marf.Classification
Class Classification

java.lang.Object
  marf.Storage.StorageManager
      marf.Classification.Classification

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, IClassification, IStorageManager

Direct Known Subclasses:

CosineSimilarityMeasure, Distance, NeuralNetwork, RandomClassification, Stochastic

public abstract class Classification
extends StorageManager
implements IClassification

Abstract Classification Module. A generic implementation of the IClassification interface. The derivatives must inherit from this class, and if they cannot, they should implement IClassification themselves.

$Id: Classification.java,v 1.50 2008/02/19 03:20:30 mokhov Exp $

Since:

0.0.1

Version:

$Revision: 1.50 $

Author:

Serguei Mokhov

See Also:

Serialized Form

Field Summary

protected double[]	adFeatureVector Local reference to the array of features, either obtained from the feature extraction module or passed directly to train() or classify.
protected IFeatureExtraction	oFeatureExtraction Reference to the enclosed FeatureExtraction object.
protected ResultSet	oResultSet Classification result set.
protected TrainingSet	oTrainingSet TrainingSet Container.

Fields inherited from class marf.Storage.StorageManager

bDumpOnNotFound, iCurrentDumpMode, oObjectToSerialize, strFilename

Fields inherited from interface marf.Classification.IClassification

MARF_INTERFACE_CODE_REVISION

Fields inherited from interface marf.Storage.IStorageManager

DUMP_BINARY, DUMP_CSV_TEXT, DUMP_GZIP_BINARY, DUMP_HTML, DUMP_SQL, DUMP_XML, MARF_INTERFACE_CODE_REVISION, STORAGE_FILE_EXTENSIONS

Constructor Summary

protected

Classification(IFeatureExtraction poFeatureExtraction) Generic Classification Constructor.

Method Summary

boolean	classify() Generic classification routine that assumes a presence of a valid non-null feature extraction module for pipeline operation.
java.lang.Object	clone() Implements Cloneable interface for the Classification object.
void	dump() Generic implementation of dump() to dump the TrainingSet.
IFeatureExtraction	getFeatureExtraction() Retrieves the features source.
static java.lang.String	getMARFSourceCodeRevision() Retrieves class' revision.
ResultSet	getResultSet() Retrieves the enclosed result set.
protected java.lang.String	getTrainingSetFilename() Constructs a global cluster file name for the TrainingSet.
static TrainingSet	loadTrainingSet(int piDumpMode, java.lang.String pstrFilename) Allows loading of the training sets for debugging and browsing purposes by external classes.
void	restore() Generic implementation of restore() for TrainingSet.
void	setFeatureExtraction(IFeatureExtraction poFeatureExtraction) Allows setting the features source.
boolean	train() Generic training routine for building/updating mean vectors in the training set.
boolean	train(double[] padFeatureVector) Generic training routine for building/updating mean vectors in the training set.

Methods inherited from class marf.Storage.StorageManager

backSynchronizeObject, dumpBinary, dumpCSV, dumpGzipBinary, dumpHTML, dumpSQL, dumpXML, enableDumpOnNotFound, equals, getDefaultExtension, getDefaultExtension, getDumpMode, getFilename, getObjectToSerialize, hashCode, restoreBinary, restoreCSV, restoreGzipBinary, restoreHTML, restoreSQL, restoreXML, setDumpMode, setFilename, toString

Methods inherited from class java.lang.Object

finalize, getClass, notify, notifyAll, wait, wait, wait

Methods inherited from interface marf.Classification.IClassification

classify, getResult

Field Detail

oFeatureExtraction

protected IFeatureExtraction oFeatureExtraction

Reference to the enclosed FeatureExtraction object.

oTrainingSet

protected TrainingSet oTrainingSet

TrainingSet Container.

adFeatureVector

protected double[] adFeatureVector

Local reference to the array of features, either obtained from the feature extraction module or passed directly to train() or classify. Used in prevention of NullPointerException, bug #1539695.

Since:

0.3.0.6

See Also:

train(double[]), IClassification.classify(double[])

oResultSet

protected ResultSet oResultSet

Classification result set. May contain one or more results (in case of similarity).

Since:

0.3.0.2

Constructor Detail

Classification

protected Classification(IFeatureExtraction poFeatureExtraction)

Generic Classification Constructor.

