Nonlinear PCA algorithm for blind separation of convolutive mixtures
Xian‐Da Zhang · Journal of Tsinghua University(Science and Technology) · 2006
A nonlinear principal component analysis(PCA) method was developed to speed up the slow convergence rate of convolutive blind source separation methods.The method is equivalent to higher-order statistical methods for some conditions.The algorithm is a recursive least-squares method which has better convergence and tracking ability than the natural gradient method and the higher-order statistical method.Simulation results illustrate the algorithm efficiency.