sequence
Class LinearTagger

java.lang.Object
  sequence.LinearTagger

All Implemented Interfaces:

Serializable

public class LinearTagger
extends Object
implements Serializable

A linear model for sequence classification. It has the form h(x) = arg_max_y f(x,y).w where x and y are sequences of identical length, and f(x,y) decomposes over pairs of positions on y.

Author:

kuzman

See Also:

Serialized Form

Field Summary

double[]

w

Constructor Summary

LinearTagger(Alphabet xAlpha, Alphabet yAlpha, SequenceFeatureFunction fxy)

Method Summary

int[]	label(SparseVector[] x) use the Viterbi algorithm to find arg_max_y f(x,y) .
void	readObject(ObjectInputStream in)
double[][][]	scores(SparseVector[] x) at each position 0<=t
void	writeObject(ObjectOutputStream out)

Methods inherited from class java.lang.Object

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

Field Detail

w

public double[] w

Constructor Detail

LinearTagger

public LinearTagger(Alphabet xAlpha,
                    Alphabet yAlpha,
                    SequenceFeatureFunction fxy)

Method Detail

scores

public double[][][] scores(SparseVector[] x)

at each position 0<=t

Parameters:

x -

Returns:

result[t][ytm1][yt] = f(x,ytm1,yt).w

label

public int[] label(SparseVector[] x)

use the Viterbi algorithm to find arg_max_y f(x,y) . w

Parameters:

x -

Returns:

y that maximizes f(x,y) . w

writeObject

public void writeObject(ObjectOutputStream out)
                 throws IOException

Throws:

IOException

readObject

public void readObject(ObjectInputStream in)
                throws IOException,
                       ClassNotFoundException

Throws:

IOException

ClassNotFoundException