de.jstacs.classifiers.differentiableSequenceScoreBased.logPrior
Class SeparateLaplaceLogPrior

java.lang.Object
  de.jstacs.algorithms.optimization.DifferentiableFunction
      de.jstacs.classifiers.differentiableSequenceScoreBased.logPrior.LogPrior
          de.jstacs.classifiers.differentiableSequenceScoreBased.logPrior.SeparateLogPrior
              de.jstacs.classifiers.differentiableSequenceScoreBased.logPrior.SeparateLaplaceLogPrior

All Implemented Interfaces:

Function, Storable, Cloneable

public class SeparateLaplaceLogPrior
extends SeparateLogPrior

Class for a LogPrior that defines a Laplace prior on the parameters of a set of DifferentiableStatisticalModels and a set of class parameters. The scale hyperparameters of the Laplace prior are determined from a set of class-specific base variances and the class variances, respectively. The formula to determine the shape hyperparameter b[i] for a parameter i from a variance v[i] is
b[i] = Math.sqrt(v[i]/2).
The variances v[i] for a parameter i of a DifferentiableStatisticalModel fun[j] are determined from the base variance v[j] as
v[i] = v[j]*funs[j].getSizeOfEventSpaceForRandomVariablesOfParameter(j) .
The mean parameters are set to 0 for the parameters of the DifferentiableStatisticalModels and to the user-specified means for the class parameters.

Author:

Jan Grau, Jens Keilwagen

Field Summary

Fields inherited from class de.jstacs.classifiers.differentiableSequenceScoreBased.logPrior.SeparateLogPrior

classMus, classVars, freeParameters, funs, vars

Fields inherited from class de.jstacs.classifiers.differentiableSequenceScoreBased.logPrior.LogPrior

UNKNOWN

Constructor Summary

SeparateLaplaceLogPrior(double[] vars, double[] classVars, double[] classMus)
Creates a new SeparateLaplaceLogPrior from a set of base variances vars, a set of class variances classVars and a set of class means classMus.

SeparateLaplaceLogPrior(StringBuffer xml)
The standard constructor for the interface Storable.

Method Summary

void	addGradientFor(double[] params, double[] vector) Adds the gradient of the log-prior using the current parameters to a given vector.
double	evaluateFunction(double[] x) Evaluates the function at a certain vector (in mathematical sense) x.
int	getDimensionOfScope() Returns the dimension of the scope of the Function.
String	getInstanceName() Returns a short instance name.
protected void	unset() Resets all internally pre-computed values, e.g.

Methods inherited from class de.jstacs.classifiers.differentiableSequenceScoreBased.logPrior.SeparateLogPrior

getNewInstance, set, toXML

Methods inherited from class de.jstacs.classifiers.differentiableSequenceScoreBased.logPrior.LogPrior

evaluateGradientOfFunction

Methods inherited from class de.jstacs.algorithms.optimization.DifferentiableFunction

findOneDimensionalMin

Methods inherited from class java.lang.Object

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

Constructor Detail

SeparateLaplaceLogPrior

public SeparateLaplaceLogPrior(double[] vars,
                               double[] classVars,
                               double[] classMus)

Creates a new SeparateLaplaceLogPrior from a set of base variances vars, a set of class variances classVars and a set of class means classMus.

Parameters:

vars - the base variances for each class

classVars - the class variances

classMus - the class means

See Also:

SeparateLogPrior.SeparateLogPrior(double[], double[], double[])

SeparateLaplaceLogPrior

public SeparateLaplaceLogPrior(StringBuffer xml)
                        throws NonParsableException

The standard constructor for the interface Storable. Creates a new SeparateLaplaceLogPrior out of its XML representation.

Parameters:

xml - the XML representation as StringBuffer

Throws:

NonParsableException - if the SeparateLaplaceLogPrior could not be reconstructed out of the XML representation (the StringBuffer could not be parsed)

See Also:

SeparateLogPrior.SeparateLogPrior(StringBuffer), Storable

Method Detail

unset

protected void unset()

Description copied from class: SeparateLogPrior

Resets all internally pre-computed values, e.g. the hyperparameters for each parameter.

Specified by:

unset in class SeparateLogPrior

addGradientFor

public void addGradientFor(double[] params,
                           double[] vector)

Description copied from class: LogPrior

Adds the gradient of the log-prior using the current parameters to a given vector.

Specified by:

addGradientFor in class LogPrior

Parameters:

params - the parameters

vector - the vector

evaluateFunction

public double evaluateFunction(double[] x)
                        throws DimensionException,
                               EvaluationException

Description copied from interface: Function

Evaluates the function at a certain vector (in mathematical sense) x.

Parameters:

x - the current vector

Returns:

the evaluation of the function

Throws:

DimensionException - if dim(x) != n, with

EvaluationException - if there was something wrong during the evaluation of the function

getDimensionOfScope

public int getDimensionOfScope()

Description copied from interface: Function

Returns the dimension of the scope of the Function.

Returns:

the dimension of the scope: n with

getInstanceName

public String getInstanceName()

Description copied from class: LogPrior

Returns a short instance name.

Specified by:

getInstanceName in class LogPrior

Returns:

a short instance name