InstanceNorm< InputDataType, OutputDataType > Class Template Reference

Declaration of the Instance Normalization layer class. More...

Public Member Functions
	InstanceNorm ()
	Create the InstanceNorm object. More...
	InstanceNorm (const size_t size, const size_t batchSize, const double eps=1e-5, const bool average=true, const double momentum=0.1)
	Create the InstanceNorm layer object with the specified parameters. More...
bool	Average () const
	Get the average parameter. More...
bool	Average ()
	Modify the average parameter. More...
template < typename eT >
void	Backward (const arma::Mat< eT > &input, const arma::Mat< eT > &gy, arma::Mat< eT > &g)
	Backward pass through the layer. More...
bool	Batchsize () const
	Get the batchSize parameter. More...
bool	Batchsize ()
	Modify the batchSize parameter. More...
OutputDataType const &	Delta () const
	Get the delta. More...
OutputDataType &	Delta ()
	Modify the delta. More...
bool	Deterministic () const
	Get the value of deterministic parameter. More...
bool &	Deterministic ()
	Modify the value of deterministic parameter. More...
double	Epsilon () const
	Get the epsilon value. More...
double	Epsilon ()
	Modify the epsilon value. More...
template < typename eT >
void	Forward (const arma::Mat< eT > &input, arma::Mat< eT > &output)
	Forward pass of the Instance Normalization layer. More...
template < typename eT >
void	Gradient (const arma::Mat< eT > &input, const arma::Mat< eT > &error, arma::Mat< eT > &gradient)
	Calculate the gradient using the output delta and the input activations. More...
OutputDataType const &	Gradient () const
	Get the gradient. More...
OutputDataType &	Gradient ()
	Modify the gradient. More...
size_t	InputSize () const
	Get the number of input units / channels. More...
size_t	InputSize ()
	Modify the input units/ channels. More...
double	Momentum () const
	Get the momentum value. More...
double	Momentum ()
	Modify the momentum value. More...
OutputDataType const &	OutputParameter () const
	Get the output parameter. More...
OutputDataType &	OutputParameter ()
	Modify the output parameter. More...
OutputDataType const &	Parameters () const
	Get the parameters. More...
OutputDataType &	Parameters ()
	Modify the parameters. More...
template < typename Archive >
void	serialize (Archive &ar, const uint32_t)
	Serialize the layer. More...
OutputDataType const &	TrainingMean () const
	Get the mean over the training data. More...
OutputDataType &	TrainingMean ()
	Modify the mean over the training data. More...
OutputDataType const &	TrainingVariance () const
	Get the variance over the training data. More...
OutputDataType &	TrainingVariance ()
	Modify the variance over the training data. More...

Detailed Description

template

<

typename

InputDataType

=

arma::mat

,

typename

OutputDataType

=

arma::mat

>

class mlpack::ann::InstanceNorm< InputDataType, OutputDataType >

Declaration of the Instance Normalization layer class.

The layer transforms the input data into zero mean and unit variance and then scales and shifts the data by parameters, gamma and beta respectively. These parameters are learnt by the network. The mean and standard-deviation are calculated per-dimension separately for each object in a mini-batch.

If deterministic is false (training), the mean and variance are calculated and the data is normalized. If it is set to true (testing) then the mean and variance accrued over the training set is used.

For more information, refer to the following paper,

@article{Ulyanov17,
  author    = {Dmitry Ulyanov, Andrea Vedaldi and
               Victor Lempitsky},
  title     = {Instance Normalization:
               The Missing Ingredient for Fast Stylization},
  year      = {2017},
  url       = {https://arxiv.org/abs/1607.08022}
}

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).

Definition at line 55 of file instance_norm.hpp.

Constructor & Destructor Documentation

InstanceNorm() [1/2]

InstanceNorm

(

)

Create the InstanceNorm object.

InstanceNorm() [2/2]

InstanceNorm	(	const size_t	size,
		const size_t	batchSize,
		const double	eps = 1e-5,
		const bool	average = true,
		const double	momentum = 0.1
	)

Create the InstanceNorm layer object with the specified parameters.

Parameters

size	The number of input units / channels.
batchSize	Size of the minibatch.
eps	The epsilon added to variance to ensure numerical stability.
average	Boolean to determine whether cumulative average is used for updating the parameters or momentum.
momentum	Parameter used to update the running mean and variance.

Member Function Documentation

Average() [1/2]

bool Average ( ) const

inline

Get the average parameter.

Definition at line 166 of file instance_norm.hpp.

Average() [2/2]

bool Average ( )

inline

Modify the average parameter.

Definition at line 168 of file instance_norm.hpp.

