jsat.classifiers.linear.kernelized

Class ALMA2K

java.lang.Object

jsat.classifiers.BaseUpdateableClassifier

jsat.classifiers.linear.kernelized.ALMA2K

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, BinaryScoreClassifier, Classifier, UpdateableClassifier, Parameterized

public class ALMA2K
extends BaseUpdateableClassifier
implements BinaryScoreClassifier, Parameterized

Provides a kernelized version of the ALMA2 algorithm. It is important to note that the number of "support vectors" ALMA may learn is unbounded.
The averaged output of all previous hyperplanes is supported at almost no overhead, and can be turned on by setting setAveraged(boolean). This information is always collected, and the output can be changed once already learned.

See: Gentile, C. (2002). A New Approximate Maximal Margin Classification Algorithm. The Journal of Machine Learning Research, 2, 213–242. Retrieved from here

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
ALMA2K(KernelTrick kernel, double alpha) Creates a new kernelized ALMA2 object

Method Summary

Modifier and Type	Method and Description
CategoricalResults	classify(DataPoint data) Performs classification on the given data point.
ALMA2K	clone()
double	getAlpha() Returns the approximation coefficient used
double	getB() Returns the B value of the ALMA algorithm
double	getC()
KernelTrick	getKernelTrick() Returns the kernel in use
double	getScore(DataPoint dp) Returns the numeric score for predicting a class of a given data point, where the sign of the value indicates which class the data point is predicted to belong to.
static Distribution	guessAlpha(DataSet d) Guesses the distribution to use for the α parameter
boolean	isAveraged() Returns whether or not the averaged or last hypothesis is used
void	setAlpha(double alpha) Alpha controls the approximation of the large margin formed by ALMA, with larger values causing more updates.
void	setAveraged(boolean averaged) ALMA2K supports taking the averaged output of all previous hypothesis weighted by the number of successful uses of the hypothesis during training.
void	setB(double B) Sets the B variable of the ALMA algorithm, this is set automatically by setAlpha(double).
void	setC(double C) Sets the C value of the ALMA algorithm.
void	setKernelTrick(KernelTrick K) Sets the kernel to use
void	setUp(CategoricalData[] categoricalAttributes, int numericAttributes, CategoricalData predicting) Prepares the classifier to begin learning from its UpdateableClassifier.update(jsat.classifiers.DataPoint, int) method.
boolean	supportsWeightedData() Indicates whether the model knows how to train using weighted data points.
void	update(DataPoint dataPoint, int targetClass) Updates the classifier by giving it a new data point to learn from.

Methods inherited from class jsat.classifiers.BaseUpdateableClassifier

getEpochs, setEpochs, train, train, trainEpochs

Methods inherited from class java.lang.Object

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

Methods inherited from interface jsat.classifiers.Classifier

train, train

Methods inherited from interface jsat.parameters.Parameterized

getParameter, getParameters

Constructor Detail

ALMA2K

public ALMA2K(KernelTrick kernel,
              double alpha)

Creates a new kernelized ALMA2 object

Parameters:

kernel - the kernel function to use

alpha - the alpha parameter of ALMA

Method Detail

clone

public ALMA2K clone()

Specified by:

clone in interface BinaryScoreClassifier

Specified by:

clone in interface Classifier

Specified by:

clone in interface UpdateableClassifier

Specified by:

clone in class BaseUpdateableClassifier

setAveraged

public void setAveraged(boolean averaged)

ALMA2K supports taking the averaged output of all previous hypothesis weighted by the number of successful uses of the hypothesis during training. This effectively reduces the variance of the classifier. It has no impact on the training / update phase, only the classification results are impacted.

Unlike most algorithms, this can be changed at any time without issue - even after the algorithm has been trained the type of output (averaged or last) can be switched on the fly.

Parameters:

averaged - true to use the averaged out, false to only use the last hypothesis

isAveraged

public boolean isAveraged()

Returns whether or not the averaged or last hypothesis is used

Returns:

whether or not the averaged or last hypothesis is used

setKernelTrick

public void setKernelTrick(KernelTrick K)

Sets the kernel to use

Parameters:

K - the kernel to use

getKernelTrick

public KernelTrick getKernelTrick()

Returns the kernel in use

Returns:

the kernel in use

setAlpha

public void setAlpha(double alpha)

Alpha controls the approximation of the large margin formed by ALMA, with larger values causing more updates. A value of 1.0 will update only on mistakes, while smaller values update if the error was not far enough away from the margin.

NOTE: Whenever alpha is set, the value of B will also be set to an appropriate value. This is not the only possible value that will lead to convergence, and can be set manually after alpha is set to another value.

Parameters:

alpha - the approximation scale in (0.0, 1.0]

getAlpha

public double getAlpha()

Returns the approximation coefficient used

Returns:

the approximation coefficient used

setB

public void setB(double B)

Sets the B variable of the ALMA algorithm, this is set automatically by setAlpha(double).

Parameters:

B - the value for B

getB

public double getB()

Returns the B value of the ALMA algorithm

Returns:

the B value of the ALMA algorithm

setC

public void setC(double C)

Sets the C value of the ALMA algorithm. The default value is the one suggested in the paper.

Parameters:

C - the C value of ALMA

getC

public double getC()

setUp

public void setUp(CategoricalData[] categoricalAttributes,
                  int numericAttributes,
                  CategoricalData predicting)

Description copied from interface: UpdateableClassifier

Prepares the classifier to begin learning from its UpdateableClassifier.update(jsat.classifiers.DataPoint, int) method.

Specified by:

setUp in interface UpdateableClassifier

Parameters:

categoricalAttributes - an array containing the categorical attributes that will be in each data point

numericAttributes - the number of numeric attributes that will be in each data point

predicting - the information for the target class that will be predicted

update

public void update(DataPoint dataPoint,
                   int targetClass)

Description copied from interface: UpdateableClassifier

Updates the classifier by giving it a new data point to learn from.

Specified by:

update in interface UpdateableClassifier

Parameters:

dataPoint - the data point to learn

targetClass - the target class of the data point

classify

public CategoricalResults classify(DataPoint data)

Description copied from interface: Classifier

Performs classification on the given data point.

Specified by:

classify in interface Classifier

Parameters:

data - the data point to classify

Returns:

the results of the classification.

getScore

public double getScore(DataPoint dp)

Description copied from interface: BinaryScoreClassifier

Returns the numeric score for predicting a class of a given data point, where the sign of the value indicates which class the data point is predicted to belong to.

Specified by:

getScore in interface BinaryScoreClassifier

Parameters:

dp - the data point to predict the class label of

Returns:

the score for the given data point

supportsWeightedData

public boolean supportsWeightedData()

Description copied from interface: Classifier

Indicates whether the model knows how to train using weighted data points. If it does, the model will train assuming the weights. The values returned by this method may change depending on the parameters set for the model.

Specified by:

supportsWeightedData in interface Classifier

Returns:

true if the model supports weighted data, false otherwise

guessAlpha

public static Distribution guessAlpha(DataSet d)

Guesses the distribution to use for the α parameter

Parameters:

d - the dataset to get the guess for

Returns:

the guess for the α parameter

See Also:

setAlpha(double)