Versioned name : SoftSign-9

Category : Activation function

Short description : SoftSign performs element-wise activation on a given input tensor.

Detailed description :

SoftSign operation is introduced in this article.

SoftSign Activation Function is a neuron activation function based on the mathematical function:

\[SoftSign(x) = \frac{x}{1+|x|}\]

Inputs :

  • 1 : data. Input tensor of type T

Outputs :

  • 1 : The resulting tensor of the same shape and type as the input tensor.

Types :

  • T : Arbitrary supported floating-point type.


<layer ... type="SoftSign">
        <port id="0">
        <port id="1">