SoftminActivation<T>
Models & Types · AiDotNet.ActivationFunctions
Implements the Softmin activation function, which is the opposite of Softmax and highlights the smallest values in a vector.
Inheritance: ActivationFunctionBase<T> → SoftminActivation<T>
For Beginners
While Softmax emphasizes the largest values in a vector, Softmin does the opposite - it gives more weight to smaller values. Think of it as a “smooth minimum” function. For example, if you have scores [5, 2, 8], Softmax would highlight 8 (the maximum), but Softmin would highlight 2 (the minimum).
Softmin is less commonly used than Softmax but can be useful in scenarios where you want to identify or emphasize the smallest values in a set, such as finding the closest points in certain distance-based algorithms or when you want to assign higher probabilities to smaller values.
Like Softmax, Softmin outputs values between 0 and 1 that sum to 1, creating a probability distribution.
How It Works
The Softmin function takes a vector of real numbers and transforms it into a probability distribution that emphasizes smaller values. It’s defined as: softmin(x_i) = exp(-x_i) / sum(exp(-x_j)) for all j in the vector.
Example
using AiDotNet.ActivationFunctions;
var activation = new SoftminActivation<double>();
double y = activation.Activate(0.5);
Console.WriteLine($"SoftminActivation: f(0.5) = {y:F4}");
Constructors
| Constructor | Summary |
|---|---|
SoftminActivation() |
Methods
| Method | Summary |
|---|---|
Matrix<T> Derivative(Vector<T> input) |
Calculates the Jacobian matrix of the Softmin function for a vector input. |
Tensor<T> Activate(Tensor<T> input) |
Applies Softmin to a tensor via engine primitives so the gradient tape records every step. |
Vector<T> Activate(Vector<T> input) |
Applies Softmin to a tensor via engine primitives so the gradient tape records every step. |