Blind source recovery using an adaptive generalized Gaussian score function

The paper discusses blind source recovery (BSR) of minimum phase and non-minimum phase mixtures of multiple source distributions using an adaptive score function. This proposed parametric score function is derived from the generalized Gaussian distribution model. An adaptive algorithm to determine the tuning parameter for the proposed score function using batch kurtosis of the BSR output is also presented. The primary advantage of the proposed framework is that it renders the adaptive estimation of the demixing network to be completely blind. No a priori information about the distribution structure of the original sources is required. Simulation examples verifying the proposed framework are also presented.