de.jstacs.classifiers.differentiableSequenceScoreBased.sampling

Class SamplingGenDisMixClassifier

java.lang.Object

de.jstacs.classifiers.AbstractClassifier

de.jstacs.classifiers.AbstractScoreBasedClassifier

de.jstacs.classifiers.differentiableSequenceScoreBased.sampling.SamplingScoreBasedClassifier

de.jstacs.classifiers.differentiableSequenceScoreBased.sampling.SamplingGenDisMixClassifier

All Implemented Interfaces:

Storable, Cloneable

public class SamplingGenDisMixClassifier
extends SamplingScoreBasedClassifier

A classifier that samples its parameters from a LogGenDisMixFunction using the Metropolis-Hastings algorithm. For details on the algorithm see SamplingScoreBasedClassifier. The LogGenDisMixFunction includes several known posterior distributions, including the posterior (LearningPrinciple.MAP) and the supervised posterior (LearningPrinciple.MSP). For non-uniform values of the mixture parameters beta the distribution we sample from is less well defined, although sampling is possible in general.

Author:

Jan Grau

Nested Class Summary

Nested classes/interfaces inherited from class de.jstacs.classifiers.differentiableSequenceScoreBased.sampling.SamplingScoreBasedClassifier

SamplingScoreBasedClassifier.DiffSMSamplingComponent, SamplingScoreBasedClassifier.SamplingScheme

Nested classes/interfaces inherited from class de.jstacs.classifiers.AbstractScoreBasedClassifier

AbstractScoreBasedClassifier.DoubleTableResult

Field Summary

Fields inherited from class de.jstacs.classifiers.differentiableSequenceScoreBased.sampling.SamplingScoreBasedClassifier

burnInTest, currentParameters, currentScore, initParameters, lastParameters, lastScore, params, previousParameters, scoringFunctions

Constructor Summary

Constructors

Constructor and Description
SamplingGenDisMixClassifier(SamplingGenDisMixClassifierParameterSet params, BurnInTest burnInTest, double[] classVariances, LogPrior prior, double[] beta, SamplingDifferentiableStatisticalModel... scoringFunctions) Creates a new SamplingGenDisMixClassifier using the external parameters params, a burn-in test, a set of sampling variances for the different classes, a prior on the parameters, weights beta for the three components of the LogGenDisMixFunction, i.e., likelihood, conditional likelihood, and prior, and scoring functions that model the distribution for each of the classes.
SamplingGenDisMixClassifier(SamplingGenDisMixClassifierParameterSet params, BurnInTest burnInTest, double[] classVariances, LogPrior prior, LearningPrinciple principle, SamplingDifferentiableStatisticalModel... scoringFunctions) Creates a new SamplingGenDisMixClassifier using the external parameters params, a burn-in test, a set of sampling variances for the different classes, a prior on the parameters, a learning principle, and scoring functions that model the distribution for each of the classes.
SamplingGenDisMixClassifier(StringBuffer xml) Creates a new SamplingGenDisMixClassifier from its XML-representation

Method Summary

Methods

Modifier and Type	Method and Description
protected void	extractFurtherClassifierInfosFromXML(StringBuffer xml) Extracts further information of a classifier from an XML representation.
GenDisMixClassifier	getClassifierForBestParameters(GenDisMixClassifierParameterSet params) Returns a standard, i.e., non-sampling, GenDisMixClassifier, where the parameters are set to those that yielded the maximum value of the objective functions among all sampled parameter values.
GenDisMixClassifier	getClassifierForMeanParameters(GenDisMixClassifierParameterSet params, boolean testBurnIn, int minBurnInSteps) Returns a standard, i.e., non-sampling, GenDisMixClassifier, where the parameters are set to the mean values over all sampled parameter values in the stationary phase.
protected DiffSSBasedOptimizableFunction	getFunction(DataSet[] data, double[][] weights) Returns the function that should be sampled from.
protected StringBuffer	getFurtherClassifierInfos() This method returns further information of a classifier as a StringBuffer.
protected String	getXMLTag() Returns the String that is used as tag for the XML representation of the classifier.
protected double	modifyFunctionValue(double value) Allows for a modification of the value returned by the function obtained by SamplingScoreBasedClassifier.getFunction(DataSet[], double[][]).

Methods inherited from class de.jstacs.classifiers.differentiableSequenceScoreBased.sampling.SamplingScoreBasedClassifier

doOneSamplingStep, doSingleSampling, getBestParameters, getClassifierAnnotation, getDeleteOnExit, getInstanceName, getMeanParameters, getNumericalCharacteristics, getSamplingComponent, getScore, getScores, getTempDir, init, isInitialized, joinAndSetParameterFiles, precomputeBurnInLength, sample, sampleNSteps, setDeleteOnExit, setInitParameters, setTempDir, train

Methods inherited from class de.jstacs.classifiers.AbstractScoreBasedClassifier

check, check, classify, classify, clone, createDefaultClassWeights, getClassWeight, getClassWeights, getMultiClassScores, getNumberOfClasses, getPValue, getPValue, getResults, getScore, setClassWeights, setClassWeights, setThresholdClassWeights

Methods inherited from class de.jstacs.classifiers.AbstractClassifier

classify, evaluate, evaluate, getAlphabetContainer, getCharacteristics, getLength, toXML, train

Methods inherited from class java.lang.Object

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

Constructor Detail

SamplingGenDisMixClassifier

public SamplingGenDisMixClassifier(SamplingGenDisMixClassifierParameterSet params,
                           BurnInTest burnInTest,
                           double[] classVariances,
                           LogPrior prior,
                           double[] beta,
                           SamplingDifferentiableStatisticalModel... scoringFunctions)
                            throws CloneNotSupportedException

Creates a new SamplingGenDisMixClassifier using the external parameters params, a burn-in test, a set of sampling variances for the different classes, a prior on the parameters, weights beta for the three components of the LogGenDisMixFunction, i.e., likelihood, conditional likelihood, and prior, and scoring functions that model the distribution for each of the classes.

