[source] # Softsign A smooth sigmoid-shaped function that squashes the input between -1 and 1. $$ {\displaystyle Softsign = {\frac {x}{1+|x|}}} $$ ## Parameters This activation function does not have any parameters. ## Example ```php use Rubix\ML\NeuralNet\ActivationFunctions\Softsign; $activationFunction = new Softsign(); ``` ## References [^1]: X. Glorot et al. (2010). Understanding the Difficulty of Training Deep Feedforward Neural Networks.