Skip to main content

Dropout

Dropout Regularization

Randomly zeros elements during training to prevent overfitting. During evaluation, the layer is a no-op (identity function).

Parameters:

  • p: Probability of an element being zeroed (0 to 1, typical: 0.1-0.5)

Shape Contract:

  • Input: [*shape] arbitrary shape
  • Output: [*shape] same shape as input

Notes:

  • Outputs are scaled by 1/(1-p) during training to maintain expected values
  • Only active during training (disabled during evaluation)
  • Common dropout rates: 0.1 (mild), 0.3 (moderate), 0.5 (strong)

Signature​

neuron Dropout(p)

Ports​

Inputs:

  • default: [*shape]

Outputs:

  • default: [*shape]

Implementation​

"from core import regularization/Dropout"
Source { source: "core", path: "regularization/Dropout" }