A neural network. More...

#include <NeuralNet.h>

Public Member Functions
	NeuralNet ()
	Constructs an empty neural network. More...

	NeuralNet (unsigned int numInputs, unsigned int numOutputs, unsigned int numHiddenLayers, unsigned int numNeuronsPerHiddenLayer, std::string activationFunctionName)
	Manually initializes a neural network. More...

	NeuralNet (NeuralNet *otherNet)
	Constructs a neural network copying another networks architecture with random weights. More...

	NeuralNet (std::ifstream *input)
	Restores a neural network from the specified file. More...

void	store (std::ofstream *output)
	Stores a neural network in the specified file. More...

std::vector< double >	getWeights ()
	Gets the weights of each neuron in the net. More...

std::vector< std::vector < std::vector< double > > >	getWeights3D ()
	Gets the weights of each neuron in the net. More...

void	setWeights (std::vector< double > w)
	Sets the weights of each input for each neuron in the net. More...

void	setWeights3D (std::vector< std::vector< std::vector< double > > > w)
	Sets the weights of each input for each neuron in the net. More...

void	randomizeWeights ()
	Randomizes all the weights of the neurons in the net. More...

std::vector< double >	getOutput (std::vector< double > input)
	Gets the outputs of the network for a set of inputs. More...

std::vector< std::vector < double > >	feedForward (std::vector< double > input)
	Gets the output of each layer of neurons as an input is fed for just as it is fed forward in getOutput. More...

std::vector< std::vector < std::vector< double > > >	getGradients (const std::vector< double > &input, const std::vector< double > &correctOutput)
	Computes the error gradients of each layer in the NeuralNet. More...

void	printWeights ()
	Prints the weights of the neurons of the layers of the net. More...

unsigned int	numberOfHiddenLayers ()
	Gets the number of hidden layers in the neural net. More...

unsigned int	numberOfInputs ()
	Gets the number of input neurons in the NeuralNet. More...

unsigned int	numberOfOutputs ()
	Gets the number of output neurons in the NeuralNet. More...

unsigned int	numberOfHiddenNeurons ()
	Gets the number of hidden neurons in the NeuralNet. More...

void	setOutputActivationFunction (std::string name)
	Sets the activation function of the output layer. More...

std::string	getHiddenActivationFunctionName ()
	Gets the name of the activation function used for hidden layers. More...

std::string	getOutputActivationFunctionName ()
	Gets the name of the activation function used for output layers. More...

void	removeNeuron (unsigned int hiddenLayerIndex, unsigned int neuronIndex)
	Removes a neuron from the network. More...

Public Attributes
std::vector< Layer >	net
	A two dimensional network of neurons. More...

Detailed Description

A neural network.

A flexible implementation designed to be usable in a wide scope of projects.

Constructor & Destructor Documentation

NeuralNet::NeuralNet ( )

Constructs an empty neural network.

NeuralNet::NeuralNet	(	unsigned int	numInputs,
		unsigned int	numOutputs,
		unsigned int	numHiddenLayers,
		unsigned int	numNeuronsPerHiddenLayer,
		std::string	activationFunctionName
	)

Manually initializes a neural network.

Parameters

numInputs	the number of inputs to the network (or neurons in the input layer)
numOutputs	the number of outputs to the network (or neurons in the output layer)
numHiddenLayers	the number of hidden layers
numNeuronsPerHiddenLayer	the number of neurons per hidden layer.
activationFunctionName	the name of the activation function to be used in the NN

NeuralNet::NeuralNet ( NeuralNet * otherNet )

explicit

Constructs a neural network copying another networks architecture with random weights.

The new network has the same number of inputs, outputs, hidden layers, and neurons per hidden layer as the network given.

Parameters

otherNet the network to be copied

NeuralNet::NeuralNet ( std::ifstream * input )

explicit

Restores a neural network from the specified file.

