Activation Functions

Pointwise nonlinearities applied to neural field activation. More...

Collaboration diagram for Activation Functions:

Classes
struct	dnf_composer::element::AbsSigmoidFunction
	Rational (absolute-value) sigmoid — cedar's default activation function. More...
struct	dnf_composer::element::ActivationFunction
	Abstract base for activation functions applied to a neural field. More...
struct	dnf_composer::element::HeavisideFunction
	Heaviside step function. More...
struct	dnf_composer::element::SigmoidFunction
	Logistic (exponential) sigmoid. More...

Enumerations
enum	dnf_composer::element::ActivationFunctionType : int { dnf_composer::element::SIGMOID , dnf_composer::element::HEAVISIDE , dnf_composer::element::ABSSIGMOID }
	Identifies the concrete activation function type. More...

Detailed Description

Pointwise nonlinearities applied to neural field activation.

Enumeration Type Documentation

ActivationFunctionType

enum dnf_composer::element::ActivationFunctionType : int

Identifies the concrete activation function type.

Cross-framework naming:

dnf-composer	cedar	cosivina
SIGMOID	ExpSigmoid	sigmoid
ABSSIGMOID	AbsSigmoid	—
HEAVISIDE	HeavisideSigmoid	—

Enumerator
SIGMOID	Logistic (exponential) sigmoid.
HEAVISIDE	Binary step function.
ABSSIGMOID	Rational (absolute-value) sigmoid approximation — cedar's default.