edu.illinois.cs.cogcomp.lbjava.learn

Class SparseWeightVector

java.lang.Object

edu.illinois.cs.cogcomp.lbjava.learn.SparseWeightVector

All Implemented Interfaces:

Serializable, Cloneable

Direct Known Subclasses:

BiasedWeightVector, NaiveBayes.NaiveBayesVector, RandomWeightVector, SparseAveragedPerceptron.AveragedWeightVector

public class SparseWeightVector
extends Object
implements Cloneable, Serializable

This class is used as a weight vector in sparse learning algorithms. Features are associated with Doubles and/or with double[]s representing the weights of the features they produce. Features not appearing in the vector are assumed to have the defaultWeight.

Author:

Nick Rizzolo

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected static int	defaultCapacity The initial capacity for weights if not specified otherwise.
protected static double	defaultWeight When a feature appears in an example but not in this vector, it is assumed to have this weight.
protected edu.illinois.cs.cogcomp.core.datastructures.vectors.DVector	weights The weights in the vector indexed by their Lexicon key.

Constructor Summary

Constructors

Constructor and Description
SparseWeightVector() Simply instantiates weights.
SparseWeightVector(double[] w) Simply initializes weights.
SparseWeightVector(edu.illinois.cs.cogcomp.core.datastructures.vectors.DVector w) Simply initializes weights.

Method Summary

Modifier and Type	Method and Description
void	clear() Empties the weight map.
Object	clone() Returns a copy of this SparseWeightVector in which the weights variable has been cloned deeply.
double	dot(int[] exampleFeatures, double[] exampleValues) Takes the dot product of this SparseWeightVector with the argument vector, using the hard coded default weight.
double	dot(int[] exampleFeatures, double[] exampleValues, double defaultW) Takes the dot product of this SparseWeightVector with the argument vector, using the specified default weight when one is not yet present in this vector.
SparseWeightVector	emptyClone() Returns a new, empty weight vector with the same parameter settings as this one.
double	getWeight(int featureIndex) Returns the weight of the given feature.
double	getWeight(int featureIndex, double defaultW) Returns the weight of the given feature.
Object[]	pairwiseMultiply(int[] exampleFeatures, double[] exampleValues, double defaultW, boolean inverse) The strength of each feature in the argument vector is multiplied by the corresponding weight in this weight vector and the result is returned as an array of arrays.
void	read(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessInputStream in) Reads the representation of a weight vector with this object's run-time type from the given stream, overwriting the data in this object.
static SparseWeightVector	readWeightVector(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessInputStream in) Reads the binary representation of a weight vector of any type from the given stream.
void	scaledAdd(int[] exampleFeatures, double[] exampleValues) Self-modifying vector addition.
void	scaledAdd(int[] exampleFeatures, double[] exampleValues, double factor) Self-modifying vector addition where the argument vector is first scaled by the given factor.
void	scaledAdd(int[] exampleFeatures, double[] exampleValues, double factor, double defaultW) Self-modifying vector addition where the argument vector is first scaled by the given factor.
void	scaledMultiply(int[] exampleFeatures, double[] exampleValues, double factor) Self-modifying vector multiplication where the argument vector is first scaled by the given factor.
void	scaledMultiply(int[] exampleFeatures, double[] exampleValues, double factor, double defaultW) Self-modifying vector multiplication where the argument vector is first scaled by the given factor.
protected void	setWeight(int featureIndex, double w) Sets the weight of the given feature.
protected void	setWeight(int featureIndex, double w, double defaultW) Sets the weight of the given feature.
int	size() Returns the length of the weight vector.
String	toString() Creates a string representation of this SparseWeightVector.
String	toString(Lexicon lex) Creates a string representation of this SparseWeightVector.
void	toStringJustWeights(PrintStream out) Outputs a textual representation of this SparseWeightVector to a stream just like write(PrintStream), but without the "Begin" and "End" annotations.
void	toStringJustWeights(PrintStream out, int min, Lexicon lex) Outputs a textual representation of this SparseWeightVector to a stream just like write(PrintStream), but without the "Begin" and "End" annotations.
void	write(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessOutputStream out) Writes the weight vector's internal representation in binary form.
void	write(PrintStream out) Outputs the contents of this SparseWeightVector into the specified PrintStream.
void	write(PrintStream out, Lexicon lex) Outputs the contents of this SparseWeightVector into the specified PrintStream.

