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

Cosine Embedding Loss function is used for measuring whether two inputs are similar or dissimilar, using the cosine distance, and is typically used for learning nonlinear embeddings or semi-supervised learning. More...

#include <cosine_embedding_loss.hpp>

Public Member Functions
	CosineEmbeddingLoss (const double margin=0.0, const bool similarity=true, const bool takeMean=false)
	Create the CosineEmbeddingLoss object. More...
template<typename PredictionType , typename TargetType >
PredictionType::elem_type	Forward (const PredictionType &prediction, const TargetType &target)
	Ordinary feed forward pass of a neural network. 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...
InputDataType &	InputParameter () const
	Get the input parameter.
InputDataType &	InputParameter ()
	Modify the input parameter.
OutputDataType &	OutputParameter () const
	Get the output parameter.
OutputDataType &	OutputParameter ()
	Modify the output parameter.
OutputDataType &	Delta () const
	Get the delta.
OutputDataType &	Delta ()
	Modify the delta.
bool	TakeMean () const
	Get the value of takeMean.
bool &	TakeMean ()
	Modify the value of takeMean.
double	Margin () const
	Get the value of margin.
double &	Margin ()
	Modify the value of takeMean.
bool	Similarity () const
	Get the value of similarity hyperparameter.
bool &	Similarity ()
	Modify the value of takeMean.
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::CosineEmbeddingLoss< InputDataType, OutputDataType >

Cosine Embedding Loss function is used for measuring whether two inputs are similar or dissimilar, using the cosine distance, and is typically used for learning nonlinear embeddings or semi-supervised learning.

\begin{eqnarray*} f(x) = 1 - cos(x1, x2) , for y = 1 f(x) = max(0, cos(x1, x2) - margin) , for y = -1 \end{eqnarray*}

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

CosineEmbeddingLoss()

template<typename InputDataType , typename OutputDataType >

mlpack::ann::CosineEmbeddingLoss< InputDataType, OutputDataType >::CosineEmbeddingLoss	(	const double	margin = 0.0,
		const bool	similarity = true,
		const bool	takeMean = false
	)

Create the CosineEmbeddingLoss object.

Parameters

margin	Increases cosine distance in case of dissimilarity. Refer definition of cosine-embedding-loss above.
similarity	Determines whether to use similarity or dissimilarity for comparision.
takeMean	Boolean variable to specify whether to take mean or not. Specifies reduction method i.e. sum or mean corresponding to 0 and 1 respectively. Default value = 0.

Member Function Documentation

Backward()

template<typename InputDataType , typename OutputDataType >

template<typename PredictionType , typename TargetType , typename LossType >

void mlpack::ann::CosineEmbeddingLoss< InputDataType, OutputDataType >::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()

template<typename InputDataType , typename OutputDataType >

template<typename PredictionType , typename TargetType >

PredictionType::elem_type mlpack::ann::CosineEmbeddingLoss< InputDataType, OutputDataType >::Forward	(	const PredictionType &	prediction,
		const TargetType &	target
	)

Ordinary feed forward pass of a neural network.

Parameters

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

---

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

• src/mlpack/methods/ann/loss_functions/cosine_embedding_loss.hpp
• src/mlpack/methods/ann/loss_functions/cosine_embedding_loss_impl.hpp