gov.sandia.cognition.learning.algorithm.nearest.NearestNeighborExhaustive Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of cognitive-foundry Show documentation
Show all versions of cognitive-foundry Show documentation
A single jar with all the Cognitive Foundry components.
/*
* File: NearestNeighborExhaustive.java
* Authors: Justin Basilico
* Company: Sandia National Laboratories
* Project: Cognitive Foundry
*
* Copyright March 24, 2008, Sandia Corporation.
* Under the terms of Contract DE-AC04-94AL85000, there is a non-exclusive
* license for use of this work by or on behalf of the U.S. Government. Export
* of this program may require a license from the United States Government.
* See CopyrightHistory.txt for complete details.
*
*/
package gov.sandia.cognition.learning.algorithm.nearest;
import gov.sandia.cognition.annotation.PublicationReference;
import gov.sandia.cognition.annotation.PublicationType;
import gov.sandia.cognition.learning.algorithm.SupervisedBatchLearner;
import gov.sandia.cognition.learning.data.InputOutputPair;
import gov.sandia.cognition.learning.function.distance.DefaultDivergenceFunctionContainer;
import gov.sandia.cognition.math.DivergenceFunction;
import java.util.Collection;
import java.util.LinkedList;
/**
* The {@code NearestNeighborExhaustive} class implements a simple evaluator
* that looks up a given input object in a collection of input-output pair
* examples and returns the output associated with the most similar input.
* Similarity is judged using the given divergence function. This is an
* implementation of the standard nearest-neighbor algorithm, which is a corner
* case of the k-nearest neighbor algorithm, implemented in the
* {@code KNearestNeighbor} class.
*
* @param The input type for nearest neighbor.
* @param The output type for nearest neighbor.
* @author Justin Basilico
* @since 2.1
* @see KNearestNeighbor
*/
@PublicationReference(
author="Wikipedia",
title="k-nearest neighbor algorithm",
type=PublicationType.WebPage,
year=2008,
url="http://en.wikipedia.org/wiki/K-nearest_neighbor_algorithm"
)
public class NearestNeighborExhaustive
extends AbstractNearestNeighbor
{
/** The data that nearest-neighbor is performed over. */
protected LinkedList> data;
/**
* Creates a new instance of {@code NearestNeighborExhaustive}.
*/
public NearestNeighborExhaustive()
{
this(null);
}
/**
* Creates a new instance of {@code NearestNeighborExhaustive}.
*
* @param divergenceFunction The divergence function to use.
*/
public NearestNeighborExhaustive(
final DivergenceFunction divergenceFunction)
{
this(divergenceFunction, null);
}
/**
* Creates a new instance of {@code NearestNeighborExhaustive}.
*
* @param divergenceFunction The divergence function to use.
* @param data The data to perform nearest neighbor over.
*/
public NearestNeighborExhaustive(
final DivergenceFunction divergenceFunction,
final Collection> data)
{
super(divergenceFunction);
this.setData(new LinkedList>());
if (data != null)
{
this.getData().addAll(data);
}
}
@Override
public NearestNeighborExhaustive clone()
{
@SuppressWarnings("unchecked")
NearestNeighborExhaustive clone =
(NearestNeighborExhaustive) super.clone();
clone.setData( new LinkedList>() );
if( this.getData() != null )
{
clone.getData().addAll( this.getData() );
}
return clone;
}
/**
* Evaluates the object to do nearest-neighbor lookup for the given input.
*
* @param input The input to evaluate.
* @return The output associated with the input that is the most similar
* (least divergent) to the input.
*/
public OutputType evaluate(
final InputType input)
{
// We need to find the best example and best divergence in the data.
InputOutputPair bestExample = null;
double bestDivergence = Double.MAX_VALUE;
// Go through all the data.
for( InputOutputPair example
: getData() )
{
// Calculate the divergence.
final double divergence =
this.divergenceFunction.evaluate(input, example.getInput());
if (bestExample == null || divergence < bestDivergence)
{
bestExample = example;
bestDivergence = divergence;
}
// else - Not the best.
}
if (bestExample != null)
{
// Return the output associated with the best example.
return bestExample.getOutput();
}
else
{
// No output.
return null;
}
}
public LinkedList> getData()
{
return this.data;
}
/**
* Sets the data that the object performs nearest-neighbor lookup on.
*
* @param data The data to perform nearest-neighbor lookup on.
*/
public void setData(
final LinkedList> data)
{
this.data = data;
}
/**
* The {@code NearestNeighborExhaustive.Learner} class implements a batch learner for
* the {@code NearestNeighborExhaustive} class.
*
* @param The input type for nearest neighbor.
* @param The output type for nearest neighbor.
* @author Justin Basilico
* @since 2.1
*/
public static class Learner
extends DefaultDivergenceFunctionContainer
implements SupervisedBatchLearner>
{
/**
* Creates a new instance of {@code NearestNeighborExhaustive.Learner}.
*/
public Learner()
{
this(null);
}
/**
* Creates a new instance of {@code NearestNeighborExhaustive.Learner}.
*
* @param divergenceFunction The divergence function to use.
*/
public Learner(
final DivergenceFunction
divergenceFunction)
{
super(divergenceFunction);
}
public NearestNeighborExhaustive learn(
Collection> data )
{
// Just pass the divergence function and the data to the nearest
// neighbor object.
return new NearestNeighborExhaustive(
this.getDivergenceFunction(), data );
}
}
}