BSS algorithm by diffusing nonparameteric density estimator

Nonparametric diffusion mixing estimator (DME) based blind signal separation (BSS) algorithm is proposed under the framework of natural gradient optimization method. In order to improve the performance of signal separation by BSS, the probability distributions of source signals must be described as accurately as possible. In this paper, we use the new data-driven bandwidth selection method based MDE to estimate the probability distributions of sources, which can improve the performance of fixed-width kernel density estimator (FKDE). The MDE is inspired via a Langevin diffusion process. As a result, the proposed algorithm has a wider application and do not need to assume the parametric nonlinear functions as them. The effectiveness of the proposed algorithm has been confirmed by simulation experiments.