edu.stanford.nlp.maxent.iis
Class LambdaSolve

java.lang.Object
  |
  +--edu.stanford.nlp.maxent.iis.LambdaSolve

Direct Known Subclasses:

LambdaSolveBinary, LambdaSolveTagger

public class LambdaSolve

extends Object

This is the main class that does the core computation in IIS.

Field Summary

double	eps
boolean	fixedFnumXY This flag is true if all (x,y) have the same f# in which case the newton equation solving is avoided
byte[][]	fnumArr This contains the number of features active for each pair (x,y)
double[]	ftieldArr This is an array of empirical expectations for the features
double[]	lambda These are the model parameters that have to be learned
boolean[]	lambda_converged
double	newtonerr
Problem	p
double[][]	probConds Conditional probabilities.
static boolean	smooth
double[]	zlambda Normalization factors, one for each x

Constructor Summary

LambdaSolve()

LambdaSolve(Problem p1, double eps1, double nerr1)

LambdaSolve(String filename)
reads the prameters from a file

Method Summary

boolean	checkCorrectness() Check whether the constraints are satisfied, the probabilities sum to one, etc.
protected double	fnum(int x, int y)
double[]	getDerivatives() assuming we have the lambdas in the array and we need only the derivatives now
void	ImprovedIterative() Iterate until convergence.
void	ImprovedIterative(int iters) Does a fixed number of IIS iterations
double	logLikelihood()
double	logLikelihoodScratch() calculate the log likelihood from scrath , hashing the conditional probabilities in pcond which we will use for the gradient later
static void	main(String[] args) With arguments, this will print out the lambda paramters of a bunch of .lam files (which are assumed to all be the same size).
double	pcond(int y, int x)
void	print() Print out p(y|x) for all pairs to the standard output.
void	read(String filename) Read the model parameters from a file ( only the lambdas )
void	save_lambdas(String filename) Writes the lambdas to the file.
void	save(String filename) Saves xSize,ySize and the lambdas to the file.

Methods inherited from class java.lang.Object

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

Field Detail

lambda

public double[] lambda

These are the model parameters that have to be learned

lambda_converged

public boolean[] lambda_converged

eps

public double eps

newtonerr

public double newtonerr

fixedFnumXY

public boolean fixedFnumXY

This flag is true if all (x,y) have the same f# in which case the newton equation solving is avoided

p

public Problem p

probConds

public double[][] probConds

Conditional probabilities.

zlambda

public double[] zlambda

Normalization factors, one for each x

fnumArr

public byte[][] fnumArr

This contains the number of features active for each pair (x,y)

ftieldArr

public double[] ftieldArr

This is an array of empirical expectations for the features

smooth

public static boolean smooth

Constructor Detail

LambdaSolve

public LambdaSolve(Problem p1,
                   double eps1,
                   double nerr1)

LambdaSolve

public LambdaSolve(String filename)

reads the prameters from a file

LambdaSolve

public LambdaSolve()

Method Detail

ImprovedIterative

public void ImprovedIterative()

Iterate until convergence. I usually use the other method that does a fixed number of iterations.

ImprovedIterative

public void ImprovedIterative(int iters)

Does a fixed number of IIS iterations

pcond

public double pcond(int y,
                    int x)

fnum

protected double fnum(int x,
                      int y)

checkCorrectness

public boolean checkCorrectness()

Check whether the constraints are satisfied, the probabilities sum to one, etc. Prints out a message if there is something wrong.

print

public void print()

Print out p(y|x) for all pairs to the standard output.

save

public void save(String filename)

Saves xSize,ySize and the lambdas to the file. Can be read later with read(String).

save_lambdas

public void save_lambdas(String filename)

Writes the lambdas to the file. Can be read later with readL.

read

public void read(String filename)

Read the model parameters from a file ( only the lambdas )

logLikelihood

public double logLikelihood()

Returns:

The loglikelihood of the empirical distribution as predicted by the model p.

logLikelihoodScratch

public double logLikelihoodScratch()

calculate the log likelihood from scrath , hashing the conditional probabilities in pcond which we will use for the gradient later

getDerivatives

public double[] getDerivatives()

assuming we have the lambdas in the array and we need only the derivatives now

main

public static void main(String[] args)

With arguments, this will print out the lambda paramters of a bunch of .lam files (which are assumed to all be the same size).