Medium Last updated on May 7, 2022, 1:06 a.m.

Dropout is a method used by Neural Networks to avoid **Overfitting**. In simple terms, During training a Neural Networks, we randomly choose certain neurons and set them to zero (ignore) for forward pass or backward pass.

These neurons are chosen by a probability p, i.e; individual nodes are kept with probability (1-*p)* so that a reduced network is left; incoming and outgoing edges to a dropped-out node are also removed.

- Dropout forces a neural network to learn more robust features that are useful in conjunction with many different random subsets of the other neurons.
- Dropout roughly doubles the number of iterations required to converge. However, training time for each epoch is less.
- With H hidden units, each of which can be dropped, we have 2^H possible models. In testing phase, the entire network is considered and each activation is reduced by a factor
*p.*

Inverted Dropout Method is commonly used for Drop out Implementation

Frequently Asked Questions by

Apple Amazon IBM