contextFold.learn
Class Learn
java.lang.Object
contextFold.learn.Learn
public class Learn
- extends java.lang.Object
Implements the machine-learning algorithm for inferring feature weight
parameters. The learning algorithm is based on the algorithm which is described
in Sections 8 and 9 of the paper:
Koby Crammer, Ofer Dekel, Joseph Keshet, Shai Shalev-Shwartz, and Yoram Singer.
Online Passive-Aggressive Algorithms. Journal of Machine Learning Research 7 (2006) 551–585.
Field Summary |
java.io.PrintWriter |
output
|
Method Summary |
void |
costSensitivePAUpdate(RNA rna,
int[] guessFold,
int[] goldFold,
float lt)
|
float |
doIter(int iter,
java.util.List<RNA> rnas,
int averageEvery,
int niters)
|
protected int[][] |
getFolds(StructreElementsScorer Gs)
|
void |
learn(int iters,
java.util.List<RNA> rnas,
java.lang.String weightsOutputFile,
int averageEvery,
boolean shuffle,
float requiredImproveFactor,
int maxNoImprovements,
boolean keepIterationWeights)
|
void |
updateKbest(int[][] guessFolds,
int[] goldFold,
float[] margins,
int iter,
int seqnum,
int seqlen,
RNA rna)
|
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
output
public java.io.PrintWriter output
Learn
public Learn(FeatureManager featureManager,
FoldEvaluator foldEvaluator,
int kbest,
LossCalculator lossCalculator)
getFolds
protected int[][] getFolds(StructreElementsScorer Gs)
doIter
public float doIter(int iter,
java.util.List<RNA> rnas,
int averageEvery,
int niters)
updateKbest
public void updateKbest(int[][] guessFolds,
int[] goldFold,
float[] margins,
int iter,
int seqnum,
int seqlen,
RNA rna)
costSensitivePAUpdate
public void costSensitivePAUpdate(RNA rna,
int[] guessFold,
int[] goldFold,
float lt)
learn
public void learn(int iters,
java.util.List<RNA> rnas,
java.lang.String weightsOutputFile,
int averageEvery,
boolean shuffle,
float requiredImproveFactor,
int maxNoImprovements,
boolean keepIterationWeights)