org.apache.ignite.ml.recommendation.util

Class MatrixFactorizationGradient<O extends Serializable,S extends Serializable>

java.lang.Object

org.apache.ignite.ml.recommendation.util.MatrixFactorizationGradient<O,S>

Type Parameters:

O - Type of an object of recommendation.

S - Type of a subject of recommendation.

All Implemented Interfaces:

Serializable

public class MatrixFactorizationGradient<O extends Serializable,S extends Serializable>
extends Object
implements Serializable

Gradient of matrix factorization loss function.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
MatrixFactorizationGradient(Map<O,Vector> objGrad, Map<S,Vector> subjGrad, int rows) Constructs a new instance of matrix factorization gradient.

Method Summary

Modifier and Type	Method and Description
void	applyGradient(Map<O,Vector> objMatrix, Map<S,Vector> subjMatrix) Applies given gradient to recommendation model (object matrix and subject matrix) and updates this model correspondingly.
Map<O,Vector>	getObjGrad() Returns gradient of object of recommendation matrix (unmodifiable).
int	getRows() Returns number of rows the gradient calculated on.
Map<S,Vector>	getSubjGrad() Returns gradient of subject of recommendation function (unmodifiable).

Methods inherited from class java.lang.Object

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

Constructor Detail

MatrixFactorizationGradient

public MatrixFactorizationGradient(Map<O,Vector> objGrad,
                                   Map<S,Vector> subjGrad,
                                   int rows)

Constructs a new instance of matrix factorization gradient.

Parameters:

objGrad - Gradient of object of recommendation matrix.

subjGrad - Gradient of subject of recommendation matrix.

rows - Number of rows the gradient calculated on.

Method Detail

applyGradient

public void applyGradient(Map<O,Vector> objMatrix,
                          Map<S,Vector> subjMatrix)

Applies given gradient to recommendation model (object matrix and subject matrix) and updates this model correspondingly.

Parameters:

objMatrix - Object of recommendation matrix.

subjMatrix - Subject of recommendation matrix.

getObjGrad

public Map<O,Vector> getObjGrad()

Returns gradient of object of recommendation matrix (unmodifiable).

Returns:

Gradient of object of recommendation matrix.

getSubjGrad

public Map<S,Vector> getSubjGrad()

Returns gradient of subject of recommendation function (unmodifiable).

Returns:

Gradient of subject of recommendation function.

getRows

public int getRows()

Returns number of rows the gradient calculated on.

Returns:

Number of rows the gradient calculated on.