jnisvmlight
Class SVMLightInterface

java.lang.Object
  jnisvmlight.SVMLightInterface

public class SVMLightInterface
extends java.lang.Object

The main interface class that transfers the training data to the SVM-light library by a native call. Optionally takes as input an individually modified set of training parameters or an array of string paramaters that exactly simulate the command line input parameters used by the SVM-light binaries. This class can also be used for native classification calls.

Field Summary

protected TrainingParameters	m_tp
static boolean	SORT_INPUT_VECTORS Apply an in-place quicksort prior to each native training call to SVM-light.

Constructor Summary

SVMLightInterface()

Method Summary

double	classifyNative(FeatureVector doc) Performs a classifcation step as a native call to SVM-light.
static LabeledFeatureVector[]	getLabeledFeatureVectorsFromURL(java.net.URL file, int numOfLinesToSkip) Reads a set of labeled training vectors from a URL.
TrainingParameters	getTrainingParameters()
SVMLightModel	trainModel(LabeledFeatureVector[] trainingData)
SVMLightModel	trainModel(LabeledFeatureVector[] trainingData, java.lang.String[] argv)
SVMLightModel	trainModel(LabeledFeatureVector[] trainingData, TrainingParameters tp)

Methods inherited from class java.lang.Object

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

Field Detail

SORT_INPUT_VECTORS

public static boolean SORT_INPUT_VECTORS

Apply an in-place quicksort prior to each native training call to SVM-light. SVM-light requires each input feature vector to be sorted in ascending order of dimensions. Disable this option if you are sure to provide sorted vectors already.

m_tp

protected TrainingParameters m_tp

Constructor Detail

SVMLightInterface

public SVMLightInterface()

Method Detail

getLabeledFeatureVectorsFromURL

public static LabeledFeatureVector[] getLabeledFeatureVectorsFromURL(java.net.URL file,
                                                                     int numOfLinesToSkip)
                                                              throws java.text.ParseException

Reads a set of labeled training vectors from a URL. The format is compatible to the SVM-light training files.

Throws:

java.text.ParseException

classifyNative

public double classifyNative(FeatureVector doc)

Performs a classifcation step as a native call to SVM-light. If this method is used exlusively, no additional SVMLightModel object has to be kept in the Java runtime process.

getTrainingParameters

public TrainingParameters getTrainingParameters()

trainModel

public SVMLightModel trainModel(LabeledFeatureVector[] trainingData)

trainModel

public SVMLightModel trainModel(LabeledFeatureVector[] trainingData,
                                java.lang.String[] argv)

trainModel

public SVMLightModel trainModel(LabeledFeatureVector[] trainingData,
                                TrainingParameters tp)