The class to manage MLP layers of neurons. More...

#include <aims/neuralnet/multilayerperceptron.h>

Collaboration diagram for AimsMLPLayer:

[legend]

Public Member Functions
Constructor and Destructor
	AimsMLPLayer (int nneurons, int ninputs)
	Two arguments must be given to the constructor. More...

virtual	~AimsMLPLayer ()
	Destructor deletes the pointer of neuron pointers. More...

Methods
int	nNeurons () const
	Return the number of neurons in the layer. More...

AimsData< float >	activations () const
	Return the vector of activations. More...

AimsData< float >	outputs () const
	Return the vector of outputs. More...

AimsData< float >	errors () const
	Return the vector of errors. More...

AimsData< float >	nextWeights (int n) const
	Return the weights coming from input n. More...

void	setWeight (const AimsData< float > &weight, int neuron)
	Set the weights of a specific neuron of the layer. More...

void	forward (const AimsData< float > &input)
	Forward propagation on the layer. More...

float	evaluateError (const AimsData< float > &target)
	Evaluation of the error of the layer for a given target. More...

void	backPropagation (const AimsMLPLayer &nextlayer)
	Gradient back propagation on the layer. More...

void	adjust (const AimsData< float > &input, float learningrate, float momentum)
	Modification of the weights. More...

Protected Attributes
Data
int	_nNeurons
	Number of neurons. More...

AimsMLPNeuron **	_neuron
	Pointer to neuron pointers. More...

Detailed Description

The class to manage MLP layers of neurons.

Definition at line 110 of file multilayerperceptron.h.

Constructor & Destructor Documentation

◆ AimsMLPLayer()

AimsMLPLayer::AimsMLPLayer	(	int	nneurons,
		int	ninputs
	)

Two arguments must be given to the constructor.

Parameters

nneurons	number of neurons in the layer
ninputs	number of inputs on each neuron

◆ ~AimsMLPLayer()

virtual AimsMLPLayer::~AimsMLPLayer ( )

virtual

Destructor deletes the pointer of neuron pointers.

Member Function Documentation

◆ activations()

AimsData<float> AimsMLPLayer::activations ( ) const

Return the vector of activations.

◆ adjust()

void AimsMLPLayer::adjust	(	const AimsData< float > &	input,
		float	learningrate,
		float	momentum
	)

Modification of the weights.

◆ backPropagation()

void AimsMLPLayer::backPropagation ( const AimsMLPLayer & nextlayer )

Gradient back propagation on the layer.

◆ errors()

AimsData<float> AimsMLPLayer::errors ( ) const

Return the vector of errors.

◆ evaluateError()

float AimsMLPLayer::evaluateError ( const AimsData< float > & target )

Evaluation of the error of the layer for a given target.

◆ forward()

void AimsMLPLayer::forward ( const AimsData< float > & input )

Forward propagation on the layer.

◆ nextWeights()

AimsData<float> AimsMLPLayer::nextWeights ( int n ) const

Return the weights coming from input n.

◆ nNeurons()

int AimsMLPLayer::nNeurons ( ) const

Return the number of neurons in the layer.

◆ outputs()

AimsData<float> AimsMLPLayer::outputs ( ) const

Return the vector of outputs.

◆ setWeight()

void AimsMLPLayer::setWeight	(	const AimsData< float > &	weight,
		int	neuron
	)

Set the weights of a specific neuron of the layer.

Member Data Documentation

◆ _neuron

AimsMLPNeuron** AimsMLPLayer::_neuron

protected

Pointer to neuron pointers.

Definition at line 117 of file multilayerperceptron.h.

◆ _nNeurons

int AimsMLPLayer::_nNeurons