org.deeplearning4j.util

Class Viterbi

java.lang.Object

org.deeplearning4j.util.Viterbi

All Implemented Interfaces:

Serializable, Persistable

public class Viterbi
extends Object
implements Persistable

Based on the impl from: https://gist.github.com/rmcgibbo/3915977

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
Viterbi(org.nd4j.linalg.api.ndarray.INDArray possibleLabels) The possible outcomes for the chain.

Method Summary

Modifier and Type	Method and Description
Pair<Double,org.nd4j.linalg.api.ndarray.INDArray>	decode(org.nd4j.linalg.api.ndarray.INDArray labels) Decodes the given labels, assuming its a binary label matrix
Pair<Double,org.nd4j.linalg.api.ndarray.INDArray>	decode(org.nd4j.linalg.api.ndarray.INDArray labels, boolean binaryLabelMatrix) Decodes a series of labels
double	getLogMetaInstability()
double	getLogOfDiangnalTProb()
double	getLogPCorrect()
double	getLogPIncorrect()
double	getLogStates()
double	getMetaStability()
double	getpCorrect()
org.nd4j.linalg.api.ndarray.INDArray	getPossibleLabels()
int	getStates()
void	load(InputStream is)
void	setLogMetaInstability(double logMetaInstability)
void	setLogOfDiangnalTProb(double logOfDiangnalTProb)
void	setLogPCorrect(double logPCorrect)
void	setLogPIncorrect(double logPIncorrect)
void	setLogStates(double logStates)
void	setMetaStability(double metaStability)
void	setpCorrect(double pCorrect)
void	setPossibleLabels(org.nd4j.linalg.api.ndarray.INDArray possibleLabels)
void	setStates(int states)
void	write(OutputStream os)

Methods inherited from class java.lang.Object

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

Constructor Detail

Viterbi

public Viterbi(org.nd4j.linalg.api.ndarray.INDArray possibleLabels)

The possible outcomes for the chain. This should be the labels in the form of the possible outcomes (1,2,3) not the binarized label matrix

Parameters:

possibleLabels - the possible labels of the markov chain

Method Detail

decode

public Pair<Double,org.nd4j.linalg.api.ndarray.INDArray> decode(org.nd4j.linalg.api.ndarray.INDArray labels)

Decodes the given labels, assuming its a binary label matrix

Parameters:

labels - the labels as a binary label matrix

Returns:

the decoded labels and the most likely outcome of the sequence

decode

public Pair<Double,org.nd4j.linalg.api.ndarray.INDArray> decode(org.nd4j.linalg.api.ndarray.INDArray labels,
                                                                boolean binaryLabelMatrix)

Decodes a series of labels

Parameters:

labels - the labels to decode

binaryLabelMatrix - whether the label is a binary label matrix

Returns:

the most likely sequence and the sequence labels

write

public void write(OutputStream os)

Specified by:

write in interface Persistable

load

public void load(InputStream is)

Specified by:

load in interface Persistable

getMetaStability

public double getMetaStability()

setMetaStability

public void setMetaStability(double metaStability)

getpCorrect

public double getpCorrect()

setpCorrect

public void setpCorrect(double pCorrect)

getPossibleLabels

public org.nd4j.linalg.api.ndarray.INDArray getPossibleLabels()

setPossibleLabels

public void setPossibleLabels(org.nd4j.linalg.api.ndarray.INDArray possibleLabels)

getStates

public int getStates()

setStates

public void setStates(int states)

getLogPCorrect

public double getLogPCorrect()

setLogPCorrect

public void setLogPCorrect(double logPCorrect)

getLogPIncorrect

public double getLogPIncorrect()

setLogPIncorrect

public void setLogPIncorrect(double logPIncorrect)

getLogMetaInstability

public double getLogMetaInstability()

setLogMetaInstability

public void setLogMetaInstability(double logMetaInstability)

getLogOfDiangnalTProb

public double getLogOfDiangnalTProb()

setLogOfDiangnalTProb

public void setLogOfDiangnalTProb(double logOfDiangnalTProb)

getLogStates

public double getLogStates()

setLogStates

public void setLogStates(double logStates)