net Namespace Reference

Classes
class	Adadelta
	An adaptive learning rate Trainer (Zeiler). More...
class	Backpropagation
	A classic SGDTrainer. More...
class	Layer
	A layer in a NeuralNet. More...
class	NeuralNet
	A neural network. More...
class	Neuron
	A neuron in a Layer of a NeuralNet. More...
class	Pruner
	Removes unnecessary neurons from a NeuralNet. More...
class	SGDTrainer
	A classic backpropagation SGD Trainer. More...
class	Trainer
	Trains neural networks. More...

Typedefs
typedef double(*	ActivationFunction )(double)
	The type of an activation function. More...

Functions
double	sigmoid (double initialOutput)
	A sigmoid function. More...
double	binary (double initialOutput)
	A binary function. More...
double	integer (double initialOutput)
	An flooring function. More...
double	simpleLinear (double initialOutput)
	A linear function. More...
double	tanSigmoid (double initialOutput)
	A tangential sigmoid or a hyperbolic tangent function. More...
double	sigmoidDerivative (double neuronOuput)
	Derivative of the sigmoid activation function. More...
double	simpleLinearDerivative (double neuronOutput)
	Derivative of the linear activation function. More...
double	tanSigmoidDerivative (double neuronOutput)

Typedef Documentation

typedef double(* net::ActivationFunction)(double)

The type of an activation function.

Each activation function must take in a double and output a double.

Function Documentation

double net::binary ( double  initialOutput )

inline

A binary function.

Returns

Either 0 or 1. 0 if input is less than 0 and 1 otherwise.

double net::integer ( double  initialOutput )

inline

An flooring function.

Returns

Floor of the input.

double net::sigmoid ( double  initialOutput )

inline

A sigmoid function.

An "s-shaped" function that uses an activation value of 0 to return a gradient output. Similar to the hyperbolic tangent function.

Returns

between -1 and 1

double net::sigmoidDerivative ( double  neuronOuput )

inline

Derivative of the sigmoid activation function.

double net::simpleLinear ( double  initialOutput )

inline

A linear function.

Returns

Same as the input

double net::simpleLinearDerivative ( double  neuronOutput )

inline

Derivative of the linear activation function.

double net::tanSigmoid ( double  initialOutput )

inline

A tangential sigmoid or a hyperbolic tangent function.

An "s-shaped" function that uses an activation value of 0 to return a gradient output. Similar to the sigmoid function

Returns

Same as the input

double net::tanSigmoidDerivative ( double  neuronOutput )

inline