Dropout< InputDataType, OutputDataType > Class Template Reference

The dropout layer is a regularizer that randomly with probability ratio sets input values to zero and scales the remaining elements by factor 1 / (1 - ratio). More...

Public Member Functions
	Dropout (const double ratio=0.5, const bool rescale=true)
	Create the Dropout object using the specified ratio and rescale parameter. More...
template < typename eT >
void	Backward (const arma::Mat< eT > &&, arma::Mat< eT > &&gy, arma::Mat< eT > &&g)
	Ordinary feed backward pass of the dropout layer. More...
OutputDataType const &	Delta () const
	Get the detla. More...
OutputDataType &	Delta ()
	Modify the delta. More...
bool	Deterministic () const
	The value of the deterministic parameter. More...
bool &	Deterministic ()
	Modify the value of the deterministic parameter. More...
template < typename eT >
void	Forward (const arma::Mat< eT > &&input, arma::Mat< eT > &&output)
	Ordinary feed forward pass of the dropout layer. More...
InputDataType const &	InputParameter () const
	Get the input parameter. More...
InputDataType &	InputParameter ()
	Modify the input parameter. More...
OutputDataType const &	OutputParameter () const
	Get the output parameter. More...
OutputDataType &	OutputParameter ()
	Modify the output parameter. More...
double	Ratio () const
	The probability of setting a value to zero. More...
void	Ratio (const double r)
	Modify the probability of setting a value to zero. More...
bool	Rescale () const
	The value of the rescale parameter. More...
bool &	Rescale ()
	Modify the value of the rescale parameter. More...
template < typename Archive >
void	serialize (Archive &ar, const unsigned int)
	Serialize the layer. More...

Detailed Description

template

<

typename

InputDataType

=

arma::mat

,

typename

OutputDataType

=

arma::mat

>

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

The dropout layer is a regularizer that randomly with probability ratio sets input values to zero and scales the remaining elements by factor 1 / (1 - ratio).

If rescale is true the input is scaled with 1 / (1-p) when deterministic is false. In the deterministic mode (during testing), the layer just scales the output.

Note: During training you should set deterministic to false and during testing you should set deterministic to true.

For more information, see the following.

@article{Hinton2012,
  author  = {Geoffrey E. Hinton, Nitish Srivastava, Alex Krizhevsky,
             Ilya Sutskever, Ruslan Salakhutdinov},
  title   = {Improving neural networks by preventing co-adaptation of feature
             detectors},
  journal = {CoRR},
  volume  = {abs/1207.0580},
  year    = {2012},
}

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 54 of file dropout.hpp.

Constructor & Destructor Documentation

Dropout()

Dropout	(	const double	ratio = 0.5,
		const bool	rescale = true
	)

Create the Dropout object using the specified ratio and rescale parameter.

Parameters

ratio	The probability of setting a value to zero.
rescale	If true the input is rescaled when deterministic is False.

Member Function Documentation

Backward()

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

Ordinary feed backward pass of the dropout layer.

Parameters

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

Delta() [1/2]

OutputDataType const& Delta ( ) const

inline

Get the detla.

Definition at line 98 of file dropout.hpp.

Delta() [2/2]

OutputDataType& Delta ( )

inline

Modify the delta.

Definition at line 100 of file dropout.hpp.

Deterministic() [1/2]

bool Deterministic ( ) const

inline

The value of the deterministic parameter.

Definition at line 103 of file dropout.hpp.

Deterministic() [2/2]

bool& Deterministic ( )

inline

Modify the value of the deterministic parameter.

Definition at line 105 of file dropout.hpp.

Forward()

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

Ordinary feed forward pass of the dropout layer.

Parameters

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

InputParameter() [1/2]

InputDataType const& InputParameter ( ) const

inline

Get the input parameter.

Definition at line 88 of file dropout.hpp.

InputParameter() [2/2]

InputDataType& InputParameter ( )

inline

Modify the input parameter.

Definition at line 90 of file dropout.hpp.

OutputParameter() [1/2]

OutputDataType const& OutputParameter ( ) const

inline

Get the output parameter.

Definition at line 93 of file dropout.hpp.

OutputParameter() [2/2]

OutputDataType& OutputParameter ( )

inline

Modify the output parameter.

Definition at line 95 of file dropout.hpp.

Ratio() [1/2]

double Ratio ( ) const

inline

The probability of setting a value to zero.

Definition at line 108 of file dropout.hpp.

Ratio() [2/2]

void Ratio ( const double  r )

inline

Modify the probability of setting a value to zero.

Definition at line 111 of file dropout.hpp.

Rescale() [1/2]

bool Rescale ( ) const

inline

The value of the rescale parameter.

Definition at line 118 of file dropout.hpp.

Rescale() [2/2]

bool& Rescale ( )

inline

Modify the value of the rescale parameter.

Definition at line 120 of file dropout.hpp.

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

serialize()

void serialize	(	Archive &	ar,
		const unsigned	int
	)

Serialize the layer.

Referenced by Dropout< InputDataType, OutputDataType >::Rescale().

---

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

• src/mlpack/methods/ann/layer/dropout.hpp