Backward()

void Backward	(	const arma::Mat< eT > &	input,
		const arma::Mat< eT > &	gy,
		arma::Mat< eT > &	g
	)

Backward pass through the layer.

Parameters

input	The input activations
gy	The backpropagated error.
g	The calculated gradient.

Batchsize() [1/2]

bool Batchsize ( ) const

inline

Get the batchSize parameter.

Definition at line 171 of file instance_norm.hpp.

Batchsize() [2/2]

bool Batchsize ( )

inline

Modify the batchSize parameter.

Definition at line 173 of file instance_norm.hpp.

References InstanceNorm< InputDataType, OutputDataType >::serialize().

Delta() [1/2]

OutputDataType const& Delta ( ) const

inline

Get the delta.

Definition at line 125 of file instance_norm.hpp.

Delta() [2/2]

OutputDataType& Delta ( )

inline

Modify the delta.

Definition at line 127 of file instance_norm.hpp.

Deterministic() [1/2]

bool Deterministic ( ) const

inline

Get the value of deterministic parameter.

Definition at line 135 of file instance_norm.hpp.

Deterministic() [2/2]

bool& Deterministic ( )

inline

Modify the value of deterministic parameter.

Definition at line 137 of file instance_norm.hpp.

Epsilon() [1/2]

double Epsilon ( ) const

inline

Get the epsilon value.

Definition at line 155 of file instance_norm.hpp.

Epsilon() [2/2]

double Epsilon ( )

inline

Modify the epsilon value.

Definition at line 157 of file instance_norm.hpp.

Forward()

void Forward	(	const arma::Mat< eT > &	input,
		arma::Mat< eT > &	output
	)

Forward pass of the Instance Normalization layer.

Transforms the input data into zero mean and unit variance, scales the data by a factor gamma and shifts it by beta.

Parameters

input	Input data for the layer
output	Resulting output activations.

Gradient() [1/3]

void Gradient	(	const arma::Mat< eT > &	input,
		const arma::Mat< eT > &	error,
		arma::Mat< eT > &	gradient
	)

Calculate the gradient using the output delta and the input activations.

Parameters

input	The input activations
error	The calculated error
gradient	The calculated gradient.

Gradient() [2/3]

OutputDataType const& Gradient ( ) const

inline

Get the gradient.

Definition at line 130 of file instance_norm.hpp.

Gradient() [3/3]

OutputDataType& Gradient ( )

inline

Modify the gradient.

Definition at line 132 of file instance_norm.hpp.

InputSize() [1/2]

size_t InputSize ( ) const

inline

Get the number of input units / channels.

Definition at line 150 of file instance_norm.hpp.

InputSize() [2/2]

size_t InputSize ( )

inline

Modify the input units/ channels.

Definition at line 152 of file instance_norm.hpp.

Momentum() [1/2]

double Momentum ( ) const

inline

Get the momentum value.

Definition at line 161 of file instance_norm.hpp.

Momentum() [2/2]

double Momentum ( )

inline

Modify the momentum value.

Definition at line 163 of file instance_norm.hpp.

OutputParameter() [1/2]

OutputDataType const& OutputParameter ( ) const

inline

Get the output parameter.

Definition at line 118 of file instance_norm.hpp.

OutputParameter() [2/2]

OutputDataType& OutputParameter ( )

inline

Modify the output parameter.

Definition at line 122 of file instance_norm.hpp.

Parameters() [1/2]

OutputDataType const& Parameters ( ) const

inline

Get the parameters.

Definition at line 113 of file instance_norm.hpp.

Parameters() [2/2]

OutputDataType& Parameters ( )

inline

Modify the parameters.

Definition at line 115 of file instance_norm.hpp.

serialize()

void serialize	(	Archive &	ar,
		const uint32_t
	)

Serialize the layer.

Referenced by InstanceNorm< InputDataType, OutputDataType >::Batchsize().

TrainingMean() [1/2]

OutputDataType const& TrainingMean ( ) const

inline

Get the mean over the training data.

Definition at line 140 of file instance_norm.hpp.

TrainingMean() [2/2]

OutputDataType& TrainingMean ( )

inline

Modify the mean over the training data.

Definition at line 142 of file instance_norm.hpp.

TrainingVariance() [1/2]

OutputDataType const& TrainingVariance ( ) const

inline

Get the variance over the training data.

Definition at line 145 of file instance_norm.hpp.

TrainingVariance() [2/2]

OutputDataType& TrainingVariance ( )

inline

Modify the variance over the training data.

Definition at line 147 of file instance_norm.hpp.

---

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

• /home/jenkins-mlpack/mlpack.org/_src/mlpack-git/src/mlpack/methods/ann/layer/not_adapted/instance_norm.hpp