selu

dragon.vm.tensorflow.nn.selu(
  features,
  name=None,
  **kwargs
)[source]

Apply the scaled exponential linear unit. [Klambauer et.al, 2017].

\[\text{SELU}(x) = 1.0507 * \begin{cases} x, & \text{ if } x \geq 0 \\ 1.67326 * (\exp(x) - 1), & \text{ otherwise } \end{cases} \]

Examples:

x = tf.constant([-1, 0, 1], 'float32')
print(tf.nn.selu(x, inplace=False))
Parameters:
  • features (dragon.Tensor) – The tensor \(x\).
  • name (str, optional) – The operation name.
Returns:

dragon.Tensor – The output tensor.