org.apache.ignite.ml.composition.stacking

Class SimpleStackedDatasetTrainer<I,O,AM extends IgniteModel<I,O>,L>

java.lang.Object

org.apache.ignite.ml.trainers.DatasetTrainer<StackedModel<IS,IA,O,AM>,L>

org.apache.ignite.ml.composition.stacking.StackedDatasetTrainer<I,I,O,AM,L>

org.apache.ignite.ml.composition.stacking.SimpleStackedDatasetTrainer<I,O,AM,L>

Type Parameters:

I - Type of submodels input.

O - Type of aggregator model output.

AM - Type of aggregator model.

L - Type of labels.

Direct Known Subclasses:

StackedVectorDatasetTrainer

public class SimpleStackedDatasetTrainer<I,O,AM extends IgniteModel<I,O>,L>
extends StackedDatasetTrainer<I,I,O,AM,L>

DatasetTrainer with same type of input and output of submodels.

Nested Class Summary

Nested classes/interfaces inherited from class org.apache.ignite.ml.trainers.DatasetTrainer

DatasetTrainer.EmptyDatasetException

Field Summary

Fields inherited from class org.apache.ignite.ml.trainers.DatasetTrainer

envBuilder, environment

Constructor Summary

Constructors

Constructor and Description
SimpleStackedDatasetTrainer() Constructs instance of this class.
SimpleStackedDatasetTrainer(DatasetTrainer<AM,L> aggregatingTrainer, IgniteBinaryOperator<I> aggregatingInputMerger) Construct instance of this class.
SimpleStackedDatasetTrainer(DatasetTrainer<AM,L> aggregatingTrainer, IgniteBinaryOperator<I> aggregatingInputMerger, IgniteFunction<I,I> submodelInput2AggregatingInputConverter, IgniteFunction<Vector,I> vector2SubmodelInputConverter, IgniteFunction<I,Vector> submodelOutput2VectorConverter) Construct instance of this class.

Method Summary

Modifier and Type	Method and Description
<M1 extends IgniteModel<I,I>> SimpleStackedDatasetTrainer<I,O,AM,L>	addTrainer(DatasetTrainer<M1,L> trainer) Adds submodel trainer along with converters needed on training and inference stages.
SimpleStackedDatasetTrainer<I,O,AM,L>	withAggregatorInputMerger(IgniteBinaryOperator<I> merger) Specify binary operator used to merge submodels outputs to one.
SimpleStackedDatasetTrainer<I,O,AM,L>	withAggregatorTrainer(DatasetTrainer<AM,L> aggregatorTrainer) Specify aggregator trainer.
<L1> SimpleStackedDatasetTrainer<I,O,AM,L1>	withConvertedLabels(IgniteFunction<L1,L> new2Old) Creates DatasetTrainer with same training logic, but able to accept labels of given new type of labels.
SimpleStackedDatasetTrainer<I,O,AM,L>	withEnvironmentBuilder(LearningEnvironmentBuilder envBuilder) Changes learning Environment.
SimpleStackedDatasetTrainer<I,O,AM,L>	withOriginalFeaturesDropped() Drop original features during training and inference.
SimpleStackedDatasetTrainer<I,O,AM,L>	withOriginalFeaturesKept() Keep original features using IgniteFunction.identity() as submodelInput2AggregatingInputConverter.
SimpleStackedDatasetTrainer<I,O,AM,L>	withOriginalFeaturesKept(IgniteFunction<I,I> submodelInput2AggregatingInputConverter) Keep original features during training and propagate submodels input to aggregator during inference using given function.

Methods inherited from class org.apache.ignite.ml.composition.stacking.StackedDatasetTrainer

fitWithInitializedDeployingContext, isUpdateable, update, updateModel, withSubmodelOutput2VectorConverter, withVector2SubmodelInputConverter

Methods inherited from class org.apache.ignite.ml.trainers.DatasetTrainer

fit, fit, fit, fit, fit, fit, getLastTrainedModelOrThrowEmptyDatasetException, identityTrainer, learningEnvironment, update, update, update, update

Methods inherited from class java.lang.Object

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

Constructor Detail

SimpleStackedDatasetTrainer

public SimpleStackedDatasetTrainer(DatasetTrainer<AM,L> aggregatingTrainer,
                                   IgniteBinaryOperator<I> aggregatingInputMerger,
                                   IgniteFunction<I,I> submodelInput2AggregatingInputConverter,
                                   IgniteFunction<Vector,I> vector2SubmodelInputConverter,
                                   IgniteFunction<I,Vector> submodelOutput2VectorConverter)

Construct instance of this class.

