Most of the neural network aims are to train the
network in order to achieve a balance between the ability to respond correctly the
input patterns that are used for training (memorization) and the ability of giving
reasonable but not identical to that used in training (generalization).