Parameters:

params - the external parameters

burnInTest - the burn-in test, or null for no burn-in test

classVariances - the sampling variances for the parameters in the different classes

prior - the prior on the parameters

beta - The weights of the three components of the LogGenDisMixFunction

scoringFunctions - the scoring functions for the different classes

Throws:

CloneNotSupportedException - if the scoring functions could not be cloned

SamplingGenDisMixClassifier

public SamplingGenDisMixClassifier(SamplingGenDisMixClassifierParameterSet params,
                           BurnInTest burnInTest,
                           double[] classVariances,
                           LogPrior prior,
                           LearningPrinciple principle,
                           SamplingDifferentiableStatisticalModel... scoringFunctions)
                            throws CloneNotSupportedException

Creates a new SamplingGenDisMixClassifier using the external parameters params, a burn-in test, a set of sampling variances for the different classes, a prior on the parameters, a learning principle, and scoring functions that model the distribution for each of the classes.

Parameters:

params - the external parameters

burnInTest - the burn-in test, or null for no burn-in test

classVariances - the sampling variances for the parameters in the different classes

prior - the prior on the parameters

principle - the learning principle, i.e., the objective function we sample from

scoringFunctions - the scoring functions for the different classes

Throws:

CloneNotSupportedException - if the scoring functions could not be cloned

SamplingGenDisMixClassifier

public SamplingGenDisMixClassifier(StringBuffer xml)
                            throws NonParsableException

Creates a new SamplingGenDisMixClassifier from its XML-representation

Parameters:

xml - the XML-representation

Throws:

NonParsableException - if xml could not be parsed

Method Detail

getFunction

protected DiffSSBasedOptimizableFunction getFunction(DataSet[] data,
                                         double[][] weights)
                                              throws Exception

Description copied from class: SamplingScoreBasedClassifier

Returns the function that should be sampled from.

Specified by:

getFunction in class SamplingScoreBasedClassifier

Parameters:

data - the samples

weights - the weights of the sequences of the samples

Returns:

the function that should be sampled from

Throws:

Exception - if the function could not be created

modifyFunctionValue

protected double modifyFunctionValue(double value)

Description copied from class: SamplingScoreBasedClassifier

Allows for a modification of the value returned by the function obtained by SamplingScoreBasedClassifier.getFunction(DataSet[], double[][]). This is for instance necessary in case of LogGenDisMixFunction to obtain a proper posterior or supervised posterior.

Overrides:

modifyFunctionValue in class SamplingScoreBasedClassifier

Parameters:

value - the original value

Returns:

the modified value

getXMLTag

protected String getXMLTag()

Description copied from class: AbstractClassifier

Returns the String that is used as tag for the XML representation of the classifier. This method is used by the methods AbstractClassifier.fromXML(StringBuffer) and AbstractClassifier.toXML().

Specified by:

getXMLTag in class AbstractClassifier

Returns:

the String that is used as tag for the XML representation of the classifier

getFurtherClassifierInfos

protected StringBuffer getFurtherClassifierInfos()

Description copied from class: AbstractClassifier

This method returns further information of a classifier as a StringBuffer. This method is used by the method AbstractClassifier.toXML() and should not be made public.

Overrides:

getFurtherClassifierInfos in class SamplingScoreBasedClassifier

Returns:

further information of a classifier as a StringBuffer

See Also:

AbstractClassifier.toXML()

extractFurtherClassifierInfosFromXML

protected void extractFurtherClassifierInfosFromXML(StringBuffer xml)
                                             throws NonParsableException

Description copied from class: AbstractClassifier

Extracts further information of a classifier from an XML representation. This method is used by the method AbstractClassifier.fromXML(StringBuffer) and should not be made public.

Overrides:

extractFurtherClassifierInfosFromXML in class SamplingScoreBasedClassifier

Parameters:

xml - the XML representation as StringBuffer

Throws:

NonParsableException - if the information could not be parsed out of the XML representation (the StringBuffer could not be parsed)

See Also:

AbstractClassifier.fromXML(StringBuffer)

getClassifierForBestParameters

public GenDisMixClassifier getClassifierForBestParameters(GenDisMixClassifierParameterSet params)
                                                   throws Exception

Returns a standard, i.e., non-sampling, GenDisMixClassifier, where the parameters are set to those that yielded the maximum value of the objective functions among all sampled parameter values.

Parameters:

params - the external parameters of the GenDisMixClassifier

Returns:

the GenDisMixClassifier with set parameter values

Throws:

Exception - if the GenDisMixClassifier could not be created

getClassifierForMeanParameters

public GenDisMixClassifier getClassifierForMeanParameters(GenDisMixClassifierParameterSet params,
                                                 boolean testBurnIn,
                                                 int minBurnInSteps)
                                                   throws Exception

Returns a standard, i.e., non-sampling, GenDisMixClassifier, where the parameters are set to the mean values over all sampled parameter values in the stationary phase.

Parameters:

params - the external parameters of the GenDisMixClassifier

testBurnIn - if burn-in phase is tested, otherwise parameters starting from index minBurnInSteps are considered

minBurnInSteps - the minimum number of steps before the stationary phase

Returns:

the GenDisMixClassifier with set parameter values

Throws:

Exception - if the GenDisMixClassifier could not be created