Title :
Flexible parametrization of postnonlinear mixtures model in blind sources separation
Author_Institution :
Lab. de Modelization et Calcul, CNRS, Grenoble, France
fDate :
6/1/2004 12:00:00 AM
Abstract :
This letter introduces a flexible parametrization for the nonlinear compensator in the postnonlinear mixture model. The idea is to parametrize not the compensator but a set of quantiles of the distribution of its output. They are specified by the density-quantiles which are parametrized through a spline basis. Good results can be achieved with only a few splines.
Keywords :
blind source separation; parameter estimation; splines (mathematics); blind source separation; flexible density quantiles parametrization; nonlinear compensator; postnonlinear mixtures model; splines; Blind source separation; Digital filters; Distribution functions; Gaussian processes; Mutual information; Random variables; Source separation; Spline; Statistical distributions; Blind source separation; density; order statistic; postnonlinear mixture; quantiles; splines;
Journal_Title :
Signal Processing Letters, IEEE
DOI :
10.1109/LSP.2004.826662
