es.uam.eps.ir.ranksys.mf.als

Class ALSFactorizer<U,I>

java.lang.Object

es.uam.eps.ir.ranksys.mf.Factorizer<U,I>

es.uam.eps.ir.ranksys.mf.als.ALSFactorizer<U,I>

Type Parameters:

U - type of the users

I - type of the items

Direct Known Subclasses:

HKVFactorizer, PZTFactorizer

public abstract class ALSFactorizer<U,I>
extends Factorizer<U,I>

Generic alternating least-squares factorizer.

Author:

Saúl Vargas (saul.vargas@uam.es)

Constructor Summary

Constructors

Constructor and Description
ALSFactorizer(int numIter) Constructor.

Method Summary

Modifier and Type	Method and Description
protected abstract double	error(cern.colt.matrix.impl.DenseDoubleMatrix2D p, cern.colt.matrix.impl.DenseDoubleMatrix2D q, FastPreferenceData<U,I> data) Squared loss of two matrices.
double	error(Factorization<U,I> factorization, FastPreferenceData<U,I> data) Global loss of the factorization.
void	factorize(Factorization<U,I> factorization, FastPreferenceData<U,I> data) Calculates the factorization by using a previously generate matrix factorization.
Factorization<U,I>	factorize(int K, FastPreferenceData<U,I> data) Creates and calculates a factorization.
protected abstract void	set_minP(cern.colt.matrix.impl.DenseDoubleMatrix2D p, cern.colt.matrix.impl.DenseDoubleMatrix2D q, FastPreferenceData<U,I> data) User matrix least-squares step.
protected abstract void	set_minQ(cern.colt.matrix.impl.DenseDoubleMatrix2D q, cern.colt.matrix.impl.DenseDoubleMatrix2D p, FastPreferenceData<U,I> data) Item matrix least-squares step.

Methods inherited from class java.lang.Object

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

Constructor Detail

ALSFactorizer

public ALSFactorizer(int numIter)

Constructor.

Parameters:

numIter - number of least-squares calculations

Method Detail

error

public double error(Factorization<U,I> factorization,
                    FastPreferenceData<U,I> data)

Description copied from class: Factorizer

Global loss of the factorization.

Specified by:

error in class Factorizer<U,I>

Parameters:

factorization - matrix factorization

data - preference data

Returns:

the global loss

factorize

public Factorization<U,I> factorize(int K,
                                    FastPreferenceData<U,I> data)

Description copied from class: Factorizer

Creates and calculates a factorization.

Specified by:

factorize in class Factorizer<U,I>

Parameters:

K - size of the latent feature space.

data - preference data

Returns:

a matrix factorization

factorize

public void factorize(Factorization<U,I> factorization,
                      FastPreferenceData<U,I> data)

Description copied from class: Factorizer

Calculates the factorization by using a previously generate matrix factorization.

Specified by:

factorize in class Factorizer<U,I>

Parameters:

factorization - matrix factorization

data - preference data

error

protected abstract double error(cern.colt.matrix.impl.DenseDoubleMatrix2D p,
                                cern.colt.matrix.impl.DenseDoubleMatrix2D q,
                                FastPreferenceData<U,I> data)

Squared loss of two matrices.

Parameters:

p - user matrix

q - item matrix

data - preference data

Returns:

squared loss

set_minP

protected abstract void set_minP(cern.colt.matrix.impl.DenseDoubleMatrix2D p,
                                 cern.colt.matrix.impl.DenseDoubleMatrix2D q,
                                 FastPreferenceData<U,I> data)

User matrix least-squares step.

Parameters:

p - user matrix

q - item matrix

data - preference data

set_minQ

protected abstract void set_minQ(cern.colt.matrix.impl.DenseDoubleMatrix2D q,
                                 cern.colt.matrix.impl.DenseDoubleMatrix2D p,
                                 FastPreferenceData<U,I> data)

Item matrix least-squares step.

Parameters:

q - item matrix

p - user matrix

data - preference data