LogCoshLoss< InputDataType, OutputDataType > Class Template Reference

The Log-Hyperbolic-Cosine loss function is often used to improve variational auto encoder. More...

Public Member Functions
	LogCoshLoss (const double a=1.0)
	Create the Log-Hyperbolic-Cosine object with the specified parameters. More...
double	A () const
	Get the value of hyperparameter a. More...
double &	A ()
	Modify the value of hyperparameter a. More...
template < typename PredictionType , typename TargetType , typename LossType >
void	Backward (const PredictionType &prediction, const TargetType &target, LossType &loss)
	Ordinary feed backward pass of a neural network. More...
template < typename PredictionType , typename TargetType >
PredictionType::elem_type	Forward (const PredictionType &prediction, const TargetType &target)
	Computes the Log-Hyperbolic-Cosine loss function. More...
OutputDataType &	OutputParameter () const
	Get the output parameter. More...
OutputDataType &	OutputParameter ()
	Modify the output parameter. More...
template < typename Archive >
void	serialize (Archive &ar, const uint32_t)
	Serialize the loss function. More...

Detailed Description

template

<

typename

InputDataType

=

arma::mat

,

typename

OutputDataType

=

arma::mat

>

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

The Log-Hyperbolic-Cosine loss function is often used to improve variational auto encoder.

This function is the log of hyperbolic cosine of difference between true values and predicted values.

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 35 of file log_cosh_loss.hpp.

Constructor & Destructor Documentation

LogCoshLoss()

LogCoshLoss

(

const double

a = 1.0

)

Create the Log-Hyperbolic-Cosine object with the specified parameters.

Parameters

a	A double type value for smoothening loss function. It must be positive a real number, Sharpness of loss function is directly proportional to a. It can also act as a scaling factor hence making the loss function more sensitive to small losses around the origin. Default value = 1.0.

Member Function Documentation

A() [1/2]

double A ( ) const

inline

Get the value of hyperparameter a.

Definition at line 81 of file log_cosh_loss.hpp.

A() [2/2]

double& A ( )

inline

Modify the value of hyperparameter a.

Definition at line 83 of file log_cosh_loss.hpp.

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

Backward()

void Backward	(	const PredictionType &	prediction,
		const TargetType &	target,
		LossType &	loss
	)

Ordinary feed backward pass of a neural network.

Parameters

prediction	Predictions used for evaluating the specified loss function.
target	The target vector.
loss	The calculated error.

Forward()

PredictionType::elem_type Forward	(	const PredictionType &	prediction,
		const TargetType &	target
	)

Computes the Log-Hyperbolic-Cosine loss function.

Parameters

prediction	Predictions used for evaluating the specified loss function.
target	Target data to compare with.

OutputParameter() [1/2]

OutputDataType& OutputParameter ( ) const

inline

Get the output parameter.

Definition at line 76 of file log_cosh_loss.hpp.

OutputParameter() [2/2]

OutputDataType& OutputParameter ( )

inline

Modify the output parameter.

Definition at line 78 of file log_cosh_loss.hpp.

serialize()

void serialize	(	Archive &	ar,
		const uint32_t
	)

Serialize the loss function.

Referenced by LogCoshLoss< InputDataType, OutputDataType >::A().

---

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

• /home/ryan/src/mlpack.org/_src/mlpack-git/src/mlpack/methods/ann/loss_functions/log_cosh_loss.hpp