org.ai_heuristic.algo.prob

Class KullbackLeibler

java.lang.Object

org.ai_heuristic.functs.Function

org.ai_heuristic.functs.FunctionMetric

org.ai_heuristic.algo.prob.KullbackLeibler

All Implemented Interfaces:

FunctionDef, FunctionMetricDef

public class KullbackLeibler
extends FunctionMetric

This class implements the Kullback-Leibler divergence going from dataset2 to dataset1. The function uses MathStat.logBaseTwo and so the data points should be of type Float or Double. The desired result is a smaller distance, but you can use the isBetter() method to determine what value is better. Null can be returned for invalid data.

Author:

Kieran Greer

Field Summary

Fields inherited from class org.ai_heuristic.functs.Function

config, mathCompare, valueType

Constructor Summary

Constructors

Constructor and Description
KullbackLeibler() Create a new instance of Kullback-Leibler.
KullbackLeibler(java.lang.String thisValueType) Create a new instance of Kullback-Leibler.

Method Summary

Modifier and Type	Method and Description
double	evaluate(double[] p, double[] q) Evaluate the Kullback-Leibler divergence going from q to p.
ReplySet	evaluate(MetricDataset ds1, MetricDataset ds2) Evaluate the comparison of the two vectors of data and return the result.
protected void	initialise() Initialise the function values, setting the config parameters or other.
boolean	isBetter(java.lang.String valueType, java.lang.Object value1, java.lang.Object value2) Return true if value2 is better than value1, as determined by the measurements of this evaluation function.
FunctionMetric	newInstance() Create and return a new instance of the function, initialised with this function's value type valueType and mathOperation evaluator mathCompare.
boolean	sib() Return true if a smaller distance between the two vectors is better.

Methods inherited from class org.ai_heuristic.functs.FunctionMetric

evaluateCompare, lib

Methods inherited from class org.ai_heuristic.functs.Function

checkValueType, createFunction, createFunction, createFunction, evaluate, getConfigParams, innerObject, isLegalNumber, setConfigParams, setEvaluator, setValueType

Methods inherited from class java.lang.Object

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

Constructor Detail

KullbackLeibler

public KullbackLeibler()
                throws java.lang.Exception

Create a new instance of Kullback-Leibler.

Throws:

java.lang.Exception - any error.

KullbackLeibler

public KullbackLeibler(java.lang.String thisValueType)
                throws java.lang.Exception

Create a new instance of Kullback-Leibler.

Parameters:

thisValueType - the type of object being evaluated.

Throws:

java.lang.Exception - any error.

Method Detail

initialise

protected void initialise()
                   throws java.lang.Exception

Initialise the function values, setting the config parameters or other.

Specified by:

initialise in class Function

Throws:

java.lang.Exception - any error.

evaluate

public ReplySet evaluate(MetricDataset ds1,
                         MetricDataset ds2)
                  throws java.lang.Exception

Evaluate the comparison of the two vectors of data and return the result. If each list has only one object, then apply to the objects directly and not the lists.

Specified by:

evaluate in interface FunctionMetricDef

Specified by:

evaluate in class FunctionMetric

Parameters:

ds1 - first value dataset. This should store a list of sample (Float or Double) values.

ds2 - second value dataset. This should store a list of base (Float or Double) values.

Returns:

the result of applying the operator to the values.

Throws:

java.lang.Exception - any error.

evaluate

public double evaluate(double[] p,
                       double[] q)
                throws java.lang.Exception

Evaluate the Kullback-Leibler divergence going from q to p.

Parameters:

p - first set of probability values.

q - second set of probability values.

Returns:

the K-L divergence evaluation.

Throws:

java.lang.Exception - any error.

isBetter

public boolean isBetter(java.lang.String valueType,
                        java.lang.Object value1,
                        java.lang.Object value2)
                 throws java.lang.Exception

Return true if value2 is better than value1, as determined by the measurements of this evaluation function. A smaller value is better.

Specified by:

isBetter in class FunctionMetric

Parameters:

valueType - the java type of the values to be evaluated.

value1 - the first value type.

value2 - the second value type.

Returns:

true if value2 is better than value1, as determined by this evaluation function.

Throws:

java.lang.Exception - any error.

sib

public boolean sib()

Return true if a smaller distance between the two vectors is better.

Specified by:

sib in class FunctionMetric

Returns:

true if a smaller evaluation is better.

newInstance

public FunctionMetric newInstance()
                           throws java.lang.Exception

Create and return a new instance of the function, initialised with this function's value type valueType and mathOperation evaluator mathCompare.

Specified by:

newInstance in class FunctionMetric

Returns:

a new instance of this function.

Throws:

java.lang.Exception - any error.