Public Member Functions
	ConnectedNet (int inputNodes, std::vector< int > hiddenLayers, int outputNodes, WeightGenerator *weightGen=nullptr)

	~ConnectedNet ()

std::vector< double >	getOutput (std::vector< double > inputValues)

void	train (TrainingData &tData, double learningRate)

std::vector< double >	getDifference ()

Detailed Description

A fully connected Neural Network

Constructor & Destructor Documentation

Create a new ConnectedNet of given dimensions.

For example: For a 3x5x4x4x2 network one would put
inputNodes - 3
hiddenLayers - [5, 4, 4]
outputNodes - 2

Parameters

inputNodes	amount of inputs to the network
hiddenLayers	vector of dimensions of hidden layers, number at each index becomes amount of neurons in hidden layer at that index
outputNodes	amount of outputs from the network
weightGen	generator to use for initializing weights in the network, if left a nullptr it is replaced with a default random generator

ConnectedNet::~ConnectedNet ( )

Destructor for ConnectedNet, clears all memory allocated

std::vector< double > ConnectedNet::getDifference ( )

Get a vector of difference between target and output

std::vector< double > ConnectedNet::getOutput ( std::vector< double > inputValues )

Present an input to the ConnectedNet and get the output

Parameters

inputValues input vector to the net

Train the net on one input - target vector pair with speed learningrate returns the error

Parameters

tData	a piece of training data to use for training the net
learningRate	rate to adjust weights in training

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