I didn’t understand the point why dividing a3 by keep prob and later on z3 by keep prob? If we want to scale up . shouldn’t we multiply instead? [image]

Course 2 :week 1

nramon September 3, 2021, 1:01pm 2

Here’s the reason for dividing by keep_prob.

About the scale up terminology, remember that keep_prob is a number between 0 and 1.

Good luck with week one

2 Likes

Topic		Replies	Views
Dropout scaling fix (division by keep_prob) Improving Deep Neural Networks: Hyperparameter tun	3	678	September 28, 2022
Why do you divide the activations by keep_prob when you use drop Improving Deep Neural Networks: Hyperparameter tun	7	680	May 22, 2023
Regularization Improving Deep Neural Networks: Hyperparameter tun	3	592	July 15, 2023
Doubt about the implementation of inverted dropout Improving Deep Neural Networks: Hyperparameter tun	5	825	November 19, 2024
Dropout: Why divide by keep prob? Improving Deep Neural Networks: Hyperparameter tun	1	841	July 10, 2021