What is Dropout?
It is a way to perform Regularization.
Dropout is a method of randomly dropping nodes with a certain probability so that they do not participate in learning. Since nodes are omitted in the feedforward process, they are naturally excluded from backpropagation as well. Let’s see the structure of Dropout.