public GeneralizedExpectationObjectiveFunction(GeneralDataset <L, F> labeledDataset, IList <IDatum <L, F> > unlabeledDataList, IList <F> geFeatures)
{
System.Console.Out.WriteLine("Number of labeled examples:" + labeledDataset.size + "\nNumber of unlabeled examples:" + unlabeledDataList.Count);
System.Console.Out.WriteLine("Number of GE features:" + geFeatures.Count);
this.numFeatures = labeledDataset.NumFeatures();
this.numClasses = labeledDataset.NumClasses();
this.labeledDataset = labeledDataset;
this.unlabeledDataList = unlabeledDataList;
this.geFeatures = geFeatures;
this.classifier = new LinearClassifier <L, F>(null, labeledDataset.featureIndex, labeledDataset.labelIndex);
ComputeEmpiricalStatistics(geFeatures);
}
private double[] GetModelProbs(IDatum <L, F> datum)
{
double[] condDist = new double[labeledDataset.NumClasses()];
ICounter <L> probCounter = classifier.ProbabilityOf(datum);
foreach (L label in probCounter.KeySet())
{
int labelID = labeledDataset.labelIndex.IndexOf(label);
condDist[labelID] = probCounter.GetCount(label);
}
return(condDist);
}
public virtual MultinomialLogisticClassifier <L, F> TrainClassifier(GeneralDataset <L, F> dataset)
{
numClasses = dataset.NumClasses();
numFeatures = dataset.NumFeatures();
data = dataset.GetDataArray();
if (dataset is RVFDataset <object, object> )
{
dataValues = dataset.GetValuesArray();
}
else
{
dataValues = LogisticUtils.InitializeDataValues(data);
}
AugmentFeatureMatrix(data, dataValues);
labels = dataset.GetLabelsArray();
return(new MultinomialLogisticClassifier <L, F>(TrainWeights(), dataset.featureIndex, dataset.labelIndex));
}
// public static void main(String[] args) {
// List examples = new ArrayList();
// String leftLight = "leftLight";
// String rightLight = "rightLight";
// String broken = "BROKEN";
// String ok = "OK";
// Counter c1 = new ClassicCounter<>();
// c1.incrementCount(leftLight, 0);
// c1.incrementCount(rightLight, 0);
// RVFDatum d1 = new RVFDatum(c1, broken);
// examples.add(d1);
// Counter c2 = new ClassicCounter<>();
// c2.incrementCount(leftLight, 1);
// c2.incrementCount(rightLight, 1);
// RVFDatum d2 = new RVFDatum(c2, ok);
// examples.add(d2);
// Counter c3 = new ClassicCounter<>();
// c3.incrementCount(leftLight, 0);
// c3.incrementCount(rightLight, 1);
// RVFDatum d3 = new RVFDatum(c3, ok);
// examples.add(d3);
// Counter c4 = new ClassicCounter<>();
// c4.incrementCount(leftLight, 1);
// c4.incrementCount(rightLight, 0);
// RVFDatum d4 = new RVFDatum(c4, ok);
// examples.add(d4);
// Dataset data = new Dataset(examples.size());
// data.addAll(examples);
// NaiveBayesClassifier classifier = (NaiveBayesClassifier)
// new NaiveBayesClassifierFactory(200, 200, 1.0,
// LogPrior.LogPriorType.QUADRATIC.ordinal(),
// NaiveBayesClassifierFactory.CL)
// .trainClassifier(data);
// classifier.print();
// //now classifiy
// for (int i = 0; i < examples.size(); i++) {
// RVFDatum d = (RVFDatum) examples.get(i);
// Counter scores = classifier.scoresOf(d);
// System.out.println("for datum " + d + " scores are " + scores.toString());
// System.out.println(" class is " + Counters.topKeys(scores, 1));
// System.out.println(" class should be " + d.label());
// }
// }
// String trainFile = args[0];
// String testFile = args[1];
// NominalDataReader nR = new NominalDataReader();
// Map<Integer, Index<String>> indices = Generics.newHashMap();
// List<RVFDatum<String, Integer>> train = nR.readData(trainFile, indices);
// List<RVFDatum<String, Integer>> test = nR.readData(testFile, indices);
// System.out.println("Constrained conditional likelihood no prior :");
// for (int j = 0; j < 100; j++) {
// NaiveBayesClassifier<String, Integer> classifier = new NaiveBayesClassifierFactory<String, Integer>(0.1, 0.01, 0.6, LogPrior.LogPriorType.NULL.ordinal(), NaiveBayesClassifierFactory.CL).trainClassifier(train);
// classifier.print();
// //now classifiy
//
// float accTrain = classifier.accuracy(train.iterator());
// log.info("training accuracy " + accTrain);
// float accTest = classifier.accuracy(test.iterator());
// log.info("test accuracy " + accTest);
//
// }
// System.out.println("Unconstrained conditional likelihood no prior :");
// for (int j = 0; j < 100; j++) {
// NaiveBayesClassifier<String, Integer> classifier = new NaiveBayesClassifierFactory<String, Integer>(0.1, 0.01, 0.6, LogPrior.LogPriorType.NULL.ordinal(), NaiveBayesClassifierFactory.UCL).trainClassifier(train);
