PReLU

class PReLU(alpha_initializer=<mxnet.initializer.Constant object>, **kwargs)

Parametric leaky version of a Rectified Linear Unit. <https://arxiv.org/abs/1502.01852>`_ paper.

It learns a gradient when the unit is not active

(19)\[\begin{split}f\left(x\right) = \left\{ \begin{array}{lr} \alpha x & : x \lt 0 \\ x & : x \geq 0 \\ \end{array} \right.\\\end{split}\]

where alpha is a learned parameter.

Parameters:

alpha_initializer : Initializer

Initializer for the embeddings matrix.

Inputs:

• data: input tensor with arbitrary shape.

Outputs:

• out: output tensor with the same shape as data.