Blind source separation using measure on copulas
The paper introduces a novel BSS algorithm for instantaneous mixtures of both independent and dependent sources. This approach is based on the minimization of Kullback-Leibler divergence between copula densities. This latter takes advantage of copulas to model the dependency structure of the source components. The new algorithm can efficiently achieve good separation standard BSS methods fail. Simulation results are presented showing the convergence and the efficiency of the proposed algorithms.