cc.mallet.fst
Class SumLatticeFactory

java.lang.Object
  cc.mallet.fst.SumLatticeFactory

All Implemented Interfaces:

java.io.Serializable

Direct Known Subclasses:

SumLatticeBeam.Factory, SumLatticeDefault.Factory, SumLatticeScaling.Factory

public abstract class SumLatticeFactory
extends java.lang.Object
implements java.io.Serializable

Provides factory methods to create inference engine for training a transducer.

See Also:

Serialized Form

Constructor Summary

SumLatticeFactory()

Method Summary

SumLattice	newSumLattice(Transducer trans, Sequence input)
SumLattice	newSumLattice(Transducer trans, Sequence input, Sequence output)
SumLattice	newSumLattice(Transducer trans, Sequence input, Sequence output, Transducer.Incrementor incrementor)
SumLattice	newSumLattice(Transducer trans, Sequence input, Sequence output, Transducer.Incrementor incrementor, boolean saveXis)
abstract SumLattice	newSumLattice(Transducer trans, Sequence input, Sequence output, Transducer.Incrementor incrementor, boolean saveXis, LabelAlphabet outputAlphabet) Returns a SumLattice object to run forward-backward.
SumLattice	newSumLattice(Transducer trans, Sequence input, Sequence output, Transducer.Incrementor incrementor, LabelAlphabet outputAlphabet)
SumLattice	newSumLattice(Transducer trans, Sequence input, Transducer.Incrementor incrementor)

Methods inherited from class java.lang.Object

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

Constructor Detail

SumLatticeFactory

public SumLatticeFactory()

Method Detail

newSumLattice

public SumLattice newSumLattice(Transducer trans,
                                Sequence input)

newSumLattice

public SumLattice newSumLattice(Transducer trans,
                                Sequence input,
                                Transducer.Incrementor incrementor)

newSumLattice

public SumLattice newSumLattice(Transducer trans,
                                Sequence input,
                                Sequence output)

newSumLattice

public SumLattice newSumLattice(Transducer trans,
                                Sequence input,
                                Sequence output,
                                Transducer.Incrementor incrementor)

newSumLattice

public SumLattice newSumLattice(Transducer trans,
                                Sequence input,
                                Sequence output,
                                Transducer.Incrementor incrementor,
                                LabelAlphabet outputAlphabet)

newSumLattice

public SumLattice newSumLattice(Transducer trans,
                                Sequence input,
                                Sequence output,
                                Transducer.Incrementor incrementor,
                                boolean saveXis)

newSumLattice

public abstract SumLattice newSumLattice(Transducer trans,
                                         Sequence input,
                                         Sequence output,
                                         Transducer.Incrementor incrementor,
                                         boolean saveXis,
                                         LabelAlphabet outputAlphabet)

Returns a SumLattice object to run forward-backward.

Parameters:

trans - Transducer model

input - Input sequence

output - If output is null then the forward-backward is not constrained to match the output

incrementor - If null then do not update the weights

saveXis - If true then save the transition weights as well

outputAlphabet - If outputAlphabet is non-null, this will create a LabelVector for each position in the output sequence indicating the probability distribution over possible outputs at that time index.