Training MLPs layer-by-layer with the information potential.

Dongxin Xu, Jos C. Prncipe

Browse the full IJCNN paper archive.