Uses of Class
de.fu_berlin.ties.classify.PredictionDistribution

Packages that use PredictionDistribution
de.fu_berlin.ties.classify This package provides functionality for classification of texts and feature vectors. 
de.fu_berlin.ties.classify.winnow This package contains the Winnow classification algorithm and related algorithms and classes. 
de.fu_berlin.ties.combi This package provides combination strategies for combining sequential classification decisions. 
de.fu_berlin.ties.demo This package contains demo code for showing how the system works. 
de.fu_berlin.ties.filter This packages provides generic filtering and rewriting functionality. 
 

Uses of PredictionDistribution in de.fu_berlin.ties.classify
 

Methods in de.fu_berlin.ties.classify that return PredictionDistribution
 PredictionDistribution Classifier.classify(FeatureVector features, Set candidateClasses)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes.
 PredictionDistribution TrainableClassifier.classify(FeatureVector features, Set candidateClasses)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes.
protected  PredictionDistribution TieClassifier.doClassify(FeatureVector features, Set candidateClasses, ContextMap context)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes.
protected  PredictionDistribution OneAgainstTheRestClassifier.doClassify(FeatureVector features, Set candidateClasses, ContextMap context)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes. This implementation combines the predictions for the positive class of all involved inner classifiers.
protected  PredictionDistribution MultiBinaryClassifier.doClassify(FeatureVector features, Set candidateClasses, ContextMap context)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes. This implementation combines the predictions for the foreground of all involved inner classifiers.
protected  PredictionDistribution MetaClassifier.doClassify(FeatureVector features, Set candidateClasses, ContextMap context)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes.
protected  PredictionDistribution ExternalClassifier.doClassify(FeatureVector features, Set candidateClasses, ContextMap context)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes.
protected  PredictionDistribution MoonClassifier.doClassify(FeatureVector features, Set candidateClasses, ContextMap context)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes.
protected abstract  PredictionDistribution TrainableClassifier.doClassify(FeatureVector features, Set candidateClasses, ContextMap context)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes.
 PredictionDistribution Reranker.rerank(PredictionDistribution origDist)
          Reranks the predictions in a distribution.
 PredictionDistribution TrainableClassifier.trainOnError(FeatureVector features, String targetClass, Set candidateClasses)
          Handles error-driven learning ("train only errors"): the specified feature vector is trained into the model only if the predicted class for the feature vector differs from the specified target class.
 

Methods in de.fu_berlin.ties.classify with parameters of type PredictionDistribution
protected  boolean TieClassifier.doTrainOnError(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          The core of the TrainableClassifier.trainOnError(FeatureVector, String, Set) method.
protected  boolean MetaClassifier.doTrainOnError(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          The core of the TrainableClassifier.trainOnError(FeatureVector, String, Set) method.
protected  boolean TrainableClassifier.doTrainOnError(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          The core of the TrainableClassifier.trainOnError(FeatureVector, String, Set) method.
 PredictionDistribution Reranker.rerank(PredictionDistribution origDist)
          Reranks the predictions in a distribution.
protected  boolean TieClassifier.shouldTrain(String targetClass, PredictionDistribution predDist, ContextMap context)
          Invoked by TrainableClassifier.trainOnError(FeatureVector, String, Set) to decide whether to train an instance.
protected  boolean MetaClassifier.shouldTrain(String targetClass, PredictionDistribution predDist, ContextMap context)
          Invoked by TrainableClassifier.trainOnError(FeatureVector, String, Set) to decide whether to train an instance.
protected  boolean ExternalClassifier.shouldTrain(String targetClass, PredictionDistribution predDist, ContextMap context)
          This implementation uses reinforcement training, if a thick threshold is configured.
protected  boolean TrainableClassifier.shouldTrain(String targetClass, PredictionDistribution predDist, ContextMap context)
          Invoked by TrainableClassifier.trainOnError(FeatureVector, String, Set) to decide whether to train an instance.
protected  boolean TieClassifier.trainOnErrorHook(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          Subclasses can implement this hook for more refined error-driven learning.
protected  boolean OneAgainstTheRestClassifier.trainOnErrorHook(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          Subclasses can implement this hook for more refined error-driven learning.
protected  boolean MultiBinaryClassifier.trainOnErrorHook(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          Subclasses can implement this hook for more refined error-driven learning.
protected  boolean MetaClassifier.trainOnErrorHook(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          Subclasses can implement this hook for more refined error-driven learning.
protected  boolean MoonClassifier.trainOnErrorHook(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          Subclasses can implement this hook for more refined error-driven learning.
protected  boolean TrainableClassifier.trainOnErrorHook(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          Subclasses can implement this hook for more refined error-driven learning.
 

