In this paper, a new algorithm for blind inversion of Wiener
systems is presented. The algorithm is based on minimization
of mutual information of the output samples. This minimization
is done through a Minimization-Projection (MP)
approach, using a nonparametric “gradient” of mutual information.