BNNS.ActivationFunction.softmax
An activation function that returns the softmax function of its input.
Declaration
case softmaxDiscussion
This constant defines an activation function that returns values using the following formula:
[Image]
The softmax function transforms a vector of real numbers into a vector of probabilities. Each probability in the result is in the range 0…1, and the sum of the probabilities is 1.
For example, given and array that contains the values [3.0, 5.0, 1.0, 6.0, 2.0, 1.0, 4.0], the softmax function calculates the following values:
Inputs | Outputs |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Changing the fourth element from 6.0 to 10.0 increases its probability to almost 1.0:
Inputs | Outputs |
|---|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
See Also
Activation Functions
BNNS.ActivationFunction.absBNNS.ActivationFunction.celu(alpha:)BNNS.ActivationFunction.clamp(bounds:)BNNS.ActivationFunction.clampedLeakyRectifiedLinear(alpha:beta:)BNNS.ActivationFunction.elu(alpha:)BNNS.ActivationFunction.geluApproximation(alpha:beta:)BNNS.ActivationFunction.geluApproximation2(alpha:beta:)BNNS.ActivationFunction.gumbel(alpha:beta:)BNNS.ActivationFunction.gumbelMax(alpha:beta:)BNNS.ActivationFunction.hardShrink(alpha:)BNNS.ActivationFunction.hardSigmoid(alpha:beta:)BNNS.ActivationFunction.hardSwish(alpha:beta:)BNNS.ActivationFunction.identityBNNS.ActivationFunction.leakyRectifiedLinear(alpha:)BNNS.ActivationFunction.linear(alpha:)