Implementation of a standard bidirectional recurrent neural network container. More...

#include <brnn.hpp>

Public Types
using	NetworkType = BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers... >
	Convenience typedef for the internal model construction.

Public Member Functions
	BRNN (const size_t rho, const bool single=false, OutputLayerType outputLayer=OutputLayerType(), MergeLayerType mergeLayer=new MergeLayerType(), MergeOutputType mergeOutput=new MergeOutputType(), InitializationRuleType initializeRule=InitializationRuleType())
	Create the BRNN object. More...

template<typename OptimizerType >
std::enable_if< HasMaxIterations< OptimizerType, size_t &(OptimizerType::*)()>::value, void >::type	WarnMessageMaxIterations (OptimizerType &optimizer, size_t samples) const
	Check if the optimizer has MaxIterations() parameter, if it does then check if it's value is less than the number of datapoints in the dataset. More...

template<typename OptimizerType >
std::enable_if< !HasMaxIterations< OptimizerType, size_t &(OptimizerType::*)()>::value, void >::type	WarnMessageMaxIterations (OptimizerType &optimizer, size_t samples) const
	Check if the optimizer has MaxIterations() parameter, if it doesn't then simply return from the function. More...

template<typename OptimizerType >
double	Train (arma::cube predictors, arma::cube responses, OptimizerType &optimizer)
	Train the bidirectional recurrent neural network on the given input data using the given optimizer. More...

template<typename OptimizerType = ens::StandardSGD>
double	Train (arma::cube predictors, arma::cube responses)
	Train the bidirectional recurrent neural network on the given input data. More...

void	Predict (arma::cube predictors, arma::cube &results, const size_t batchSize=256)
	Predict the responses to a given set of predictors. More...

double	Evaluate (const arma::mat &parameters, const size_t begin, const size_t batchSize, const bool deterministic)
	Evaluate the bidirectional recurrent neural network with the given parameters. More...

double	Evaluate (const arma::mat &parameters, const size_t begin, const size_t batchSize)
	Evaluate the bidirectional recurrent neural network with the given parameters. More...

template<typename GradType >
double	EvaluateWithGradient (const arma::mat &parameters, const size_t begin, GradType &gradient, const size_t batchSize)
	Evaluate the bidirectional recurrent neural network with the given parameters. More...

void	Gradient (const arma::mat &parameters, const size_t begin, arma::mat &gradient, const size_t batchSize)
	Evaluate the gradient of the bidirectional recurrent neural network with the given parameters, and with respect to only one point in the dataset. More...

void	Shuffle ()
	Shuffle the order of function visitation. More...

template<class LayerType , class... Args>
void	Add (Args... args)

void	Add (LayerTypes< CustomLayers... > layer)

size_t	NumFunctions () const
	Return the number of separable functions. (number of predictor points).

const arma::mat &	Parameters () const
	Return the initial point for the optimization.

arma::mat &	Parameters ()
	Modify the initial point for the optimization.

const size_t &	Rho () const
	Return the maximum length of backpropagation through time.

size_t &	Rho ()
	Modify the maximum length of backpropagation through time.

const arma::cube &	Responses () const
	Get the matrix of responses to the input data points.

arma::cube &	Responses ()
	Modify the matrix of responses to the input data points.

const arma::cube &	Predictors () const
	Get the matrix of data points (predictors).

arma::cube &	Predictors ()
	Modify the matrix of data points (predictors).

void	Reset ()
	Reset the state of the network. More...

void	ResetParameters ()
	Reset the module information (weights/parameters).

template<typename Archive >
void	serialize (Archive &ar, const uint32_t)
	Serialize the model.

Detailed Description

template<typename OutputLayerType = NegativeLogLikelihood<>, typename MergeLayerType = Concat<>, typename MergeOutputType = LogSoftMax<>, typename InitializationRuleType = RandomInitialization, typename... CustomLayers>
class mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >

Implementation of a standard bidirectional recurrent neural network container.

Template Parameters

OutputLayerType	The output layer type used to evaluate the network.
InitializationRuleType	Rule used to initialize the weight matrix.