Parameters:

aggregatingTrainer - Aggregator trainer.

aggregatingInputMerger - Function used to merge submodels outputs into one.

submodelInput2AggregatingInputConverter - Function used to convert input of submodel to output of submodel this function is used if user chooses to keep original features.

SimpleStackedDatasetTrainer

public SimpleStackedDatasetTrainer(DatasetTrainer<AM,L> aggregatingTrainer,
                                   IgniteBinaryOperator<I> aggregatingInputMerger)

Construct instance of this class.

Parameters:

aggregatingTrainer - Aggregator trainer.

aggregatingInputMerger - Function used to merge submodels outputs into one.

SimpleStackedDatasetTrainer

public SimpleStackedDatasetTrainer()

Constructs instance of this class.

Method Detail

addTrainer

public <M1 extends IgniteModel<I,I>> SimpleStackedDatasetTrainer<I,O,AM,L> addTrainer(DatasetTrainer<M1,L> trainer)

Adds submodel trainer along with converters needed on training and inference stages.

Overrides:

addTrainer in class StackedDatasetTrainer<I,I,O,AM extends IgniteModel<I,O>,L>

Parameters:

trainer - Submodel trainer.

Returns:

This object.

withAggregatorTrainer

public SimpleStackedDatasetTrainer<I,O,AM,L> withAggregatorTrainer(DatasetTrainer<AM,L> aggregatorTrainer)

Specify aggregator trainer.

Overrides:

withAggregatorTrainer in class StackedDatasetTrainer<I,I,O,AM extends IgniteModel<I,O>,L>

Parameters:

aggregatorTrainer - Aggregator trainer.

Returns:

This object.

withOriginalFeaturesDropped

public SimpleStackedDatasetTrainer<I,O,AM,L> withOriginalFeaturesDropped()

Drop original features during training and inference.

Overrides:

withOriginalFeaturesDropped in class StackedDatasetTrainer<I,I,O,AM extends IgniteModel<I,O>,L>

Returns:

This object.

withOriginalFeaturesKept

public SimpleStackedDatasetTrainer<I,O,AM,L> withOriginalFeaturesKept(IgniteFunction<I,I> submodelInput2AggregatingInputConverter)

Keep original features during training and propagate submodels input to aggregator during inference using given function. Note that if this object is on, training will be done on vector obtaining from concatenating features passed to submodels trainers and outputs of submodels converted to vectors, this can, for example influence aggregator model input vector dimension (if IS = Vector), or, more generally, some IS parameters which are not reflected just by its type. So converter should be written accordingly.

Overrides:

withOriginalFeaturesKept in class StackedDatasetTrainer<I,I,O,AM extends IgniteModel<I,O>,L>

Parameters:

submodelInput2AggregatingInputConverter - Function used to propagate submodels input to aggregator.

Returns:

This object.

withAggregatorInputMerger

public SimpleStackedDatasetTrainer<I,O,AM,L> withAggregatorInputMerger(IgniteBinaryOperator<I> merger)

Specify binary operator used to merge submodels outputs to one.

Overrides:

withAggregatorInputMerger in class StackedDatasetTrainer<I,I,O,AM extends IgniteModel<I,O>,L>

Parameters:

merger - Binary operator used to merge submodels outputs to one.

Returns:

This object.

withEnvironmentBuilder

public SimpleStackedDatasetTrainer<I,O,AM,L> withEnvironmentBuilder(LearningEnvironmentBuilder envBuilder)

Changes learning Environment.

Overrides:

withEnvironmentBuilder in class StackedDatasetTrainer<I,I,O,AM extends IgniteModel<I,O>,L>

Parameters:

envBuilder - Learning environment builder.

withConvertedLabels

public <L1> SimpleStackedDatasetTrainer<I,O,AM,L1> withConvertedLabels(IgniteFunction<L1,L> new2Old)

Creates DatasetTrainer with same training logic, but able to accept labels of given new type of labels.

Overrides:

withConvertedLabels in class DatasetTrainer<StackedModel<I,I,O,AM extends IgniteModel<I,O>>,L>

Type Parameters:

L1 - New labels type.

Parameters:

new2Old - Converter of new labels to old labels.

Returns:

DatasetTrainer with same training logic, but able to accept labels of given new type of labels.

withOriginalFeaturesKept

public SimpleStackedDatasetTrainer<I,O,AM,L> withOriginalFeaturesKept()

Keep original features using IgniteFunction.identity() as submodelInput2AggregatingInputConverter.

Returns:

This object.