ActivationPropertyAttribute

Attributes · AiDotNet.Attributes

Declares mathematical properties of an activation function for automatic test generation and cataloging. The source generator reads these to configure invariant test parameters.

Inheritance: Attribute → ActivationPropertyAttribute

Constructors

Constructor Summary
ActivationPropertyAttribute()

Properties

Property Summary
ComputeCost Cost { get; set; } Relative computational cost.
bool HasLearnableParameters { get; set; } Whether it has learnable parameters (PReLU).
bool IsBounded { get; set; } Whether the output is bounded.
bool IsDifferentiable { get; set; } Whether it’s differentiable everywhere.
bool IsMonotonic { get; set; } Whether the activation is monotonically non-decreasing.
bool IsStochastic { get; set; } Whether the activation uses randomness during training (e.g., RReLU).
bool IsVectorActivation { get; set; } Whether this operates on vectors (Softmax) vs scalars (ReLU).
bool ZeroPreserving { get; set; } Whether Activate(0) produces exactly 0.
double BoundLower { get; set; } Lower bound of the output range for bounded activations.
double BoundUpper { get; set; } Upper bound of the output range for bounded activations.