mlpack::ann::Reparametrization< InputDataType, OutputDataType > Class Template Reference

Implementation of the Reparametrization layer class. More...

#include <reparametrization.hpp>

Public Member Functions
	Reparametrization ()
	Create the Reparametrization object.
	Reparametrization (const size_t latentSize, const bool stochastic=true, const bool includeKl=true, const double beta=1)
	Create the Reparametrization layer object using the specified sample vector size. More...
	Reparametrization (const Reparametrization &layer)
	Copy Constructor.
	Reparametrization (Reparametrization &&layer)
	Move Constructor.
Reparametrization &	operator= (const Reparametrization &layer)
	Copy assignment operator.
Reparametrization &	operator= (Reparametrization &&layer)
	Move assignment operator.
template<typename eT >
void	Forward (const arma::Mat< eT > &input, arma::Mat< eT > &output)
	Ordinary feed forward pass of a neural network, evaluating the function f(x) by propagating the activity forward through f. More...
template<typename eT >
void	Backward (const arma::Mat< eT > &input, const arma::Mat< eT > &gy, arma::Mat< eT > &g)
	Ordinary feed backward pass of a neural network, calculating the function f(x) by propagating x backwards trough f. More...
OutputDataType const &	OutputParameter () const
	Get the output parameter.
OutputDataType &	OutputParameter ()
	Modify the output parameter.
OutputDataType const &	Delta () const
	Get the delta.
OutputDataType &	Delta ()
	Modify the delta.
size_t const &	OutputSize () const
	Get the output size.
size_t &	OutputSize ()
	Modify the output size.
double	Loss ()
	Get the KL divergence with standard normal.
bool	Stochastic () const
	Get the value of the stochastic parameter.
bool	IncludeKL () const
	Get the value of the includeKl parameter.
double	Beta () const
	Get the value of the beta hyperparameter.
size_t	InputShape () const
template<typename Archive >
void	serialize (Archive &ar, const uint32_t)
	Serialize the layer.

Detailed Description

template<typename InputDataType = arma::mat, typename OutputDataType = arma::mat>
class mlpack::ann::Reparametrization< InputDataType, OutputDataType >

Implementation of the Reparametrization layer class.

This layer samples from the given parameters of a normal distribution.

This class also supports beta-VAE, a state-of-the-art framework for automated discovery of interpretable factorised latent representations from raw image data in a completely unsupervised manner.

For more information, refer the following paper.

@article{ICLR2017,
  title   = {beta-VAE: Learning basic visual concepts with a constrained
             variational framework},
  author  = {Irina Higgins, Loic Matthey, Arka Pal, Christopher Burgess,
             Xavier Glorot, Matthew Botvinick, Shakir Mohamed and
             Alexander Lerchner | Google DeepMind},
  journal = {2017 International Conference on Learning Representations(ICLR)},
  year    = {2017},
  url     = {https://deepmind.com/research/publications/beta-VAE-Learning-Basic-Visual-Concepts-with-a-Constrained-Variational-Framework}
}

Template Parameters

InputDataType	Type of the input data (arma::colvec, arma::mat, arma::sp_mat or arma::cube).
OutputDataType	Type of the output data (arma::colvec, arma::mat, arma::sp_mat or arma::cube).

Constructor & Destructor Documentation

Reparametrization()

template<typename InputDataType , typename OutputDataType >

mlpack::ann::Reparametrization< InputDataType, OutputDataType >::Reparametrization	(	const size_t	latentSize,
		const bool	stochastic = true,
		const bool	includeKl = true,
		const double	beta = 1
	)

Create the Reparametrization layer object using the specified sample vector size.

Parameters

latentSize	The number of output latent units.
stochastic	Whether we want random sample or constant.
includeKl	Whether we want to include KL loss in backward function.
beta	The beta (hyper)parameter for beta-VAE mentioned above.

Member Function Documentation

Backward()

template<typename InputDataType , typename OutputDataType >

template<typename eT >

void mlpack::ann::Reparametrization< InputDataType, OutputDataType >::Backward	(	const arma::Mat< eT > &	input,
		const arma::Mat< eT > &	gy,
		arma::Mat< eT > &	g
	)

Ordinary feed backward pass of a neural network, calculating the function f(x) by propagating x backwards trough f.

Using the results from the feed forward pass.

Parameters

input	The propagated input activation.
gy	The backpropagated error.
g	The calculated gradient.

Forward()

template<typename InputDataType , typename OutputDataType >

template<typename eT >

void mlpack::ann::Reparametrization< InputDataType, OutputDataType >::Forward	(	const arma::Mat< eT > &	input,
		arma::Mat< eT > &	output
	)

Ordinary feed forward pass of a neural network, evaluating the function f(x) by propagating the activity forward through f.

Parameters

input	Input data used for evaluating the specified function.
output	Resulting output activation.

---

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

• src/mlpack/methods/ann/layer/layer_types.hpp
• src/mlpack/methods/ann/layer/reparametrization.hpp
• src/mlpack/methods/ann/layer/reparametrization_impl.hpp