Dropout (neural networks)

Common card · No. 478563

Dropout is a regularization technique for reducing overfitting in artificial neural networks by preventing complex co-adaptations on training data. The technique consists of randomly setting the inputs and/or outputs of neurons to zero. It is performed during the training process of a neural network, and is normally disabled during inference. A related older technique is dilution, which randomly decreases weights towards zero.

Read on Wikipedia · Image Daniel Voigt Godoy, CC BY 4.0