weka.dl4j.earlystopping

Class EarlyStopping

java.lang.Object

weka.dl4j.earlystopping.EarlyStopping

All Implemented Interfaces:

java.io.Serializable, weka.core.OptionHandler

public class EarlyStopping
extends java.lang.Object
implements weka.core.OptionHandler, java.io.Serializable

Early stopping implementation to stop training after N epochs without loss improvement on a separate validation set.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
EarlyStopping()
EarlyStopping(int maxEpochsNoImprovement, double validationSetPercentage) Constructor setting maxEpochsNoImprovement and validation split

Method Summary

Modifier and Type	Method and Description
boolean	evaluate(org.deeplearning4j.nn.graph.ComputationGraph model) Evaluate a model and check if the training should continue.
int	getMaxEpochsNoImprovement()
java.lang.String[]	getOptions() Gets the current settings of the Classifier.
org.nd4j.linalg.dataset.api.iterator.DataSetIterator	getValDataSetIterator() Get the validation dataset iterator
double	getValidationSetPercentage()
java.lang.String	globalInfo() Returns a string describing this search method
void	init(org.nd4j.linalg.dataset.api.iterator.DataSetIterator dsIt) Initialize the underlying dl4j EarlyStopping object
java.util.Enumeration<weka.core.Option>	listOptions() Returns an enumeration describing the available options.
void	setMaxEpochsNoImprovement(int maxEpochsNoIMprovement)
void	setOptions(java.lang.String[] options) Parses a given list of options.
void	setValidationSetPercentage(double p)

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface weka.core.OptionHandler

makeCopy

Constructor Detail

EarlyStopping

public EarlyStopping()

EarlyStopping

public EarlyStopping(int maxEpochsNoImprovement,
                     double validationSetPercentage)

Constructor setting maxEpochsNoImprovement and validation split

Parameters:

maxEpochsNoImprovement - Maximum numer of epochs with no improvement

validationSetPercentage - Validation split percentage

Method Detail

init

public void init(org.nd4j.linalg.dataset.api.iterator.DataSetIterator dsIt)

Initialize the underlying dl4j EarlyStopping object

Parameters:

dsIt - DataSet trainIterator of the validation set

evaluate

public boolean evaluate(org.deeplearning4j.nn.graph.ComputationGraph model)

Evaluate a model and check if the training should continue. Returns false if the score has not improved for the given number of epochs. Else true

Parameters:

model - Model to be evaluated against the validation set

Returns:

If training should continue or not

getMaxEpochsNoImprovement

public int getMaxEpochsNoImprovement()

setMaxEpochsNoImprovement

@OptionMetadata(displayName="max epochs with no improvement",
                description="Terminate after N epochs in which the model has shown no improvement (default = 0).",
                commandLineParamName="maxEpochsNoImprovement",
                commandLineParamSynopsis="-maxEpochsNoImprovement <int>",
                displayOrder=0)
public void setMaxEpochsNoImprovement(int maxEpochsNoIMprovement)

getValidationSetPercentage

public double getValidationSetPercentage()

setValidationSetPercentage

@OptionMetadata(displayName="validation set percentage",
                description="Percentage of training set to use for validation (default = 0).",
                commandLineParamName="valPercentage",
                commandLineParamSynopsis="-valPercentage <float>",
                displayOrder=1)
public void setValidationSetPercentage(double p)

getValDataSetIterator

public org.nd4j.linalg.dataset.api.iterator.DataSetIterator getValDataSetIterator()

Get the validation dataset iterator

Returns:

DataSetIterator for the validation set

listOptions

public java.util.Enumeration<weka.core.Option> listOptions()

Returns an enumeration describing the available options.

Specified by:

listOptions in interface weka.core.OptionHandler

Returns:

an enumeration of all the available options.

getOptions

public java.lang.String[] getOptions()

Gets the current settings of the Classifier.

Specified by:

getOptions in interface weka.core.OptionHandler

Returns:

an array of strings suitable for passing to setOptions

setOptions

public void setOptions(java.lang.String[] options)
                throws java.lang.Exception

Parses a given list of options.

Specified by:

setOptions in interface weka.core.OptionHandler

Parameters:

options - the list of options as an array of strings

Throws:

java.lang.Exception - if an option is not supported

globalInfo

public java.lang.String globalInfo()

Returns a string describing this search method

Returns:

a description of the search method suitable for displaying in the explorer/experimenter gui