Package org.tribuo.classification.ensemble

Class VotingCombiner

java.lang.Object

org.tribuo.classification.ensemble.VotingCombiner

All Implemented Interfaces:

com.oracle.labs.mlrg.olcut.config.Configurable, com.oracle.labs.mlrg.olcut.provenance.Provenancable<com.oracle.labs.mlrg.olcut.provenance.ConfiguredObjectProvenance>, Serializable, EnsembleCombiner<Label>

public final class VotingCombiner
extends Object
implements EnsembleCombiner<Label>

A combiner which performs a weighted or unweighted vote across the predicted labels.

This uses the most likely prediction from each ensemble member, unlike FullyWeightedVotingCombiner which uses the full distribution of predictions for each ensemble member.

See Also:

• Serialized Form

Constructor Summary

Constructors

Constructor	Description
VotingCombiner()	Constructs a voting combiner.

Method Summary

Modifier and Type	Method	Description
Prediction<Label>	combine(ImmutableOutputInfo<Label> outputInfo, List<Prediction<Label>> predictions)	Combine the predictions.
Prediction<Label>	combine(ImmutableOutputInfo<Label> outputInfo, List<Prediction<Label>> predictions, float[] weights)	Combine the supplied predictions.
ONNXNode	exportCombiner(ONNXNode input)	Exports this voting combiner to ONNX.
<T extends ONNXRef<?>> ONNXNode	exportCombiner(ONNXNode input, T weight)	Exports this voting combiner to ONNX
com.oracle.labs.mlrg.olcut.provenance.ConfiguredObjectProvenance	getProvenance()
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 Details

VotingCombiner

public VotingCombiner()

Constructs a voting combiner.

Method Details

combine

public Prediction<Label> combine(ImmutableOutputInfo<Label> outputInfo,
 List<Prediction<Label>> predictions)

Description copied from interface: EnsembleCombiner

Combine the predictions.

Specified by:

combine in interface EnsembleCombiner<Label>

Parameters:

outputInfo - The output domain.

predictions - The predictions to combine.

Returns:

The ensemble prediction.

combine

public Prediction<Label> combine(ImmutableOutputInfo<Label> outputInfo,
 List<Prediction<Label>> predictions,
 float[] weights)

Description copied from interface: EnsembleCombiner

Combine the supplied predictions. predictions.size() must equal weights.length.

Specified by:

combine in interface EnsembleCombiner<Label>

Parameters:

outputInfo - The output domain.

predictions - The predictions to combine.

weights - The weights to use for each prediction.

Returns:

The ensemble prediction.

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>

exportCombiner

public ONNXNode exportCombiner(ONNXNode input)

Exports this voting combiner to ONNX.

The input should be a 3-tensor [batch_size, num_outputs, num_ensemble_members].

Specified by:

exportCombiner in interface EnsembleCombiner<Label>

Parameters:

input - The input tensor to combine.

Returns:

the final node proto representing the voting operation.

exportCombiner

public <T extends ONNXRef<?>> ONNXNode exportCombiner(ONNXNode input,
 T weight)

Exports this voting combiner to ONNX

The input should be a 3-tensor [batch_size, num_outputs, num_ensemble_members].

Specified by:

exportCombiner in interface EnsembleCombiner<Label>

Type Parameters:

T - The type of the weights input reference.

Parameters:

input - The input tensor to combine.

weight - The combination weight node.

Returns:

the final node proto representing the voting operation.