A CLASS OF APPROACHES FOR BLIND SOURCE SEPARATION BASED ON MULTIVARIATE DENSITY ESTIMATION
一类基于多变量密度估计的盲源分离方法

A class of learning algorithms is drived for blind separation of independent source signals in this paper. These algorithms are based on minimizing a contrast function defined in terms of the Kullback-Leibler distance. By utilizing the technique of multivariate density esti-mation, two types of separating algorithms are obtained. Simulations illustrate the effectiveness of the algorithms.