gov.sandia.cognition.statistics.bayesian

Class RejectionSampling.DefaultUpdater<ObservationType,ParameterType>

java.lang.Object

gov.sandia.cognition.util.AbstractCloneableSerializable

gov.sandia.cognition.statistics.bayesian.RejectionSampling.DefaultUpdater<ObservationType,ParameterType>

Type Parameters:

ObservationType - Type of observation

ParameterType - Type of parameters to infer

All Implemented Interfaces:

RejectionSampling.Updater<ObservationType,ParameterType>, CloneableSerializable, java.io.Serializable, java.lang.Cloneable

Enclosing class:

RejectionSampling<ObservationType,ParameterType>

public static class RejectionSampling.DefaultUpdater<ObservationType,ParameterType>
extends AbstractCloneableSerializable
implements RejectionSampling.Updater<ObservationType,ParameterType>

Default ImportanceSampling Updater that uses a BayesianParameter to compute the quantities of interest.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>>	conjuctive Defines the parameter that connects the conditional and prior distributions.
protected int	proposals Number of proposals suggested
protected java.lang.Double	scale Scale factor to multiply the sampler function by to envelop the conjunctive distribution.

Constructor Summary

Constructors

Constructor and Description
DefaultUpdater() Default constructor.
DefaultUpdater(BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive) Creates a new instance of DefaultUpdater
DefaultUpdater(BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive, java.lang.Double scale, ProbabilityFunction<ParameterType> sampler) Creates a new instance of DefaultUpdater
DefaultUpdater(BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive, ProbabilityFunction<ParameterType> sampler) Creates a new instance of DefaultUpdater

Method Summary

Modifier and Type	Method and Description
double	computeAcceptanceProbability(ParameterType parameter, java.lang.Iterable<? extends ObservationType> data) Computes the probability of accepting the parameter for the given data.
UnivariateGaussian.PDF	computeGaussianSampler(java.lang.Iterable<? extends ObservationType> data, java.util.Random random, int numSamples) Computes a Gaussian sample for the parameter, assuming it has is a Double, using importance sampling.
double	computeOptimalScale(java.lang.Iterable<? extends ObservationType> data) Computes the optimal scale factor for enveloping the conjunctive distribution with the sampler function given the data
BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>>	getConjuctive() Getter for conjunctive
int	getProposals() Getter for proposals
ProbabilityFunction<ParameterType>	getSampler() Getter for sampler
java.lang.Double	getScale() Getter for scale
ParameterType	makeProposal(java.util.Random random) Samples from the parameter prior
void	setConjuctive(BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive) Setter for conjunctive
protected void	setProposals(int proposals) Setter for proposals
void	setSampler(ProbabilityFunction<ParameterType> sampler) Setter for sampler
void	setScale(java.lang.Double scale) Setter for scale

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

Methods inherited from interface gov.sandia.cognition.util.CloneableSerializable

clone

Field Detail

conjuctive

protected BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive

Defines the parameter that connects the conditional and prior distributions.

scale

protected java.lang.Double scale

Scale factor to multiply the sampler function by to envelop the conjunctive distribution.

proposals

protected int proposals

Number of proposals suggested

Constructor Detail

DefaultUpdater

public DefaultUpdater()

Default constructor.

DefaultUpdater

public DefaultUpdater(BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive)

Creates a new instance of DefaultUpdater

Parameters:

conjuctive - Defines the parameter that connects the conditional and prior distributions.

DefaultUpdater

public DefaultUpdater(BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive,
                      ProbabilityFunction<ParameterType> sampler)

Creates a new instance of DefaultUpdater

Parameters:

conjuctive - Defines the parameter that connects the conditional and prior distributions.

sampler - Distribution from which we sample and envelop the conjunctive distribution.

DefaultUpdater

public DefaultUpdater(BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive,
                      java.lang.Double scale,
                      ProbabilityFunction<ParameterType> sampler)

Creates a new instance of DefaultUpdater

Parameters:

conjuctive - Defines the parameter that connects the conditional and prior distributions.

scale - Scale factor to multiply the sampler function by to envelop the conjunctive distribution.

sampler - Distribution from which we sample and envelop the conjunctive distribution.

Method Detail

computeAcceptanceProbability

public double computeAcceptanceProbability(ParameterType parameter,
                                           java.lang.Iterable<? extends ObservationType> data)

Description copied from interface: RejectionSampling.Updater

Computes the probability of accepting the parameter for the given data.

Specified by:

computeAcceptanceProbability in interface RejectionSampling.Updater<ObservationType,ParameterType>

Parameters:

parameter - Parameter to consider

data - Data to consider.

Returns:

Probability of accepting the parameter

makeProposal

public ParameterType makeProposal(java.util.Random random)

Description copied from interface: RejectionSampling.Updater

Samples from the parameter prior

Specified by:

makeProposal in interface RejectionSampling.Updater<ObservationType,ParameterType>

Parameters:

random - Random number generator.

Returns:

Location of the proposed sample

getConjuctive

public BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> getConjuctive()

Getter for conjunctive

Returns:

Defines the parameter that connects the conditional and prior distributions.

setConjuctive

public void setConjuctive(BayesianParameter<ParameterType,? extends ProbabilityFunction<ObservationType>,? extends ProbabilityFunction<ParameterType>> conjuctive)

Setter for conjunctive

Parameters:

conjuctive - Defines the parameter that connects the conditional and prior distributions.

computeOptimalScale

public double computeOptimalScale(java.lang.Iterable<? extends ObservationType> data)

Computes the optimal scale factor for enveloping the conjunctive distribution with the sampler function given the data

Parameters:

data - Data to consider

Returns:

optimal scale factor for enveloping the conjunctive distribution with the sampler function given the data

computeGaussianSampler

public UnivariateGaussian.PDF computeGaussianSampler(java.lang.Iterable<? extends ObservationType> data,
                                                     java.util.Random random,
                                                     int numSamples)

Computes a Gaussian sample for the parameter, assuming it has is a Double, using importance sampling.

Parameters:

data - (Sub)set of the data to use to estimate the Gaussian

random - Random number generator

numSamples - Number of samples to create the Gaussian... doesn't need to be very large.

Returns:

Gaussian that has the appropriate mean and variance to generate parameters.

getScale

public java.lang.Double getScale()

Getter for scale

Returns:

Scale factor to multiply the sampler function by to envelop the conjunctive distribution.

setScale

public void setScale(java.lang.Double scale)

Setter for scale

Parameters:

scale - Scale factor to multiply the sampler function by to envelop the conjunctive distribution.

getProposals

public int getProposals()

Getter for proposals

Returns:

Number of proposals suggested

setProposals

protected void setProposals(int proposals)

Setter for proposals

Parameters:

proposals - Number of proposals suggested

getSampler

public ProbabilityFunction<ParameterType> getSampler()

Getter for sampler

Returns:

Distribution from which we sample and envelop the conjunctive distribution.

setSampler

public void setSampler(ProbabilityFunction<ParameterType> sampler)

Setter for sampler

Parameters:

sampler - Distribution from which we sample and envelop the conjunctive distribution.