DLS Course 2 Hyperparameter tuning week 1

Why we can’t simultaneously employ L2 regularization as well as drop out method while training our neural network?

Hi, @manan:

You can! Adding a penalty term to the loss function and dropping units are not mutually exclusive :slight_smile:

It’s even mentioned in the Dropout paper:



Okay Thank you. Is there any definite answer for which regularization technique works best in general for very deep neural networks?

I think you’ll have to find out what works best for your problem :slight_smile: