edu.udo.cs.yale.operator.learner
Interface Learner

All Known Implementing Classes:

AbstractLearner, AbstractMetaLearner, AbstractMySVMLearner, AdaBoost, AverageLearner, Bagging, BayBoostStream, BayesianBoosting, BestRuleInduction, ClassificationByRegression, DefaultLearner, EvoSVM, GenericWekaEnsembleLearner, GenericWekaLearner, GenericWekaMetaLearner, GPLearner, IteratingGSS, JMySVMLearner, LibSVMLearner, MultiCriterionDecisionStumps, MyKLRLearner, NaiveBayes, PSOSVM, RVMLearner

public interface Learner

A Learner is an operator that encapsulates the learning step of a machine learning method. Some Learners may be capable of estimating the performance of the generated model. In that case, they additionally return a PerformanceVector. Furthermore some learner can calculate weights of the used attributes which can also be delivered.

Version:

$Id: Learner.java,v 2.18 2006/08/03 14:39:34 ingomierswa Exp $

Author:

Ingo Mierswa

Method Summary

PerformanceVector	getEstimatedPerformance() Most learners should throw an exception if they are not able to estimate the learning performance.
java.lang.String	getName() Returns the name of the learner.
AttributeWeights	getWeights(ExampleSet eSet) Most learners should throw an exception if they are not able to calculate attribute weights.
Model	learn(ExampleSet exampleSet) Trains a model.
boolean	shouldCalculateWeights() Most learners will return false since they are not able to calculate attribute weights.
boolean	shouldEstimatePerformance() Most learners will return false since they are not able to estimate the learning performance.
boolean	supportsCapability(LearnerCapability capability) Checks for Learner capabilities.

Method Detail

learn

Model learn(ExampleSet exampleSet)
            throws OperatorException

Trains a model. This method should be called by apply() and is implemented by subclasses.

Throws:

OperatorException

getName

java.lang.String getName()

Returns the name of the learner.

supportsCapability

boolean supportsCapability(LearnerCapability capability)

Checks for Learner capabilities. Should return true if the given capability is supported.

shouldEstimatePerformance

boolean shouldEstimatePerformance()

Most learners will return false since they are not able to estimate the learning performance. However, if a learning scheme is able to calculate the performance (e.g. Xi-Alpha estimation of a SVM) it should return true.

getEstimatedPerformance

PerformanceVector getEstimatedPerformance()
                                          throws OperatorException

Most learners should throw an exception if they are not able to estimate the learning performance. However, if a learning scheme is able to calculate the performance (e.g. Xi-Alpha estimation of a SVM) it should return a performance vector containing the estimated performance.

Throws:

OperatorException

shouldCalculateWeights

boolean shouldCalculateWeights()

Most learners will return false since they are not able to calculate attribute weights. However, if a learning scheme is able to calculate weights (e.g. the normal vector of a SVM) it should return true.

getWeights

AttributeWeights getWeights(ExampleSet eSet)
                            throws OperatorException

Most learners should throw an exception if they are not able to calculate attribute weights. However, if a learning scheme is able to calculate weights (e.g. the normal vector of a SVM) it should return an AttributeWeights object.

Throws:

OperatorException