TY - JOUR
T1 - Blind source separation using block-coordinate relative Newton method
AU - Bronstein, Alexander M.
AU - Bronstein, Michael M.
AU - Zibulevsky, Michael
N1 - Funding Information:
The authors would like to acknowledge support for this project by the Ollendorff Minerva Center and the HASSIP Research Network Program HPRN-CT-2002-00285, sponsored by the European Commission.
PY - 2004/8
Y1 - 2004/8
N2 - Presented here is a block-coordinate version of the relative Newton method, recently proposed for quasi-maximum likelihood blind source separation. Special structure of the Hessian matrix allows performing block-coordinate Newton descent efficiently. Simulations show that typically our method converges in near constant number of iterations (order of 10) independently of the problem size.
AB - Presented here is a block-coordinate version of the relative Newton method, recently proposed for quasi-maximum likelihood blind source separation. Special structure of the Hessian matrix allows performing block-coordinate Newton descent efficiently. Simulations show that typically our method converges in near constant number of iterations (order of 10) independently of the problem size.
KW - Blind source separation
KW - Block-coordinate optimization
KW - Newton algorithm
KW - Quasi-maximum likelihood
UR - http://www.scopus.com/inward/record.url?scp=3042852446&partnerID=8YFLogxK
U2 - 10.1016/j.sigpro.2004.05.019
DO - 10.1016/j.sigpro.2004.05.019
M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???
AN - SCOPUS:3042852446
SN - 0165-1684
VL - 84
SP - 1447
EP - 1459
JO - Signal Processing
JF - Signal Processing
IS - 8
ER -