Uses of PredictionDistribution in de.fu_berlin.ties.classify.winnow
 

Subclasses of PredictionDistribution in de.fu_berlin.ties.classify.winnow
 class WinnowDistribution
          A distribution over the classes predicted by a Winnow classifier.
 

Methods in de.fu_berlin.ties.classify.winnow that return PredictionDistribution
protected  PredictionDistribution Winnow.doClassify(FeatureVector features, Set candidateClasses, ContextMap context)
          Classifies an item that is represented by a feature vector by choosing the most probable class among a set of candidate classes.
 

Methods in de.fu_berlin.ties.classify.winnow with parameters of type PredictionDistribution
protected  boolean Winnow.trainOnErrorHook(PredictionDistribution predDist, FeatureVector features, String targetClass, Set candidateClasses, ContextMap context)
          Hook implementing error-driven learning, promoting and demoting weights as required.
 

Uses of PredictionDistribution in de.fu_berlin.ties.combi
 

Methods in de.fu_berlin.ties.combi with parameters of type PredictionDistribution
 CombinationState TrivialStrategy.translateResult(PredictionDistribution[] predictions, TokenDetails details)
          Translates the classes predicted by the classifiers in a decision for the caller.
 CombinationState OpenCloseStrategy.translateResult(PredictionDistribution[] predictions, TokenDetails details)
          Translates the classes predicted by the classifiers in a decision for the caller.
 CombinationState InsideOutsideStrategy.translateResult(PredictionDistribution[] predictions, TokenDetails details)
          Translates the classes predicted by the classifiers in a decision for the caller.
 CombinationState BeginEndStrategy.translateResult(PredictionDistribution[] predictions, TokenDetails details)
          Translates the classes predicted by the classifiers in a decision for the caller.
abstract  CombinationState CombinationStrategy.translateResult(PredictionDistribution[] predictions, TokenDetails details)
          Translates the classes predicted by the classifiers in a decision for the caller.
 CombinationState BeginAfterStrategy.translateResult(PredictionDistribution[] predictions, TokenDetails details)
          Translates the classes predicted by the classifiers in a decision for the caller.
 void BeginEndStrategy.updateState(CombinationState newState, PredictionDistribution[] predictions, TokenDetails details)
          Updates the current state (the class of the last item).
 void CombinationStrategy.updateState(CombinationState newState, PredictionDistribution[] predictions, TokenDetails details)
          Updates the current state (the class of the last item).
 

Uses of PredictionDistribution in de.fu_berlin.ties.demo
 

Constructors in de.fu_berlin.ties.demo with parameters of type PredictionDistribution
FilterResult(PredictionDistribution myPredDist, String myText, TokenizingExtractor extractor, Map<String,List<Float>> myFeatureWeights)
          Creates a new instance.
 

Uses of PredictionDistribution in de.fu_berlin.ties.filter
 

Methods in de.fu_berlin.ties.filter that return PredictionDistribution
 PredictionDistribution TrainableFilter.trainIfNecessary(Element element, boolean decision)
          Trains the correct decision for an element by calling the TrainableClassifier.trainOnError(FeatureVector, String, java.util.Set) method on the stored trainable classifier.
 



Copyright © 2003-2007 Christian Siefkes. All Rights Reserved.