Teem::FeedForwardNeuralNetwork Class Reference
[Controllers and neural networks]

#include <FeedForwardNeuralNetwork.h>

Inheritance diagram for Teem::FeedForwardNeuralNetwork:

List of all members.

---

Detailed Description

Class that implement a simple multilayer feedforward neural network without using the abstract NN architecture.

Public Member Functions
	FeedForwardNeuralNetwork (size_t inputCount, size_t outputCount, const std::string &root)
	Constructor, create a neural network with inputCount inputs and outputCount outputs.
virtual	~FeedForwardNeuralNetwork ()
	Destructor.
virtual void	setInput (unsigned index, double val)
	Set the input values to the neural network.
virtual double	getInput (unsigned index) const
	Return the input value.
virtual unsigned	getInputCount () const
	Returns the number of input.
virtual double	getOutput (unsigned index)
	Read the output values.
virtual void	step ()
	Propagate the input values to the output through all the layers.
size_t	layerNum ()
	Return the number of layers.
size_t	layerSize (size_t layer)
	Return the size of the nth layer. Layer 0 is the fist layer after input.
size_t	inputNum ()
	Return the number of input.
size_t	outputNum ()
	Return the number of output.
void	setWeight (size_t toLayer, size_t from, size_t to, double w)
	Set the weight of a particular synapse to a particular layer.
double	getWeight (size_t toLayer, size_t from, size_t to) const
	Get the value for a particular weight.
void	setBiasWeight (size_t toLayer, size_t to, double w)
	Set weight from bias to neuron.
double	getBiasWeight (size_t toLayer, size_t to) const
	Get weight from bias to neuron.
virtual void	randomize (double from, double to)
	Put random weights with uniform distribution.
Protected Types
typedef double(*	ActivationFunction )(double x, double b)
	Activation function y = g(x).
Static Protected Member Functions
static double	TanhForwardActivationFunction (double x, double b)
	Tanh activation function.
Protected Attributes
size_t	inputCount
	number of input
size_t	outputCount
	number of output
size_t	layerCount
	number of layer
Ishtar::Variable< unsigned >	hiddenLayerCount
	number of hidden layer
std::vector< size_t >	layerSizes
	size of layers
Ishtar::Variable< double >	biasValue
	value of the bias neuron
Ishtar::Variable< std::string >	activationFunction
	name of the activation function
Ishtar::Variable< double >	activationFunctionParameter
	parameter for the activation function
std::vector< Matrix< double > >	weights
	weight matrix for each layer
std::vector< std::valarray< double > >	biasWeights
	weight from the bias to each layer
std::vector< std::valarray< double > >	activations
	activation of each neuron
std::vector< std::valarray< double > >	outputs
	output of each neuron
std::valarray< double >	input
	input vector
ActivationFunction	forwardActFunc
	pointer to activation function
Classes
class	ActivationFunctor
	Functor to compute activation function on std::valarray<double> unsing std::for_each. More...

---

Member Typedef Documentation

typedef double(* Teem::FeedForwardNeuralNetwork::ActivationFunction)(double x, double b) [protected]

Activation function y = g(x). Parameters: x  activation b  activation function parameter (slope) Returns: output

---

Member Function Documentation

void Teem::FeedForwardNeuralNetwork::setWeight (  size_t  toLayer, size_t  from, size_t  to, double  w )  [inline]

Set the weight of a particular synapse to a particular layer. Using layer 0, synpases from input to layer 0 are set.

---

The documentation for this class was generated from the following files:

• controller/nn/FeedForwardNeuralNetwork.h
• controller/nn/FeedForwardNeuralNetwork.cpp

---