Employing adaptive functions and maximum entropy principle for nonlinear blind source deconvolution

In this paper we present the results of applying adaptive nonlinearities and maximum entropy principle to identify an inverting filter for the post nonlinear blind source deconvolution problem. The filter is a cascade of a linear FIR matrix and a nonlinear memoryless componentwise system. Cubic splines and polynomials have been selected as adaptive parametric functions. A fast frequency implementation is also proposed.