Leaky Rectified Linear Units are activation functions that output x when x > 0 or a small leakage value determined as the input times the leakage coefficient when x < 0. The amount of leakage is controlled by the leakage parameter.
|1||leakage||0.1||float||The amount of leakage as a proportion of the input value to allow to pass through when not inactivated.|
use Rubix\ML\NeuralNet\ActivationFunctions\LeakyReLU; $activationFunction = new LeakyReLU(0.3);
- A. L. Maas et al. (2013). Rectifier Nonlinearities Improve Neural Network Acoustic Models.