Methods inherited from class java.lang.Object

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

Field Detail

defaultWeight

protected static final double defaultWeight

When a feature appears in an example but not in this vector, it is assumed to have this weight.

See Also:

Constant Field Values

defaultCapacity

protected static final int defaultCapacity

The initial capacity for weights if not specified otherwise.

See Also:

Constant Field Values

weights

protected edu.illinois.cs.cogcomp.core.datastructures.vectors.DVector weights

The weights in the vector indexed by their Lexicon key.

Constructor Detail

SparseWeightVector

public SparseWeightVector()

Simply instantiates weights.

SparseWeightVector

public SparseWeightVector(double[] w)

Simply initializes weights.

Parameters:

w - An array of weights.

SparseWeightVector

public SparseWeightVector(edu.illinois.cs.cogcomp.core.datastructures.vectors.DVector w)

Simply initializes weights.

Parameters:

w - A vector of weights.

Method Detail

getWeight

public double getWeight(int featureIndex)

Returns the weight of the given feature.

Parameters:

featureIndex - The feature index.

Returns:

The weight of the feature.

getWeight

public double getWeight(int featureIndex,
                        double defaultW)

Returns the weight of the given feature.

Parameters:

featureIndex - The feature index.

defaultW - The default weight.

Returns:

The weight of the feature.

setWeight

protected void setWeight(int featureIndex,
                         double w)

Sets the weight of the given feature.

Parameters:

featureIndex - The feature index.

w - The new weight.

setWeight

protected void setWeight(int featureIndex,
                         double w,
                         double defaultW)

Sets the weight of the given feature.

Parameters:

featureIndex - The feature index.

w - The new weight.

defaultW - The default weight.

dot

public double dot(int[] exampleFeatures,
                  double[] exampleValues)

Takes the dot product of this SparseWeightVector with the argument vector, using the hard coded default weight.

Parameters:

exampleFeatures - The example's feature indices.

exampleValues - The example's feature values.

Returns:

The computed dot product.

dot

public double dot(int[] exampleFeatures,
                  double[] exampleValues,
                  double defaultW)

Takes the dot product of this SparseWeightVector with the argument vector, using the specified default weight when one is not yet present in this vector.

Parameters:

exampleFeatures - The example's feature indices.

exampleValues - The example's feature values.

defaultW - The default weight.

Returns:

The computed dot product.

scaledAdd

public void scaledAdd(int[] exampleFeatures,
                      double[] exampleValues)

Self-modifying vector addition.

Parameters:

exampleFeatures - The example's feature indices.

exampleValues - The example's feature values.

scaledAdd

public void scaledAdd(int[] exampleFeatures,
                      double[] exampleValues,
                      double factor)

Self-modifying vector addition where the argument vector is first scaled by the given factor. The default weight is used to initialize new feature weights.

Parameters:

exampleFeatures - The example's feature indices.

exampleValues - The example's feature values.

factor - The scaling factor.

scaledAdd

public void scaledAdd(int[] exampleFeatures,
                      double[] exampleValues,
                      double factor,
                      double defaultW)

Self-modifying vector addition where the argument vector is first scaled by the given factor.

Parameters:

exampleFeatures - The example's feature indices.

exampleValues - The example's feature values.

factor - The scaling factor.

defaultW - An initial weight for previously unseen features.

scaledMultiply

public void scaledMultiply(int[] exampleFeatures,
                           double[] exampleValues,
                           double factor)

Self-modifying vector multiplication where the argument vector is first scaled by the given factor. The default weight is used to initialize new feature weights.

Parameters:

exampleFeatures - The example's feature indices.

exampleValues - The example's feature values.

factor - The scaling factor.

scaledMultiply

public void scaledMultiply(int[] exampleFeatures,
                           double[] exampleValues,
                           double factor,
                           double defaultW)

Self-modifying vector multiplication where the argument vector is first scaled by the given factor.

