-B 20050120 CZ eng I - Internal Report
Koldovský, Zbyněk - Tichavský, Petr - Oja, E.
Efficient Variant of Algorithm FastICA for Independent Component Analysis Attaining the Cramér-Rao Lower Bound.
Praha: ÚTIA AV ČR, 2005. 27 s. 2005/10.
R&D Projects: GA MŠMT 1M0572
Institutional research plan: CEZ:AV0Z10750506
Keywords : independent component analysis * blind source separation * blind deconvolution
Subject RIV: BB - Applied Statistics, Operational Research

FastICA is one of the most popular algorithms for Independent Component Analysis, demixing a set of statistically independent sources that have been mixed linearly. A key question is how accurate the method is for finite data samples. We propose an improved version of the FastICA algorithm which is asymptotically efficient. i.e. its accuracy given by the residual error variance attains the Cramér-Rao lower bound. The error is thus as small as possible.

V práci je navržena nová varianta algoritmu FastICA, která je asymptoticky eficientní, tj. její přesnost se blíží Rao-Cramerově hranici, za předpokladu že pravděpodobnostní rozložení separovaných signálů je z třídy zobecněných Gaussovských distribucí. V simulacích je navržená metoda porovnávána se známým algoritmem JADE a s neparametrickou ICA.
Permanent Link: http://hdl.handle.net/11104/0131472