ai.libs.mlplan.core

Class AbstractMLPlanSingleLabelBuilder

java.lang.Object

ai.libs.mlplan.core.AbstractMLPlanBuilder

ai.libs.mlplan.core.AbstractMLPlanSingleLabelBuilder

All Implemented Interfaces:

ai.libs.jaicore.basic.ILoggingCustomizable, IMLPlanBuilder

Direct Known Subclasses:

MLPlanSKLearnBuilder, MLPlanWekaBuilder

public abstract class AbstractMLPlanSingleLabelBuilder
extends AbstractMLPlanBuilder

Field Summary

Fields

Modifier and Type	Field and Description
protected static ai.libs.jaicore.ml.core.evaluation.measure.IMeasure<java.lang.Double,java.lang.Double>	LOSS_FUNCTION
protected static int	SEARCH_NUM_MC_ITERATIONS
protected static double	SEARCH_TRAIN_FOLD_SIZE
protected static int	SELECTION_NUM_MC_ITERATIONS
protected static double	SELECTION_TRAIN_FOLD_SIZE

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	AbstractMLPlanSingleLabelBuilder()

Method Summary

Modifier and Type	Method and Description
protected ai.libs.jaicore.ml.weka.dataset.splitter.IDatasetSplitter	getDefaultDatasetSplitter()
AbstractMLPlanSingleLabelBuilder	withMonteCarloCrossValidationInSearchPhase(int numIterations, double trainFoldSize, ai.libs.jaicore.ml.core.evaluation.measure.IMeasure<java.lang.Double,java.lang.Double> lossFunction) Configure ML-Plan to use MCCV for the given number of iterations, train fold size and loss function in the search phase.
AbstractMLPlanSingleLabelBuilder	withMonteCarloCrossValidationInSelectionPhase(int numIterations, double trainFoldSize, ai.libs.jaicore.ml.core.evaluation.measure.IMeasure<java.lang.Double,java.lang.Double> lossFunction) Configure ML-Plan to use MCCV for the given number of iterations, train fold size and loss function in the selection phase.
AbstractMLPlanSingleLabelBuilder	withPerformanceMeasure(ai.libs.jaicore.ml.core.evaluation.measure.IMeasure<java.lang.Double,java.lang.Double> lossFunction) Sets the performance measure to evaluate a candidate solution's generalization performance.

Methods inherited from class ai.libs.mlplan.core.AbstractMLPlanBuilder

build, build, forMeka, forSKLearn, forWeka, getAlgorithmConfig, getCandidateEvaluationTimeOut, getClassifierEvaluationInSearchPhase, getClassifierEvaluationInSelectionPhase, getClassifierFactory, getComponents, getDBAdapter, getHASCOFactory, getLoggerName, getNodeEvaluationTimeOut, getPerformanceMeasureName, getRequestedInterface, getSearchEvaluatorFactory, getSearchSelectionDatasetSplitter, getSearchSpaceConfigFile, getSelectionEvaluatorFactory, getTimeOut, getUseCache, prepareNodeEvaluatorInFactoryWithData, setLoggerName, setPerformanceMeasureName, withAlgorithmConfig, withAlgorithmConfigFile, withCandidateEvaluationTimeOut, withClassifierFactory, withDataset, withDatasetSplitterForSearchSelectionSplit, withNodeEvaluationTimeOut, withNumCpus, withPreferredComponentsFile, withPreferredNodeEvaluator, withRandomCompletionBasedBestFirstSearch, withRequestedInterface, withSearchFactory, withSearchPhaseEvaluatorFactory, withSearchSpaceConfigFile, withSelectionPhaseEvaluatorFactory, withTimeOut

Methods inherited from class java.lang.Object

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

Field Detail

SEARCH_NUM_MC_ITERATIONS

protected static final int SEARCH_NUM_MC_ITERATIONS

See Also:

Constant Field Values

SEARCH_TRAIN_FOLD_SIZE

protected static final double SEARCH_TRAIN_FOLD_SIZE

See Also:

Constant Field Values

SELECTION_NUM_MC_ITERATIONS

protected static final int SELECTION_NUM_MC_ITERATIONS

See Also:

Constant Field Values

SELECTION_TRAIN_FOLD_SIZE

protected static final double SELECTION_TRAIN_FOLD_SIZE

See Also:

Constant Field Values

LOSS_FUNCTION

protected static final ai.libs.jaicore.ml.core.evaluation.measure.IMeasure<java.lang.Double,java.lang.Double> LOSS_FUNCTION

Constructor Detail

AbstractMLPlanSingleLabelBuilder

protected AbstractMLPlanSingleLabelBuilder()

Method Detail

withMonteCarloCrossValidationInSearchPhase

public AbstractMLPlanSingleLabelBuilder withMonteCarloCrossValidationInSearchPhase(int numIterations,
                                                                                   double trainFoldSize,
                                                                                   ai.libs.jaicore.ml.core.evaluation.measure.IMeasure<java.lang.Double,java.lang.Double> lossFunction)

Configure ML-Plan to use MCCV for the given number of iterations, train fold size and loss function in the search phase.

Parameters:

numIterations - The number of iterations of the MCCV.

trainFoldSize - The portion of the data that is to be used as training data.

lossFunction - The loss function to evaluate the performance of the classifier.

Returns:

The builder object.

withMonteCarloCrossValidationInSelectionPhase

public AbstractMLPlanSingleLabelBuilder withMonteCarloCrossValidationInSelectionPhase(int numIterations,
                                                                                      double trainFoldSize,
                                                                                      ai.libs.jaicore.ml.core.evaluation.measure.IMeasure<java.lang.Double,java.lang.Double> lossFunction)

Configure ML-Plan to use MCCV for the given number of iterations, train fold size and loss function in the selection phase.

Parameters:

numIterations - The number of iterations of the MCCV.

trainFoldSize - The portion of the data that is to be used as training data.

lossFunction - The loss function to evaluate the performance of the classifier.

Returns:

The builder object.

withPerformanceMeasure

public AbstractMLPlanSingleLabelBuilder withPerformanceMeasure(ai.libs.jaicore.ml.core.evaluation.measure.IMeasure<java.lang.Double,java.lang.Double> lossFunction)

Sets the performance measure to evaluate a candidate solution's generalization performance. Caution: This resets the evaluators to MCCV for both search and selection phase if these are not already MCCVs.

Parameters:

lossFunction - The loss function to be used.

Returns:

The builder object.

getDefaultDatasetSplitter

protected ai.libs.jaicore.ml.weka.dataset.splitter.IDatasetSplitter getDefaultDatasetSplitter()