diva.sketch.classification

Class WeightSet

java.lang.Object

diva.sketch.classification.WeightSet

Direct Known Subclasses:

GaussianWeightSet

public abstract class WeightSet
extends Object

A WeightSet object represents a training type (e.g. square, class, circle, etc.) It has a mu vector and a sigma vector. Each element in the mu vector represents the average value (e.g. mean) for a particular feature of this class, and the corresponding sigma value indicates how much the value of the feature may vary (e.g. standard deviation). However, the exact statistical calculation for mu and sigma vary depending on the type of classification algorithms.

Version:

$Revision: 1.5 $

Author:

Heloise Hse (hwawen@eecs.berkeley.edu), Michael Shilman (michaels@eecs.berkeley.edu)

Constructor Summary

Constructors

Constructor and Description
WeightSet(String type) Construct a WeightSet of the specified type.

Method Summary

Modifier and Type	Method and Description
void	addExample(FeatureSet f) Add an example to this classifier.
void	clearExamples() Clear the examples.
protected abstract FeatureSet	computeMu() Calculate the mu values for each feature in this class.
protected abstract FeatureSet	computeSigma() Calculate the sigma values for each feature in this class.
protected void	debug(String s)
protected Iterator	examples() Return an iterator over the examples of this classifier.
int	getExampleCount() Return the number of examples.
FeatureSet	getMuValues() Return the mu values.
FeatureSet	getSigmaValues() Return the sigma values.
String	getType() Return the type of this classifier.
String	toString() The text representation of this weight set.
void	train() Train on the examples by computing the mu and sigma values for each feature.

Methods inherited from class java.lang.Object

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

Constructor Detail

WeightSet

public WeightSet(String type)

Construct a WeightSet of the specified type.

Method Detail

addExample

public void addExample(FeatureSet f)

Add an example to this classifier. The example will be used to compute mu and sigma.

clearExamples

public void clearExamples()

Clear the examples. Once the training is complete it is a waste of memory to retain all of the examples.

computeMu

protected abstract FeatureSet computeMu()
                                 throws ClassifierException

Calculate the mu values for each feature in this class. Return the feature set representing the mu values.

Throws:

ClassifierException

computeSigma

protected abstract FeatureSet computeSigma()
                                    throws ClassifierException

Calculate the sigma values for each feature in this class. Return the feature set representing the sigma values.

Throws:

ClassifierException

debug

protected void debug(String s)

examples

protected Iterator examples()

Return an iterator over the examples of this classifier. The examples are cleared once training is complete, so this method will return an empty iterator if it is called post-training.

getExampleCount

public int getExampleCount()

Return the number of examples.

getMuValues

public FeatureSet getMuValues()

Return the mu values.

getSigmaValues

public FeatureSet getSigmaValues()

Return the sigma values.

getType

public String getType()

Return the type of this classifier.

train

public void train()
           throws ClassifierException

Train on the examples by computing the mu and sigma values for each feature. This method clears the examples once the training is complete, to avoid wasted space.

Throws:

ClassifierException

toString

public String toString()

The text representation of this weight set.

Overrides:

toString in class Object