Parameters:

poFeatureExtraction - FeatureExtraction module reference

Method Detail

train

public boolean train()
              throws ClassificationException

Generic training routine for building/updating mean vectors in the training set. Assumes presence of a non-null feature extraction module for pipelining. Can be overridden, and if the overriding classifier is using TrainingSet, it should call super.train();

Specified by:

train in interface IClassification

Returns:

true if training was successful (i.e. mean vector was updated); false otherwise

Throws:

ClassificationException - if there was a problem while training

See Also:

TrainingSet

train

public boolean train(double[] padFeatureVector)
              throws ClassificationException

Generic training routine for building/updating mean vectors in the training set. Can be overridden, and if the overriding classifier is using TrainingSet, it should call super.train();

Specified by:

train in interface IClassification

Parameters:

padFeatureVector - feature vector to train on

Returns:

true if training was successful (i.e. mean vector was updated); false otherwise

Throws:

ClassificationException - if there was a problem while training

Since:

0.3.0.6

See Also:

TrainingSet

classify

public boolean classify()
                 throws ClassificationException

Generic classification routine that assumes a presence of a valid non-null feature extraction module for pipeline operation.

Specified by:

classify in interface IClassification

Returns:

true if classification was successful; false otherwise

Throws:

ClassificationException - if there was an error while classifying

Since:

0.3.0.6

See Also:

IClassification.classify()

dump

public void dump()
          throws StorageException

Generic implementation of dump() to dump the TrainingSet.

Specified by:

dump in interface IStorageManager

Overrides:

dump in class StorageManager

Throws:

StorageException - if there's a problem saving training set to disk

Since:

0.2.0

See Also:

IStorageManager.DUMP_GZIP_BINARY, IStorageManager.DUMP_BINARY, StorageManager.dumpGzipBinary(), StorageManager.dumpCSV(), StorageManager.dumpBinary(), StorageManager.dumpXML(), StorageManager.dumpHTML(), StorageManager.dumpSQL(), StorageManager.backSynchronizeObject(), StorageManager.iCurrentDumpMode

restore

public void restore()
             throws StorageException

Generic implementation of restore() for TrainingSet.

Specified by:

restore in interface IStorageManager

Overrides:

restore in class StorageManager

Throws:

StorageException - if there is a problem loading the training set from disk

Since:

0.2.0

See Also:

IStorageManager.DUMP_GZIP_BINARY, IStorageManager.DUMP_BINARY, StorageManager.dumpGzipBinary(), StorageManager.dumpCSV(), StorageManager.dumpBinary(), StorageManager.dumpXML(), StorageManager.dumpHTML(), StorageManager.dumpSQL(), StorageManager.backSynchronizeObject(), StorageManager.iCurrentDumpMode

loadTrainingSet

public static TrainingSet loadTrainingSet(int piDumpMode,
                                          java.lang.String pstrFilename)
                                   throws StorageException

Allows loading of the training sets for debugging and browsing purposes by external classes.

Parameters:

piDumpMode -

pstrFilename -

Returns:

loaded training set bean if I/O was successful

Throws:

StorageException

Since:

0.3.0.6

getResultSet

public ResultSet getResultSet()

Retrieves the enclosed result set.

Specified by:

getResultSet in interface IClassification

Returns:

the enclosed ResultSet object

Since:

0.3.0.2

getTrainingSetFilename

protected java.lang.String getTrainingSetFilename()

Constructs a global cluster file name for the TrainingSet.

Filename is constructed using fully-qualified class of either TrainingSet or a classifier name with global clustering info such as preprocessing and feature extraction methods, so that only that cluster can be reloaded after.

May be overridden by the derivatives when necessary.

Returns:

String, filename

Since:

0.2.0

getFeatureExtraction

public IFeatureExtraction getFeatureExtraction()

Retrieves the features source.

Specified by:

getFeatureExtraction in interface IClassification

Returns:

returns the FeatureExtraction reference

Since:

0.3.0.4

setFeatureExtraction

public void setFeatureExtraction(IFeatureExtraction poFeatureExtraction)

Allows setting the features source.

Specified by:

setFeatureExtraction in interface IClassification

Parameters:

poFeatureExtraction - the FeatureExtraction object to set

Since:

0.3.0.4

clone

public java.lang.Object clone()

Implements Cloneable interface for the Classification object. The contained FeatureExtraction isn't cloned at this point, and is just assigned to the clone.

Overrides:

clone in class StorageManager

Since:

0.3.0.5

See Also:

Object.clone()

getMARFSourceCodeRevision

public static java.lang.String getMARFSourceCodeRevision()

Retrieves class' revision.

Returns:

revision string

Since:

0.3.0.2