IdentityActivation<T>
Models & Types · AiDotNet.ActivationFunctions
Implements the Identity activation function for neural networks.
Inheritance: ActivationFunctionBase<T> → IdentityActivation<T>
Implements: IFusedActivation
For Beginners
The Identity activation function is the simplest activation function - it returns exactly what you give it.
When you pass a value through this function:
- Input of 2 ? Output of 2
- Input of -3.5 ? Output of -3.5
- And so on…
Think of it like a straight line on a graph where y = x.
While most neural networks use non-linear activation functions (like ReLU or Sigmoid) to model complex patterns, the Identity function can be useful in:
- The output layer of regression problems (when predicting continuous values)
- Testing or debugging neural networks
- Certain network architectures where you want values to pass through unchanged
Example
using AiDotNet.ActivationFunctions;
var activation = new IdentityActivation<double>();
double y = activation.Activate(0.5);
Console.WriteLine($"IdentityActivation: f(0.5) = {y:F4}");
Constructors
| Constructor | Summary |
|---|---|
IdentityActivation() |
Methods
| Method | Summary |
|---|---|
Matrix<T> Derivative(Vector<T> input) |
Calculates the derivative (gradient) of the Identity function for a vector of values. |
T Activate(T input) |
Returns the input tensor unchanged — no allocation, preserves tape chain. |
T Derivative(T input) |
Calculates the derivative (gradient) of the Identity function for a vector of values. |
Tensor<T> Activate(Tensor<T> input) |
Returns the input tensor unchanged — no allocation, preserves tape chain. |
Vector<T> Activate(Vector<T> input) |
Returns the input tensor unchanged — no allocation, preserves tape chain. |