cc.mallet.fst.semi_supervised.constraints
Interface GEConstraint

All Known Implementing Classes:

OneLabelGEConstraints, OneLabelKLGEConstraints, OneLabelL2GEConstraints, OneLabelL2RangeGEConstraints, SelfTransitionGEConstraint, TwoLabelGEConstraints, TwoLabelKLGEConstraints, TwoLabelL2GEConstraints

public interface GEConstraint

Interface for GE constraint that considers either one or two states.

Author:

Gregory Druck

Method Summary

void	computeExpectations(java.util.ArrayList<SumLattice> lattices) Compute expectations using cached lattices.
GEConstraint	copy() This is used in multi-threading.
double	getCompositeConstraintFeatureValue(FeatureVector input, int inputPosition, int srcIndex, int destIndex) Computes the composite constraint feature value (over all constraint features) for FeatureVector fv and a transition from state li1 to li2.
double	getValue() Returns the total constraint value.
boolean	isOneStateConstraint()
void	preProcess(FeatureVector input) Gives the constraint the option to do some caching using only the FeatureVector.
java.util.BitSet	preProcess(InstanceList data)
void	setStateLabelMap(StateLabelMap map) Sets that map between the state indices and label indices.
void	zeroExpectations() Zero expectation values.

Method Detail

getCompositeConstraintFeatureValue

double getCompositeConstraintFeatureValue(FeatureVector input,
                                          int inputPosition,
                                          int srcIndex,
                                          int destIndex)

Computes the composite constraint feature value (over all constraint features) for FeatureVector fv and a transition from state li1 to li2.

Parameters:

input - FeatureVector on transition

inputPosition - Position of input in sequence

srcIndex - Source state index for transition

destIndex - Destination state index for transition

Returns:

Constraint feature value

getValue

double getValue()

Returns the total constraint value.

Returns:

Constraint value

computeExpectations

void computeExpectations(java.util.ArrayList<SumLattice> lattices)

Compute expectations using cached lattices.

Parameters:

lattices - Cached SumLattices

data - Unlabeled data

zeroExpectations

void zeroExpectations()

Zero expectation values. Called before re-computing gradient.

setStateLabelMap

void setStateLabelMap(StateLabelMap map)

Sets that map between the state indices and label indices.

Parameters:

map - StateLabelMap

copy

GEConstraint copy()

This is used in multi-threading.

Returns:

A copy of the GEConstraint.

isOneStateConstraint

boolean isOneStateConstraint()

Returns:

true if constraint feature only considers one state

preProcess

java.util.BitSet preProcess(InstanceList data)

Parameters:

data - Unlabeled data

Returns:

Returns a bitset of the size of the data, with the bit set if a constraint feature fires in that instance.

preProcess

void preProcess(FeatureVector input)

Gives the constraint the option to do some caching using only the FeatureVector. For example, the constrained input features could be cached.

Parameters:

input - FeatureVector input