MutualInformationFeatureSelectionExtractor (ClearTK TimeML 2.0.0 API)

java.lang.Object
- org.cleartk.ml.feature.transform.TrainableExtractor_ImplBase<OUTCOME_T>
- - org.cleartk.ml.feature.selection.FeatureSelectionExtractor<OUTCOME_T>
  - - org.cleartk.ml.feature.selection.MutualInformationFeatureSelectionExtractor<OUTCOME_T,FOCUS_T>

All Implemented Interfaces:

Predicate<Feature>, FeatureExtractor1<FOCUS_T>, TrainableExtractor<OUTCOME_T>
```
public class MutualInformationFeatureSelectionExtractor<OUTCOME_T,FOCUS_T extends Annotation>
extends FeatureSelectionExtractor<OUTCOME_T>
implements FeatureExtractor1<FOCUS_T>
```
Copyright (c) 2012, Regents of the University of Colorado
All rights reserved.
Selects features via mutual information statistics between the features extracted from its sub-extractor and the outcome values they are paired with in classification instances.

Author:

Lee Becker

Nested Class Summary

Nested Classes
Modifier and Type	Class and Description
`static class`	`MutualInformationFeatureSelectionExtractor.CombineScoreMethod` Specifies how scores for each outcome should be combined/aggregated into a single score
`static class`	`MutualInformationFeatureSelectionExtractor.MutualInformationStats<OUTCOME_T>` Helper class for aggregating and computing mutual information statistics

Field Summary

Fields
Modifier and Type Field and Description

protected boolean isTrained
- Fields inherited from class org.cleartk.ml.feature.transform.TrainableExtractor_ImplBase
  name

Fields
Modifier and Type	Field and Description
`protected boolean`	`isTrained`

Constructor Summary

Constructors
Constructor and Description
`MutualInformationFeatureSelectionExtractor(String name, FeatureExtractor1<FOCUS_T> extractor)`
`MutualInformationFeatureSelectionExtractor(String name, FeatureExtractor1<FOCUS_T> extractor, int numFeatures)`
`MutualInformationFeatureSelectionExtractor(String name, FeatureExtractor1<FOCUS_T> extractor, MutualInformationFeatureSelectionExtractor.CombineScoreMethod combineMeasureType, double smoothingCount, int numFeatures)`

Method Summary

Methods
Modifier and Type	Method and Description
`boolean`	`apply(Feature feature)`
`List<Feature>`	`extract(JCas view, FOCUS_T focusAnnotation)` Some feature extractors will require some specific number of annotations that is greater than one.
`List<String>`	`getSelectedFeatures()`
`void`	`load(URI uri)` Loads statistics from location URI
`String`	`nameFeature(Feature feature)`
`void`	`save(URI uri)` Saves statistics from train in location URI
`void`	`train(Iterable<Instance<OUTCOME_T>> instances)` In the prototypical case, train takes a collection of instances and computes statistics over the values such as computing mean, standard deviation, TF*IDF, etc...

Methods inherited from class org.cleartk.ml.feature.selection.FeatureSelectionExtractor
transform

Methods inherited from class org.cleartk.ml.feature.transform.TrainableExtractor_ImplBase
isTransformable, selectTransformableFeatures

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface com.google.common.base.Predicate
equals

- Field Detail
  - isTrained
```
protected boolean isTrained
```
- Constructor Detail
  - MutualInformationFeatureSelectionExtractor
```
public MutualInformationFeatureSelectionExtractor(String name,
                                          FeatureExtractor1<FOCUS_T> extractor)
```
  - MutualInformationFeatureSelectionExtractor
```
public MutualInformationFeatureSelectionExtractor(String name,
                                          FeatureExtractor1<FOCUS_T> extractor,
                                          int numFeatures)
```
  - MutualInformationFeatureSelectionExtractor
```
public MutualInformationFeatureSelectionExtractor(String name,
                                          FeatureExtractor1<FOCUS_T> extractor,
                                          MutualInformationFeatureSelectionExtractor.CombineScoreMethod combineMeasureType,
                                          double smoothingCount,
                                          int numFeatures)
```
- Method Detail
  - apply
```
public boolean apply(Feature feature)
```
    Specified by:
    
    apply in interface Predicate<Feature>
  - extract
```
public List<Feature> extract(JCas view,
                    FOCUS_T focusAnnotation)
                      throws CleartkExtractorException
```
    Description copied from interface: FeatureExtractor1
    
    Some feature extractors will require some specific number of annotations that is greater than one. Such extractors should not implement this interface.
    
    Specified by:
    
    extract in interface FeatureExtractor1<FOCUS_T extends Annotation>
    
    Parameters:
    view - the current view of the document.
    focusAnnotation - the annotation for which to collect features.
    
    Returns:
    a list of features generated by the extractor for the focus annotation.
    
    Throws:
    
    CleartkExtractorException
  - getSelectedFeatures
```
public final List<String> getSelectedFeatures()
```
  - load
```
public void load(URI uri)
          throws IOException
```
    Description copied from interface: TrainableExtractor
    
    Loads statistics from location URI
    
    Specified by:
    
    load in interface TrainableExtractor<OUTCOME_T>
    
    Throws:
    
    IOException
  - nameFeature
```
public String nameFeature(Feature feature)
```
  - save
```
public void save(URI uri)
          throws IOException
```
    Description copied from interface: TrainableExtractor
    
    Saves statistics from train in location URI
    
    Specified by:
    
    save in interface TrainableExtractor<OUTCOME_T>
    
    Throws:
    
    IOException
  - train
```
public void train(Iterable<Instance<OUTCOME_T>> instances)
```
    Description copied from interface: TrainableExtractor
    
    In the prototypical case, train takes a collection of instances and computes statistics over the values such as computing mean, standard deviation, TF*IDF, etc...
    
    Specified by:
    
    train in interface TrainableExtractor<OUTCOME_T>
    
    Parameters:
    instances - - URI pointing to the output location for saving statistics

Class MutualInformationFeatureSelectionExtractor<OUTCOME_T,FOCUS_T extends Annotation>

Nested Class Summary

Field Summary

Fields inherited from class org.cleartk.ml.feature.transform.TrainableExtractor_ImplBase

Constructor Summary

Method Summary

Methods inherited from class org.cleartk.ml.feature.selection.FeatureSelectionExtractor

Methods inherited from class org.cleartk.ml.feature.transform.TrainableExtractor_ImplBase

Methods inherited from class java.lang.Object

Methods inherited from interface com.google.common.base.Predicate

Field Detail

isTrained

Constructor Detail

MutualInformationFeatureSelectionExtractor

MutualInformationFeatureSelectionExtractor

MutualInformationFeatureSelectionExtractor

Method Detail

apply

extract

getSelectedFeatures

load

nameFeature

save

train