fr.lip6.jkernelmachines.projection

Class KernelPCA<T>

java.lang.Object

fr.lip6.jkernelmachines.projection.KernelPCA<T>

public class KernelPCA<T>
extends java.lang.Object

Kernel principal component analysis, using generic datatypes.

Author:

picard

Constructor Summary

Constructors

Constructor and Description
KernelPCA(Kernel<T> k)

Method Summary

Modifier and Type	Method and Description
Kernel<T>	getKernel() Get the kernel use in the KernelPCA
double	getMean() Get the mean of the current Kernel
double[][]	getProjectors() Get the projector coefficients obtained after learning
double[]	getWhiteningCoefficients() Get the whitening coefficient
TrainingSample<double[]>	project(TrainingSample<T> t)
TrainingSample<double[]>	project(TrainingSample<T> t, boolean whitening)
java.util.List<TrainingSample<double[]>>	projectList(java.util.List<TrainingSample<T>> list) Performs the projection on a list of samples.
java.util.List<TrainingSample<double[]>>	projectList(java.util.List<TrainingSample<T>> list, boolean whitening) Performs the projection on a list of samples with optional whitening (unitary covariance matrix).
void	setKernel(Kernel<T> kernel) Set the current Kernel
void	train(java.util.List<TrainingSample<T>> list)

Methods inherited from class java.lang.Object

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

Constructor Detail

KernelPCA

public KernelPCA(Kernel<T> k)

Method Detail

train

public void train(java.util.List<TrainingSample<T>> list)

project

public TrainingSample<double[]> project(TrainingSample<T> t,
                                        boolean whitening)

project

public TrainingSample<double[]> project(TrainingSample<T> t)

projectList

public java.util.List<TrainingSample<double[]>> projectList(java.util.List<TrainingSample<T>> list)

Performs the projection on a list of samples.

Parameters:

list - the list of input samples

Returns:

a new list with projected samples

projectList

public java.util.List<TrainingSample<double[]>> projectList(java.util.List<TrainingSample<T>> list,
                                                            boolean whitening)

Performs the projection on a list of samples with optional whitening (unitary covariance matrix).

Parameters:

list - the list of input samples

whitening - option to perform a whitened projection

Returns:

a new list with projected samples

getKernel

public Kernel<T> getKernel()

Get the kernel use in the KernelPCA

Returns:

the current kernel

setKernel

public void setKernel(Kernel<T> kernel)

Set the current Kernel

Parameters:

kernel - the kernel

getMean

public double getMean()

Get the mean of the current Kernel

Returns:

the mean of the kernel over the training set

getProjectors

public double[][] getProjectors()

Get the projector coefficients obtained after learning

Returns:

the kernel expansion coefficients

getWhiteningCoefficients

public double[] getWhiteningCoefficients()

Get the whitening coefficient

Returns:

the whitening coefficients