Independent component analysis through direct estimation of the mutual information

0410338 20240103182207.120060210235959.9 951-22-5017-9 Independent component analysis through direct estimation of the mutual information Helsinki IEEE Signal Processing Society 2000 6 s. Independent Component Analysis and Blind Signal Separation. Proceedings69-74PajunenP.KarhunenJ. cav_un_auth*0212645 Darbellay G. A. CH cav_un_auth*0101212 Tichavský Petr UTIA-B Department of Stochastic Informatics Ústav teorie informace a automatizace AV ČR, v. v. i. 12B KSK1075601 GA AV CZ cav_un_auth*0027435 AV0Z1075907 An algorithm for the estimation of the mutual information between several signals is studied. The algorithm uses a data-dependent partitioning of the observation space and has a relatively low computational cost. The usefulness of this approach in solving the independent component analysis problem is demonstrated on the separation of a linear mixture of two and three real-world speech data sets. cav_un_auth*0212646 ICA 2000. /2./ Helsinki FI 19.06.2000-22.06.2000 BB SI http://hdl.handle.net/11104/0130428 UTIA-B 20000054 2000 2000 Helsinki IEEE Signal Processing Society 951-22-5017-9 Independent Component Analysis and Blind Signal Separation. Proceedings 69 74 Pajunen P. 340 Karhunen J. 340