Constructor & Destructor Documentation

◆ BRNN()

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::BRNN	(	const size_t	rho,
		const bool	single = `false`,
		OutputLayerType	outputLayer = `OutputLayerType()`,
		MergeLayerType *	mergeLayer = `new MergeLayerType()`,
		MergeOutputType *	mergeOutput = `new MergeOutputType()`,
		InitializationRuleType	initializeRule = `InitializationRuleType()`
	)

Create the BRNN object.

Optionally, specify which initialize rule and performance function should be used.

If you want to pass in a parameter and discard the original parameter object, be sure to use std::move to avoid unnecessary copy.

Parameters

rho	Maximum number of steps to backpropagate through time (BPTT).
single	Predict only the last element of the input sequence.
mergeLayer	Merge layer to be used to evaluate the network.
outputLayer	Output layer used to evaluate the network.
mergeOutput	Output Merge layer to be used.
initializeRule	Optional instantiated InitializationRule object for initializing the network parameter.

Member Function Documentation

◆ Evaluate() [1/2]

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

double mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::Evaluate	(	const arma::mat &	parameters,
		const size_t	begin,
		const size_t	batchSize,
		const bool	deterministic
	)

Evaluate the bidirectional recurrent neural network with the given parameters.

This function is usually called by the optimizer to train the model.

Parameters

parameters	Matrix model parameters.
begin	Index of the starting point to use for objective function evaluation.
batchSize	Number of points to be passed at a time to use for objective function evaluation.
deterministic	Whether or not to train or test the model. Note some layer act differently in training or testing mode.

◆ Evaluate() [2/2]

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

double mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::Evaluate	(	const arma::mat &	parameters,
		const size_t	begin,
		const size_t	batchSize
	)

Evaluate the bidirectional recurrent neural network with the given parameters.

This function is usually called by the optimizer to train the model. This just calls the other overload of Evaluate() with deterministic = true.

Parameters

parameters	Matrix model parameters.
begin	Index of the starting point to use for objective function evaluation.
batchSize	Number of points to be passed at a time to use for objective function evaluation.

◆ EvaluateWithGradient()

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

template<typename GradType >

double mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::EvaluateWithGradient	(	const arma::mat &	parameters,
		const size_t	begin,
		GradType &	gradient,
		const size_t	batchSize
	)

Evaluate the bidirectional recurrent neural network with the given parameters.

This function is usually called by the optimizer to train the model. This just calls the other overload of Evaluate() with deterministic = true.

Parameters

parameters	Matrix model parameters.
begin	Index of the starting point to use for objective function evaluation.
gradient	Matrix to output gradient into.
batchSize	Number of points to be passed at a time to use for objective function evaluation.

◆ Gradient()

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

void mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::Gradient	(	const arma::mat &	parameters,
		const size_t	begin,
		arma::mat &	gradient,
		const size_t	batchSize
	)

Evaluate the gradient of the bidirectional recurrent neural network with the given parameters, and with respect to only one point in the dataset.

This is useful for optimizers such as SGD, which require a separable objective function.

Parameters

parameters	Matrix of the model parameters to be optimized.
begin	Index of the starting point to use for objective function gradient evaluation.
gradient	Matrix to output gradient into.
batchSize	Number of points to be processed as a batch for objective function gradient evaluation.

◆ Predict()

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

void mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::Predict	(	arma::cube	predictors,
		arma::cube &	results,
		const size_t	batchSize = `256`
	)

Predict the responses to a given set of predictors.

The responses will reflect the output of the given output layer as returned by the output layer function.

If you want to pass in a parameter and discard the original parameter object, be sure to use std::move to avoid unnecessary copy.

The format of the data should be as follows:

each slice should correspond to a time step
each column should correspond to a data point
each row should correspond to a dimension So, e.g., predictors(i, j, k) is the i'th dimension of the j'th data point at time slice k. The responses will be in the same format.

Parameters

predictors	Input predictors.
results	Matrix to put output predictions of responses into.
batchSize	Number of points to predict at once.

◆ Reset()

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

void mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::Reset ( )

Reset the state of the network.

This ensures that all internally-held gradients are set to 0, all memory cells are reset, and the parameters matrix is the right size.