Parameters

input the file the network is to be constructed from

Member Function Documentation

std::vector< std::vector< double > > NeuralNet::feedForward ( std::vector< double > input )

Gets the output of each layer of neurons as an input is fed for just as it is fed forward in getOutput.

Parameters

input the inputs to the neural network

Returns: a 2d vector containing the outputs of each layer

std::vector< std::vector< std::vector< double > > > NeuralNet::getGradients	(	const std::vector< double > &	input,
		const std::vector< double > &	correctOutput
	)

Computes the error gradients of each layer in the NeuralNet.

Parameters

input	the input to the neural net
correctOutput	what the neural net would ideally output when fed the provided input

Returns: a 2d vector containing the gradients of each layer starting from the first layer and ending with the last

std::string NeuralNet::getHiddenActivationFunctionName ( )

Gets the name of the activation function used for hidden layers.

Returns: the string representing the activation function

std::vector< double > NeuralNet::getOutput ( std::vector< double > input )

Gets the outputs of the network for a set of inputs.

The crowning function of this class ;)

Parameters

input the inputs to the neural network

Returns: the outputs of the neural network

std::string NeuralNet::getOutputActivationFunctionName ( )

Gets the name of the activation function used for output layers.

Returns: the string representing the activation function

std::vector< double > NeuralNet::getWeights ( )

Gets the weights of each neuron in the net.

Returns: A 1d vector of neuron weights starting from the first neuron in the first layer to the last neuron in the output layer.

std::vector< std::vector< std::vector< double > > > NeuralNet::getWeights3D ( )

Gets the weights of each neuron in the net.

Returns: A 3d vector of neuron weights.

unsigned int NeuralNet::numberOfHiddenLayers ( )

Gets the number of hidden layers in the neural net.

Returns: the number of layers minus 1

unsigned int NeuralNet::numberOfHiddenNeurons ( )

Gets the number of hidden neurons in the NeuralNet.

Returns: the number of hidden neurons in the network

unsigned int NeuralNet::numberOfInputs ( )

Gets the number of input neurons in the NeuralNet.

Returns: the number of inputs to the network

unsigned int NeuralNet::numberOfOutputs ( )

Gets the number of output neurons in the NeuralNet.

Returns: the number of outputs to the network

void NeuralNet::printWeights ( )

Prints the weights of the neurons of the layers of the net.

void NeuralNet::randomizeWeights ( )

Randomizes all the weights of the neurons in the net.

void NeuralNet::removeNeuron	(	unsigned int	hiddenLayerIndex,
		unsigned int	neuronIndex
	)

Removes a neuron from the network.

Deletes all weights associated with the deleted neuron.

Returns: hiddenLayerIndex the index of the layer of the neuron to be deleted; 0 is input layer; neuronIndex the index of the neuron; same as the neuron's index in the neurons vector of its layer

void NeuralNet::setOutputActivationFunction ( std::string name )

Sets the activation function of the output layer.

Parameters

name	the name of the activation function

void NeuralNet::setWeights ( std::vector< double > w )

Sets the weights of each input for each neuron in the net.

Parameters

w	the vector of weights from which neuron weights are set, starting from the first neuron in the first layer to the last neuron in the output layer

void NeuralNet::setWeights3D ( std::vector< std::vector< std::vector< double > > > w )

Sets the weights of each input for each neuron in the net.

Parameters

w	the 3d vector of weights from which neuron weights are set

void NeuralNet::store ( std::ofstream * output )

Stores a neural network in the specified file.

Stores a neural network using ofstream. Useful for appending a network to the end of a file without overwriting it. WARNING: will not close ofstream.

Parameters

output the file the network is to be written to

Member Data Documentation

std::vector< Layer > net::NeuralNet::net

A two dimensional network of neurons.

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

/home/truell20/Documents/Fido/Fido/include/NeuralNet.h
/home/truell20/Documents/Fido/Fido/src/NeuralNet.cpp

Public Member Functions

Public Attributes

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation