org.openimaj.ml.pca

Class FeatureVectorPCA

java.lang.Object

org.openimaj.math.matrix.algorithm.pca.PrincipalComponentAnalysis

org.openimaj.ml.pca.FeatureVectorPCA

public class FeatureVectorPCA
extends PrincipalComponentAnalysis

Principal Components Analysis wrapper for FeatureVectors.

Author:

Jonathon Hare (jsh2@ecs.soton.ac.uk)

Nested Class Summary

Nested classes/interfaces inherited from class org.openimaj.math.matrix.algorithm.pca.PrincipalComponentAnalysis

PrincipalComponentAnalysis.ComponentSelector, PrincipalComponentAnalysis.EnergyThresholdComponentSelector, PrincipalComponentAnalysis.NumberComponentSelector, PrincipalComponentAnalysis.PercentageEnergyComponentSelector

Field Summary

Fields inherited from class org.openimaj.math.matrix.algorithm.pca.PrincipalComponentAnalysis

basis, eigenvalues, mean

Constructor Summary

Constructors

Constructor and Description
FeatureVectorPCA() Default constructor, using an SvdPrincipalComponentAnalysis.
FeatureVectorPCA(PrincipalComponentAnalysis inner) Construct with the given PrincipalComponentAnalysis object.

Method Summary

Methods

Modifier and Type	Method and Description
DoubleFV	generate(DoubleFV scalings) Generate a new "observation" as a linear combination of the principal components (PC): mean + PC * scaling.
void	learnBasis(Collection<? extends FeatureVector> data) Learn the PCA basis of the given feature vectors.
void	learnBasis(double[][] data) Learn the principal components of the given data array.
void	learnBasis(FeatureVector[] data) Learn the PCA basis of the given feature vectors.
protected void	learnBasisNorm(Jama.Matrix norm) Learn the PCA basis from the centered data provided.
DoubleFV	project(FeatureVector vector) Project a vector by the basis.

Methods inherited from class org.openimaj.math.matrix.algorithm.pca.PrincipalComponentAnalysis

buildNormalisedDataMatrix, generate, getBasis, getCumulativeEnergies, getCumulativeEnergy, getEigenValue, getEigenValues, getEigenVectors, getMean, getPrincipalComponent, getStandardDeviations, getStandardDeviations, learnBasis, learnBasis, project, project, selectSubset, selectSubset, selectSubsetEnergyThreshold, selectSubsetPercentageEnergy, toString

Methods inherited from class java.lang.Object

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

Constructor Detail

FeatureVectorPCA

public FeatureVectorPCA()

Default constructor, using an SvdPrincipalComponentAnalysis.

FeatureVectorPCA

public FeatureVectorPCA(PrincipalComponentAnalysis inner)

Construct with the given PrincipalComponentAnalysis object.

Parameters:

inner - PCA algorithm.

Method Detail

learnBasis

public void learnBasis(FeatureVector[] data)

Learn the PCA basis of the given feature vectors.

Parameters:

data - the feature vectors to apply PCA to.

learnBasis

public void learnBasis(Collection<? extends FeatureVector> data)

Learn the PCA basis of the given feature vectors.

Parameters:

data - the feature vectors to apply PCA to.

project

public DoubleFV project(FeatureVector vector)

Project a vector by the basis. The vector is normalised by subtracting the mean and then multiplied by the basis.

Parameters:

vector - the vector to project

Returns:

projected vector

learnBasis

public void learnBasis(double[][] data)

Description copied from class: PrincipalComponentAnalysis

Learn the principal components of the given data array. Each row corresponds to an observation with the number of dimensions equal to the number of columns.

Overrides:

learnBasis in class PrincipalComponentAnalysis

Parameters:

data - the data

learnBasisNorm

protected void learnBasisNorm(Jama.Matrix norm)

Description copied from class: PrincipalComponentAnalysis

Learn the PCA basis from the centered data provided. Each row corresponds to an observation with the number of dimensions equal to the number of columns.

Specified by:

learnBasisNorm in class PrincipalComponentAnalysis

generate

public DoubleFV generate(DoubleFV scalings)

Generate a new "observation" as a linear combination of the principal components (PC): mean + PC * scaling. If the scaling vector is shorter than the number of components, it will be zero-padded. If it is longer, it will be truncated.

Parameters:

scalings - the weighting for each PC

Returns:

generated observation