mlpack::perceptron::Perceptron< LearnPolicy, WeightInitializationPolicy, MatType > Class Template Reference

This class implements a simple perceptron (i.e., a single layer neural network). More...

#include <perceptron.hpp>

Public Member Functions
	Perceptron (const size_t numClasses=0, const size_t dimensionality=0, const size_t maxIterations=1000)
	Constructor: create the perceptron with the given number of classes and initialize the weight matrix, but do not perform any training. More...
	Perceptron (const MatType &data, const arma::Row< size_t > &labels, const size_t numClasses, const size_t maxIterations=1000)
	Constructor: constructs the perceptron by building the weights matrix, which is later used in classification. More...
	Perceptron (const Perceptron &other, const MatType &data, const arma::Row< size_t > &labels, const size_t numClasses, const arma::rowvec &instanceWeights)
	Alternate constructor which copies parameters from an already initiated perceptron. More...
void	Train (const MatType &data, const arma::Row< size_t > &labels, const size_t numClasses, const arma::rowvec &instanceWeights=arma::rowvec())
	Train the perceptron on the given data for up to the maximum number of iterations (specified in the constructor or through MaxIterations()). More...
void	Classify (const MatType &test, arma::Row< size_t > &predictedLabels)
	Classification function. More...
template<typename Archive >
void	serialize (Archive &ar, const uint32_t)
	Serialize the perceptron.
size_t	MaxIterations () const
	Get the maximum number of iterations.
size_t &	MaxIterations ()
	Modify the maximum number of iterations.
size_t	NumClasses () const
	Get the number of classes this perceptron has been trained for.
const arma::mat &	Weights () const
	Get the weight matrix.
arma::mat &	Weights ()
	Modify the weight matrix. You had better know what you are doing!
const arma::vec &	Biases () const
	Get the biases.
arma::vec &	Biases ()
	Modify the biases. You had better know what you are doing!

Detailed Description

template<typename LearnPolicy = SimpleWeightUpdate, typename WeightInitializationPolicy = ZeroInitialization, typename MatType = arma::mat>
class mlpack::perceptron::Perceptron< LearnPolicy, WeightInitializationPolicy, MatType >

This class implements a simple perceptron (i.e., a single layer neural network).

It converges if the supplied training dataset is linearly separable.

Template Parameters

LearnPolicy	Options of SimpleWeightUpdate and GradientDescent.
WeightInitializationPolicy	Option of ZeroInitialization and RandomInitialization.

Constructor & Destructor Documentation

Perceptron() [1/3]

template<typename LearnPolicy , typename WeightInitializationPolicy , typename MatType >

mlpack::perceptron::Perceptron< LearnPolicy, WeightInitializationPolicy, MatType >::Perceptron	(	const size_t	numClasses = 0,
		const size_t	dimensionality = 0,
		const size_t	maxIterations = 1000
	)

Constructor: create the perceptron with the given number of classes and initialize the weight matrix, but do not perform any training.

Construct the perceptron with the given number of classes and maximum number of iterations.

(Call the Train() function to perform training.)

Parameters

numClasses	Number of classes in the dataset.
dimensionality	Dimensionality of the dataset.
maxIterations	Maximum number of iterations for the perceptron learning algorithm.

Perceptron() [2/3]

template<typename LearnPolicy , typename WeightInitializationPolicy , typename MatType >

mlpack::perceptron::Perceptron< LearnPolicy, WeightInitializationPolicy, MatType >::Perceptron	(	const MatType &	data,
		const arma::Row< size_t > &	labels,
		const size_t	numClasses,
		const size_t	maxIterations = 1000
	)

Constructor: constructs the perceptron by building the weights matrix, which is later used in classification.

Constructor - constructs the perceptron.

The number of classes should be specified separately, and the labels vector should contain values in the range [0, numClasses - 1]. The data::NormalizeLabels() function can be used if the labels vector does not contain values in the required range.

Parameters

data	Input, training data.
labels	Labels of dataset.
numClasses	Number of classes in the dataset.
maxIterations	Maximum number of iterations for the perceptron learning algorithm.

Or rather, builds the weights matrix, which is later used in classification. It adds a bias input vector of 1 to the input data to take care of the bias weights.

Parameters

data	Input, training data.
labels	Labels of dataset.
maxIterations	Maximum number of iterations for the perceptron learning algorithm.

Perceptron() [3/3]

template<typename LearnPolicy , typename WeightInitializationPolicy , typename MatType >

mlpack::perceptron::Perceptron< LearnPolicy, WeightInitializationPolicy, MatType >::Perceptron	(	const Perceptron< LearnPolicy, WeightInitializationPolicy, MatType > &	other,
		const MatType &	data,
		const arma::Row< size_t > &	labels,
		const size_t	numClasses,
		const arma::rowvec &	instanceWeights
	)

Alternate constructor which copies parameters from an already initiated perceptron.

Parameters

other	The other initiated Perceptron object from which we copy the values from.
data	The data on which to train this Perceptron object on.
labels	The labels of data.
numClasses	Number of classes in the data.
instanceWeights	Weight vector to use while training. For boosting purposes.
other	The other initiated Perceptron object from which we copy the values from.
data	The data on which to train this Perceptron object on.
instanceWeights	Weight vector to use while training. For boosting purposes.
labels	The labels of data.

Member Function Documentation

Classify()

template<typename LearnPolicy , typename WeightInitializationPolicy , typename MatType >

void mlpack::perceptron::Perceptron< LearnPolicy, WeightInitializationPolicy, MatType >::Classify	(	const MatType &	test,
		arma::Row< size_t > &	predictedLabels
	)

Classification function.

After training, use the weights matrix to classify test, and put the predicted classes in predictedLabels.

Parameters

test	Testing data or data to classify.
predictedLabels	Vector to store the predicted classes after classifying test.

Train()

template<typename LearnPolicy , typename WeightInitializationPolicy , typename MatType >

void mlpack::perceptron::Perceptron< LearnPolicy, WeightInitializationPolicy, MatType >::Train	(	const MatType &	data,
		const arma::Row< size_t > &	labels,
		const size_t	numClasses,
		const arma::rowvec &	instanceWeights = arma::rowvec()
	)

Train the perceptron on the given data for up to the maximum number of iterations (specified in the constructor or through MaxIterations()).

Training function.

A single iteration corresponds to a single pass through the data, so if you want to pass through the dataset only once, set MaxIterations() to 1.

This training does not reset the model weights, so you can call Train() on multiple datasets sequentially.

Parameters

data	Dataset on which training should be performed.
labels	Labels of the dataset.
numClasses	Number of classes in the data.
instanceWeights	Cost matrix. Stores the cost of mispredicting instances. This is useful for boosting.

It trains on trainData using the cost matrix instanceWeights.

Parameters

data	Data to train on.
labels	Labels of data.
instanceWeights	Cost matrix. Stores the cost of mispredicting instances. This is useful for boosting.

---

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

• src/mlpack/methods/perceptron/perceptron.hpp
• src/mlpack/methods/perceptron/perceptron_impl.hpp