◆ Shuffle()

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

void mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::Shuffle ( )

Shuffle the order of function visitation.

This may be called by the optimizer.

◆ Train() [1/2]

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

template<typename OptimizerType >

double mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::Train	(	arma::cube	predictors,
		arma::cube	responses,
		OptimizerType &	optimizer
	)

Train the bidirectional recurrent neural network on the given input data using the given optimizer.

This will use the existing model parameters as a starting point for the optimization. If this is not what you want, then you should access the parameters vector directly with Parameters() and modify it as desired.

If you want to pass in a parameter and discard the original parameter object, be sure to use std::move to avoid unnecessary copy.

The format of the data should be as follows:

each slice should correspond to a time step
each column should correspond to a data point
each row should correspond to a dimension So, e.g., predictors(i, j, k) is the i'th dimension of the j'th data point at time slice k.

Template Parameters

OptimizerType Type of optimizer to use to train the model.

Parameters

predictors	Input training variables.
responses	Outputs results from input training variables.
optimizer	Instantiated optimizer used to train the model.

◆ Train() [2/2]

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

template<typename OptimizerType >

double mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::Train	(	arma::cube	predictors,
		arma::cube	responses
	)

Train the bidirectional recurrent neural network on the given input data.

By default, the SGD optimization algorithm is used, but others can be specified (such as ens::RMSprop).

This will use the existing model parameters as a starting point for the optimization. If this is not what you want, then you should access the parameters vector directly with Parameters() and modify it as desired.

If you want to pass in a parameter and discard the original parameter object, be sure to use std::move to avoid unnecessary copy.

The format of the data should be as follows:

each slice should correspond to a time step
each column should correspond to a data point
each row should correspond to a dimension So, e.g., predictors(i, j, k) is the i'th dimension of the j'th data point at time slice k.

Template Parameters

OptimizerType Type of optimizer to use to train the model.

Parameters

predictors	Input training variables.
responses	Outputs results from input training variables.

◆ WarnMessageMaxIterations() [1/2]

template<typename OutputLayerType , typename MergeLayerType , typename MergeOutputType , typename InitializationRuleType , typename... CustomLayers>

template<typename OptimizerType >

std::enable_if< !HasMaxIterations< OptimizerType, size_t &(OptimizerType::*)()>::value, void >::type mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::WarnMessageMaxIterations	(	OptimizerType &	optimizer,
		size_t	samples
	)		const

Check if the optimizer has MaxIterations() parameter, if it does then check if it's value is less than the number of datapoints in the dataset.

Template Parameters

OptimizerType Type of optimizer to use to train the model.

Parameters

optimizer	optimizer used in the training process.
samples	Number of datapoints in the dataset.

◆ WarnMessageMaxIterations() [2/2]

template<typename OutputLayerType = NegativeLogLikelihood<>, typename MergeLayerType = Concat<>, typename MergeOutputType = LogSoftMax<>, typename InitializationRuleType = RandomInitialization, typename... CustomLayers>

template<typename OptimizerType >

std::enable_if< !HasMaxIterations<OptimizerType, size_t&(OptimizerType::*)()>::value, void>::type mlpack::ann::BRNN< OutputLayerType, MergeLayerType, MergeOutputType, InitializationRuleType, CustomLayers >::WarnMessageMaxIterations	(	OptimizerType &	optimizer,
		size_t	samples
	)		const

Check if the optimizer has MaxIterations() parameter, if it doesn't then simply return from the function.

Template Parameters

OptimizerType Type of optimizer to use to train the model.

Parameters

optimizer	optimizer used in the training process.
samples	Number of datapoints in the dataset.

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

src/mlpack/methods/ann/brnn.hpp
src/mlpack/methods/ann/brnn_impl.hpp

Public Types

Public Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ BRNN()

Member Function Documentation

◆ Evaluate() [1/2]

◆ Evaluate() [2/2]

◆ EvaluateWithGradient()

◆ Gradient()

◆ Predict()

◆ Reset()

◆ Shuffle()

◆ Train() [1/2]

◆ Train() [2/2]

◆ WarnMessageMaxIterations() [1/2]

◆ WarnMessageMaxIterations() [2/2]