Parameters:

exampleFeatures - The example's feature indices.

exampleValues - The example's feature values.

factor - The scaling factor.

defaultW - An initial weight for previously unseen features.

pairwiseMultiply

public Object[] pairwiseMultiply(int[] exampleFeatures,
                                 double[] exampleValues,
                                 double defaultW,
                                 boolean inverse)

The strength of each feature in the argument vector is multiplied by the corresponding weight in this weight vector and the result is returned as an array of arrays. The first array contains the integer keys of the example's features, as indexed in the lexicon. The second array gives the double values corresponding to the product of the pairwise multiplication of the strengths of that feature.

Parameters:

exampleFeatures - The example's feature indices.

exampleValues - The example's feature values.

defaultW - An initial weight for previously unseen features.

inverse - When set to true the weight in this vector is inverted before the multiplication takes place.

Returns:

A new example vector representing the pairwise multiplication.

clear

public void clear()

Empties the weight map.

size

public int size()

Returns the length of the weight vector.

write

public void write(PrintStream out)

Outputs the contents of this SparseWeightVector into the specified PrintStream. The string representation starts with a "Begin" annotation, ends with an "End" annotation, and without a Lexicon passed as a parameter, the weights are simply printed in the order of their integer indices.

Parameters:

out - The stream to write to.

write

public void write(PrintStream out,
                  Lexicon lex)

Outputs the contents of this SparseWeightVector into the specified PrintStream. The string representation starts with a "Begin" annotation, ends with an "End" annotation, and lists each feature with its corresponding weight on the same, separate line in between.

Parameters:

out - The stream to write to.

lex - The feature lexicon.

toStringJustWeights

public void toStringJustWeights(PrintStream out)

Outputs a textual representation of this SparseWeightVector to a stream just like write(PrintStream), but without the "Begin" and "End" annotations. Without a Lexicon passed as a parameter, the weights are simply printed in the order of their integer indices.

Parameters:

out - The stream to write to.

toStringJustWeights

public void toStringJustWeights(PrintStream out,
                                int min,
                                Lexicon lex)

Outputs a textual representation of this SparseWeightVector to a stream just like write(PrintStream), but without the "Begin" and "End" annotations. With a Lexicon passed as a parameter, the feature is printed along with each weight.

Parameters:

out - The stream to write to.

min - Sets the minimum width for the textual representation of all features.

lex - The feature lexicon.

toString

public String toString()

Creates a string representation of this SparseWeightVector. This method merely returns the data computed by write(PrintStream).

Overrides:

toString in class Object

Returns:

A textual representation of this vector.

toString

public String toString(Lexicon lex)

Creates a string representation of this SparseWeightVector. This method merely returns the data computed by write(PrintStream,Lexicon).

Parameters:

lex - The feature lexicon.

Returns:

A textual representation of this vector.

write

public void write(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessOutputStream out)

Writes the weight vector's internal representation in binary form.

Parameters:

out - The output stream.

readWeightVector

public static SparseWeightVector readWeightVector(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessInputStream in)

Reads the binary representation of a weight vector of any type from the given stream. The stream is expected to first return a string containing the fully qualified class name of the weight vector. If the short value -1 appears instead, this method returns null.

This method is appropriate for reading weight vectors as written by write(ExceptionlessOutputStream).

Parameters:

in - The input stream.

Returns:

The weight vector read from the stream.

read

public void read(edu.illinois.cs.cogcomp.core.datastructures.vectors.ExceptionlessInputStream in)

Reads the representation of a weight vector with this object's run-time type from the given stream, overwriting the data in this object.

This method is appropriate for reading weight vectors as written by write(ExceptionlessOutputStream).

Parameters:

in - The input stream.

clone

public Object clone()

Returns a copy of this SparseWeightVector in which the weights variable has been cloned deeply.

Overrides:

clone in class Object

Returns:

A copy of this SparseWeightVector.

emptyClone

public SparseWeightVector emptyClone()

Returns a new, empty weight vector with the same parameter settings as this one.

Returns:

An empty weight vector.