edu.udo.cs.yale.operator.learner.igss.utility
Class InstanceAveraging

java.lang.Object
  edu.udo.cs.yale.operator.learner.igss.utility.AbstractUtility
      edu.udo.cs.yale.operator.learner.igss.utility.InstanceAveraging

All Implemented Interfaces:

Utility

Direct Known Subclasses:

Accuracy, Coverage, WRAcc

public abstract class InstanceAveraging
extends AbstractUtility

Abstract superclass for all instance-averaging functions.

Version:

$Id: InstanceAveraging.java,v 1.1 2006/10/02 16:07:49 ingomierswa Exp $

Author:

Dirk Dach

Field Summary

Fields inherited from class edu.udo.cs.yale.operator.learner.igss.utility.AbstractUtility

large, priors

Fields inherited from interface edu.udo.cs.yale.operator.learner.igss.utility.Utility

FIRST_TYPE_INDEX, LAST_TYPE_INDEX, TYPE_ACCURACY, TYPE_BINOMIAL, TYPE_LINEAR, TYPE_SQUARED, TYPE_WRACC, UTILITY_TYPES

Constructor Summary

InstanceAveraging(double[] priors, int large)
Constructor

Method Summary

double	conf(double totalWeight, double delta) Calculate confidence intervall without a specific rule for instance averaging functions.
double	conf(double totalWeight, double totalPositiveWeight, Hypothesis hypo, double delta) Calculate confidence intervall for a specific rule for instance averaging functions.
double	confidenceIntervall(double totalWeight, double totalPositiveWeight, Hypothesis hypo, double delta) Calculates the the confidence intervall for a specific hypothesis.
double	confSmallM(double totalWeight, double delta) Calculate confidence intervall without a specific rule for instance averaging functions and small m.
abstract double	variance(double totalWeight, double totalPositiveWeight, Hypothesis hypo) Calculates the empirical variance.

Methods inherited from class edu.udo.cs.yale.operator.learner.igss.utility.AbstractUtility

calculateM, confidenceIntervall, inverseNormal

Methods inherited from class java.lang.Object

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

Methods inherited from interface edu.udo.cs.yale.operator.learner.igss.utility.Utility

getUpperBound, utility

Constructor Detail

InstanceAveraging

public InstanceAveraging(double[] priors,
                         int large)

Constructor

Method Detail

confidenceIntervall

public double confidenceIntervall(double totalWeight,
                                  double totalPositiveWeight,
                                  Hypothesis hypo,
                                  double delta)

Calculates the the confidence intervall for a specific hypothesis. Uses Chernoff bounds if the number of random experiments is too small and normal approximation otherwise. This method is adapted for instance averaging utility types. Every example is considered a random experiment, because f_inst is evaluated for every example!!! This is the reason why total weight is used instead of covered weight Should be overwritten by subclasses if they make a different random experiment.

Specified by:

confidenceIntervall in interface Utility

Overrides:

confidenceIntervall in class AbstractUtility

conf

public double conf(double totalWeight,
                   double delta)

Calculate confidence intervall without a specific rule for instance averaging functions.

Specified by:

conf in class AbstractUtility

conf

public double conf(double totalWeight,
                   double totalPositiveWeight,
                   Hypothesis hypo,
                   double delta)

Calculate confidence intervall for a specific rule for instance averaging functions.

Specified by:

conf in class AbstractUtility

variance

public abstract double variance(double totalWeight,
                                double totalPositiveWeight,
                                Hypothesis hypo)

Calculates the empirical variance.

confSmallM

public double confSmallM(double totalWeight,
                         double delta)

Calculate confidence intervall without a specific rule for instance averaging functions and small m.

Specified by:

confSmallM in class AbstractUtility