Perfect learning in neural networks: on sample complexity and scaling issues of gradient descent trained multilayer perceptrons and a novel object-oriented simulation framework for scalable information processing modules.
R. Lange. University of Heidelberg, Germany, (1995)
