When using deep neural networks for classification, in order to have the ability to generalize we can define a robust architecture that overfits

data and then use techniques that provide a regularization effect, such as Dropout, Early stopping, Batch Regularization, Max$-$Norm regular

True

False