Fuzzy Backpropagation Training of Neural Networks

Adaption of parameters during neural network training according to the actual shape of the error surface is supposed to be a powerful instrument to enforce convergence and to decrease time consumption of neural network training.