This stopping criterion tracks the improvement of the error function of the training error over an interval of iterations. More...

#include <shark/Algorithms/StoppingCriteria/TrainingError.h>

Inheritance diagram for shark::TrainingError< PointType >:

Public Member Functions
	TrainingError (size_t intervalSize, double minDifference)

bool	stop (const SingleObjectiveResultSet< PointType > &set)
	returns true if training should stop More...

void	reset ()
	resets the internal state More...

Protected Attributes
std::queue< double >	m_interval
	monitored training interval More...

double	m_minDifference
	minmum difference allowed More...

size_t	m_intervalSize
	size of the interval More...

Additional Inherited Members
Public Types inherited from shark::AbstractStoppingCriterion< SingleObjectiveResultSet< PointType > >
typedef SingleObjectiveResultSet< PointType >	ResultSet

Detailed Description

template<class PointType = RealVector>
class shark::TrainingError< PointType >

This stopping criterion tracks the improvement of the error function of the training error over an interval of iterations.

If at one point, the difference between the error values of the beginning and the end of the interval are smaller than a certain value, this stopping criterion assumes convergence and stops. Of course, this may be misleading, when the algorithm temporarily gets stuck at a saddle point of the error surface. The functions assumes that the algorithm is minimizing. For details, see:

Lutz Prechelt. Early Stopping - but when? In Genevieve B. Orr and Klaus-Robert Müller: Neural Networks: Tricks of the Trade, volume 1524 of LNCS, Springer, 1997.

Definition at line 55 of file TrainingError.h.

Constructor & Destructor Documentation

◆ TrainingError()

template<class PointType = RealVector>

shark::TrainingError< PointType >::TrainingError	(	size_t	intervalSize,
		double	minDifference
	)

inline

constructs the TrainingError generalization loss

Parameters

intervalSize	size of the interval over which the progress is monitored
minDifference	minimum difference between start and end of the interval allowed before training stops

Definition at line 60 of file TrainingError.h.

References shark::TrainingError< PointType >::m_intervalSize, shark::TrainingError< PointType >::m_minDifference, and shark::TrainingError< PointType >::reset().