Package edu.stanford.nlp.classify

Class LogPrior

java.lang.Object

edu.stanford.nlp.classify.LogPrior

All Implemented Interfaces:

Serializable

public class LogPrior
extends Object
implements Serializable

A Prior for functions. Immutable.

Author:

Galen Andrew

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	LogPrior.LogPriorType

Constructor Summary

Constructors

Constructor	Description
LogPrior()
LogPrior(double[] C)	IMPORTANT NOTE: This constructor allows non-uniform regularization, but it transforms the inputs C (like the machine learning people like) to sigma (like we NLP folks like).
LogPrior(int intPrior)
LogPrior(int intPrior, double sigma, double epsilon)
LogPrior(LogPrior.LogPriorType type)
LogPrior(LogPrior.LogPriorType type, double sigma, double epsilon)

Method Summary

Modifier and Type	Method	Description
double	compute(double[] x, double[] grad)	Adjust the given grad array by adding the prior's gradient component and return the value of the logPrior
double	computeStochastic(double[] x, double[] grad, double fractionOfData)
static LogPrior	getAdaptationPrior(double[] means, LogPrior otherPrior)
double	getEpsilon()
double	getSigma()
double	getSigmaSquared()
double[]	getSigmaSquaredM()
LogPrior.LogPriorType	getType()
static LogPrior.LogPriorType	getType(String name)
void	setEpsilon(double epsilon)
void	setSigma(double sigma)
void	setSigmaSquared(double sigmaSq)
void	setSigmaSquaredM(double[] sigmaSq)

Methods inherited from class java.lang.Object

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

Constructor Detail

LogPrior

public LogPrior()

LogPrior

public LogPrior(int intPrior)

LogPrior

public LogPrior(LogPrior.LogPriorType type)

LogPrior

public LogPrior(int intPrior,
                double sigma,
                double epsilon)

LogPrior

public LogPrior(LogPrior.LogPriorType type,
                double sigma,
                double epsilon)

LogPrior

public LogPrior(double[] C)

IMPORTANT NOTE: This constructor allows non-uniform regularization, but it transforms the inputs C (like the machine learning people like) to sigma (like we NLP folks like). C = 1/\sigma^2

Method Detail

getType

public static LogPrior.LogPriorType getType(String name)

getAdaptationPrior

public static LogPrior getAdaptationPrior(double[] means,
                                          LogPrior otherPrior)

getType

public LogPrior.LogPriorType getType()

getSigma

public double getSigma()

getSigmaSquared

public double getSigmaSquared()

getSigmaSquaredM

public double[] getSigmaSquaredM()

getEpsilon

public double getEpsilon()

setSigma

public void setSigma(double sigma)

setSigmaSquared

public void setSigmaSquared(double sigmaSq)

setSigmaSquaredM

public void setSigmaSquaredM(double[] sigmaSq)

setEpsilon

public void setEpsilon(double epsilon)

computeStochastic

public double computeStochastic(double[] x,
                                double[] grad,
                                double fractionOfData)

compute

public double compute(double[] x,
                      double[] grad)

Adjust the given grad array by adding the prior's gradient component and return the value of the logPrior

Parameters:

x - the input point

grad - the gradient array

Returns:

the value