An Information-Maximization Approach to Blind Separation and Blind Deconvolution
We derive a new self-organizing learning algorithm that maximizes the information transferred in a network of nonlinear units. The algorithm does not assume any knowledge of the...