org.tribuo.multilabel.sgd.objectives

Class Hinge

java.lang.Object

org.tribuo.multilabel.sgd.objectives.Hinge

All Implemented Interfaces:

com.oracle.labs.mlrg.olcut.config.Configurable, com.oracle.labs.mlrg.olcut.provenance.Provenancable<com.oracle.labs.mlrg.olcut.provenance.ConfiguredObjectProvenance>, SGDObjective<SGDVector>, MultiLabelObjective

public final class Hinge
extends Object
implements MultiLabelObjective

Hinge loss, scores the correct value margin and any incorrect predictions -margin. By default the margin is 1.0.

The Hinge loss does not generate a probabilistic model, and uses a NoopNormalizer.

Constructor Summary

Constructors

Constructor and Description
Hinge() Construct a hinge objective with a margin of 1.0.
Hinge(double margin) Construct a hinge objective with the supplied margin.

Method Summary

Modifier and Type	Method and Description
VectorNormalizer	getNormalizer() Returns a new NoopNormalizer.
com.oracle.labs.mlrg.olcut.provenance.ConfiguredObjectProvenance	getProvenance()
boolean	isProbabilistic() Returns false.
com.oracle.labs.mlrg.olcut.util.Pair<Double,SGDVector>	lossAndGradient(SGDVector truth, SGDVector prediction) Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.
double	threshold() The default prediction threshold for creating the output.
String	toString()

Methods inherited from class java.lang.Object

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

Methods inherited from interface com.oracle.labs.mlrg.olcut.config.Configurable

postConfig

Constructor Detail

Hinge

public Hinge(double margin)

Construct a hinge objective with the supplied margin.

Parameters:

margin - The margin to use.

Hinge

public Hinge()

Construct a hinge objective with a margin of 1.0.

Method Detail

lossAndGradient

public com.oracle.labs.mlrg.olcut.util.Pair<Double,SGDVector> lossAndGradient(SGDVector truth,
                                                                              SGDVector prediction)

Returns a Pair of Double and SGDVector representing the loss and per label gradients respectively.

Specified by:

lossAndGradient in interface SGDObjective<SGDVector>

Parameters:

truth - The true label id.

prediction - The prediction for each label id.

Returns:

The loss and per label gradient.

getNormalizer

public VectorNormalizer getNormalizer()

Returns a new NoopNormalizer.

Specified by:

getNormalizer in interface MultiLabelObjective

Returns:

The vector normalizer.

isProbabilistic

public boolean isProbabilistic()

Returns false.

Specified by:

isProbabilistic in interface MultiLabelObjective

Returns:

False.

threshold

public double threshold()

Description copied from interface: MultiLabelObjective

The default prediction threshold for creating the output.

Specified by:

threshold in interface MultiLabelObjective

Returns:

The threshold.

toString

public String toString()

Overrides:

toString in class Object

getProvenance

public com.oracle.labs.mlrg.olcut.provenance.ConfiguredObjectProvenance getProvenance()

Specified by:

getProvenance in interface com.oracle.labs.mlrg.olcut.provenance.Provenancable<com.oracle.labs.mlrg.olcut.provenance.ConfiguredObjectProvenance>