Dropout Regularization

What is Dropout Regularization Technique?

A technique of regularization which removes nodes and their links in DNN randomly.

Dropout Regularization Features

Dropout reduces overfitting and allows DNN to be trained more quickly. During learning, different nodes are removed from each parameter update cycle. It also increases the efficiency of certain activities of classification.