org.encog.ml.hmm.alog

Class ForwardBackwardScaledCalculator

java.lang.Object

org.encog.ml.hmm.alog.ForwardBackwardCalculator

org.encog.ml.hmm.alog.ForwardBackwardScaledCalculator

public class ForwardBackwardScaledCalculator
extends ForwardBackwardCalculator

The forward-backward algorithm is an inference algorithm for hidden Markov models which computes the posterior marginals of all hidden state variables given a sequence of observations. This version makes use of scaling, and will not generate underflows with long sequences.

Nested Class Summary

Nested classes/interfaces inherited from class org.encog.ml.hmm.alog.ForwardBackwardCalculator

ForwardBackwardCalculator.Computation

Field Summary

Fields inherited from class org.encog.ml.hmm.alog.ForwardBackwardCalculator

alpha, beta, probability

Constructor Summary

Constructors

Constructor and Description
ForwardBackwardScaledCalculator(MLDataSet oseq, HiddenMarkovModel hmm)
ForwardBackwardScaledCalculator(MLDataSet oseq, HiddenMarkovModel hmm, java.util.EnumSet<ForwardBackwardCalculator.Computation> flags)

Method Summary

Modifier and Type	Method and Description
protected void	computeAlpha(HiddenMarkovModel hmm, MLDataSet oseq) Compute alpha.
protected void	computeBeta(HiddenMarkovModel hmm, MLDataSet oseq) Compute the beta step.
double	lnProbability()

Methods inherited from class org.encog.ml.hmm.alog.ForwardBackwardCalculator

alphaElement, betaElement, computeAlphaInit, computeAlphaStep, computeBetaStep, probability

Methods inherited from class java.lang.Object

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

Constructor Detail

ForwardBackwardScaledCalculator

public ForwardBackwardScaledCalculator(MLDataSet oseq,
                                       HiddenMarkovModel hmm)

ForwardBackwardScaledCalculator

public ForwardBackwardScaledCalculator(MLDataSet oseq,
                                       HiddenMarkovModel hmm,
                                       java.util.EnumSet<ForwardBackwardCalculator.Computation> flags)

Method Detail

computeAlpha

protected void computeAlpha(HiddenMarkovModel hmm,
                            MLDataSet oseq)

Description copied from class: ForwardBackwardCalculator

Compute alpha.

Overrides:

computeAlpha in class ForwardBackwardCalculator

Parameters:

hmm - The hidden markov model.

oseq - The sequence.

computeBeta

protected void computeBeta(HiddenMarkovModel hmm,
                           MLDataSet oseq)

Description copied from class: ForwardBackwardCalculator

Compute the beta step.

Overrides:

computeBeta in class ForwardBackwardCalculator

Parameters:

hmm - The hidden markov model.

oseq - The sequence.

lnProbability

public double lnProbability()