gov.sandia.cognition.learning.data.feature

Class FeatureHashing

java.lang.Object

gov.sandia.cognition.util.AbstractCloneableSerializable

gov.sandia.cognition.learning.data.feature.FeatureHashing

All Implemented Interfaces:

Evaluator<Vector,Vector>, VectorFactoryContainer, VectorFunction, VectorOutputEvaluator<Vector,Vector>, CloneableSerializable, java.io.Serializable, java.lang.Cloneable

@PublicationReference(title="Feature Hashing for Large Scale Multitask Learning",
                      author={"Kilian Weinberger","Anirban Dasgupta","Josh Attenberg","John Langford","Alex Smola"},
                      year=2009,
                      type=Conference,
                      publication="Proceedings of the 26th Annual International Conference on Machine Learning (ICML)",
                      url="http://arxiv.org/pdf/0902.2206.pdf")
public class FeatureHashing
extends AbstractCloneableSerializable
implements VectorFunction, VectorOutputEvaluator<Vector,Vector>, VectorFactoryContainer

Implements a function that applies vector feature hashing. This is also known as the "hashing trick". It takes in a vector (typically sparse) and then applies hashing to the non-zero values of the vector to produce another vector, which is usually of a much lower dimensionality. It does this by hashing each index and then using the sign of the hashed value to either increment or decrement the hashed index by the value.

Since:

3.4.2

Author:

Justin Basilico

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
static int	DEFAULT_OUTPUT_DIMENSIONALITY The default output dimensionality is 100.
protected HashFunction	hashFunction The hashing function to use.
protected int	outputDimensionality The output size of the hash.
protected VectorFactory<?>	vectorFactory Vector factory to use.

Constructor Summary

Constructors

Constructor and Description
FeatureHashing() Creates a new FeatureHashing.
FeatureHashing(int outputDimensionality) Creates a new FeatureHashing with the given output size.
FeatureHashing(int outputDimensionality, HashFunction hashFunction, VectorFactory<?> vectorFactory) Creates a new FeatureHashing with the given parameters.

Method Summary

Modifier and Type	Method and Description
Vector	evaluate(Vector input) Evaluates the function on the given input and returns the output.
HashFunction	getHashFunction() Gets the hash function to use.
int	getOutputDimensionality() Gets the expected dimensionality of the output vector of the evaluator, if it is known.
VectorFactory<?>	getVectorFactory() Gets the vector factory the object to use to create new vectors.
protected int	hash(int index) Applies the hashing function to the index.
void	setHashFunction(HashFunction hashFunction) Gets the hash function to use.
void	setOutputDimensionality(int outputDimensionality) Sets the output dimensionality, which is the size of the output vector that the input is hashed into.
void	setVectorFactory(VectorFactory<?> vectorFactory) Sets the vector factory to use.

Methods inherited from class gov.sandia.cognition.util.AbstractCloneableSerializable

clone

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_OUTPUT_DIMENSIONALITY

public static final int DEFAULT_OUTPUT_DIMENSIONALITY

The default output dimensionality is 100.

See Also:

Constant Field Values

outputDimensionality

protected int outputDimensionality

The output size of the hash.

hashFunction

protected HashFunction hashFunction

The hashing function to use.

vectorFactory

protected VectorFactory<?> vectorFactory

Vector factory to use.

Constructor Detail

FeatureHashing

public FeatureHashing()

Creates a new FeatureHashing.

FeatureHashing

public FeatureHashing(int outputDimensionality)

Creates a new FeatureHashing with the given output size.

Parameters:

outputDimensionality - The output dimensionality. Cannot be negative.

FeatureHashing

public FeatureHashing(int outputDimensionality,
                      HashFunction hashFunction,
                      VectorFactory<?> vectorFactory)

Creates a new FeatureHashing with the given parameters.

Parameters:

outputDimensionality - The output dimensionality. Cannot be negative.

hashFunction - The hash function to use.

vectorFactory - The vector factory to use.

Method Detail

evaluate

public Vector evaluate(Vector input)

Description copied from interface: Evaluator

Evaluates the function on the given input and returns the output.

Specified by:

evaluate in interface Evaluator<Vector,Vector>

Parameters:

input - The input to evaluate.

Returns:

The output produced by evaluating the input.

hash

protected int hash(int index)

Applies the hashing function to the index.

Parameters:

index - The index to hash.

Returns:

The hashed value.

getOutputDimensionality

public int getOutputDimensionality()

Description copied from interface: VectorOutputEvaluator

Gets the expected dimensionality of the output vector of the evaluator, if it is known. If it is not known, -1 is returned.

Specified by:

getOutputDimensionality in interface VectorOutputEvaluator<Vector,Vector>

Returns:

The expected dimensionality of the output vector of the evaluator, or -1 if it is not known.

setOutputDimensionality

public void setOutputDimensionality(int outputDimensionality)

Sets the output dimensionality, which is the size of the output vector that the input is hashed into.

Parameters:

outputDimensionality - The output dimensionality. Cannot be negative.

getHashFunction

public HashFunction getHashFunction()

Gets the hash function to use.

Returns:

The hash function to use.

setHashFunction

public void setHashFunction(HashFunction hashFunction)

Gets the hash function to use.

Parameters:

hashFunction - The hash function to use.

getVectorFactory

public VectorFactory<?> getVectorFactory()

Description copied from interface: VectorFactoryContainer

Gets the vector factory the object to use to create new vectors.

Specified by:

getVectorFactory in interface VectorFactoryContainer

Returns:

The vector factory.

setVectorFactory

public void setVectorFactory(VectorFactory<?> vectorFactory)

Sets the vector factory to use.

Parameters:

vectorFactory - The vector factory.