// classifier.print();
// //now classify
//
// float accTrain = classifier.accuracy(train.iterator());
// log.info("training accuracy " + accTrain);
// float accTest = classifier.accuracy(test.iterator());
// log.info("test accuracy " + accTest);
// }
// }
public virtual NaiveBayesClassifier <L, F> TrainClassifier(GeneralDataset <L, F> dataset)
{
if (dataset is RVFDataset)
{
throw new Exception("Not sure if RVFDataset runs correctly in this method. Please update this code if it does.");
}
return(TrainClassifier(dataset.GetDataArray(), dataset.labels, dataset.NumFeatures(), dataset.NumClasses(), dataset.labelIndex, dataset.featureIndex));
}
public BiasedLogConditionalObjectiveFunction(GeneralDataset <object, object> dataset, double[][] confusionMatrix, LogPrior prior)
: this(dataset.NumFeatures(), dataset.NumClasses(), dataset.GetDataArray(), dataset.GetLabelsArray(), confusionMatrix, prior)
{
}
public virtual SVMLightClassifier <L, F> TrainClassifierBasic(GeneralDataset <L, F> dataset)
{
IIndex <L> labelIndex = dataset.LabelIndex();
IIndex <F> featureIndex = dataset.featureIndex;
bool multiclass = (dataset.NumClasses() > 2);
try
{
// this is the file that the model will be saved to
File modelFile = File.CreateTempFile("svm-", ".model");
if (deleteTempFilesOnExit)
{
modelFile.DeleteOnExit();
}
// this is the file that the svm light formated dataset
// will be printed to
File dataFile = File.CreateTempFile("svm-", ".data");
if (deleteTempFilesOnExit)
{
dataFile.DeleteOnExit();
}
// print the dataset
PrintWriter pw = new PrintWriter(new FileWriter(dataFile));
dataset.PrintSVMLightFormat(pw);
pw.Close();
// -v 0 makes it not verbose
// -m 400 gives it a larger cache, for faster training
string cmd = (multiclass ? svmStructLearn : (useSVMPerf ? svmPerfLearn : svmLightLearn)) + " -v " + svmLightVerbosity + " -m 400 ";
// set the value of C if we have one specified
if (C > 0.0)
{
cmd = cmd + " -c " + C + " ";
}
else
{
// C value
if (useSVMPerf)
{
cmd = cmd + " -c " + 0.01 + " ";
}
}
//It's required to specify this parameter for SVM perf
// Alpha File
if (useAlphaFile)
{
File newAlphaFile = File.CreateTempFile("svm-", ".alphas");
if (deleteTempFilesOnExit)
{
newAlphaFile.DeleteOnExit();
}
cmd = cmd + " -a " + newAlphaFile.GetAbsolutePath();
if (alphaFile != null)
{
cmd = cmd + " -y " + alphaFile.GetAbsolutePath();
}
alphaFile = newAlphaFile;
}
// File and Model Data
cmd = cmd + " " + dataFile.GetAbsolutePath() + " " + modelFile.GetAbsolutePath();
if (verbose)
{
logger.Info("<< " + cmd + " >>");
}
/*Process p = Runtime.getRuntime().exec(cmd);
*
* p.waitFor();
*
* if (p.exitValue() != 0) throw new RuntimeException("Error Training SVM Light exit value: " + p.exitValue());
* p.destroy(); */
SystemUtils.Run(new ProcessBuilder(whitespacePattern.Split(cmd)), new PrintWriter(System.Console.Error), new PrintWriter(System.Console.Error));
if (doEval)
{
File predictFile = File.CreateTempFile("svm-", ".pred");
if (deleteTempFilesOnExit)
{
predictFile.DeleteOnExit();
}
string evalCmd = (multiclass ? svmStructClassify : (useSVMPerf ? svmPerfClassify : svmLightClassify)) + " " + dataFile.GetAbsolutePath() + " " + modelFile.GetAbsolutePath() + " " + predictFile.GetAbsolutePath();
if (verbose)
{
logger.Info("<< " + evalCmd + " >>");
}
SystemUtils.Run(new ProcessBuilder(whitespacePattern.Split(evalCmd)), new PrintWriter(System.Console.Error), new PrintWriter(System.Console.Error));
}
// read in the model file
Pair <double, ClassicCounter <int> > weightsAndThresh = ReadModel(modelFile, multiclass);
double threshold = weightsAndThresh.First();
ClassicCounter <Pair <F, L> > weights = ConvertWeights(weightsAndThresh.Second(), featureIndex, labelIndex, multiclass);
ClassicCounter <L> thresholds = new ClassicCounter <L>();
if (!multiclass)
{
thresholds.SetCount(labelIndex.Get(0), -threshold);
thresholds.SetCount(labelIndex.Get(1), threshold);
}
SVMLightClassifier <L, F> classifier = new SVMLightClassifier <L, F>(weights, thresholds);
if (doEval)
{
File predictFile = File.CreateTempFile("svm-", ".pred2");
if (deleteTempFilesOnExit)
{
predictFile.DeleteOnExit();
}
PrintWriter pw2 = new PrintWriter(predictFile);
NumberFormat nf = NumberFormat.GetNumberInstance();
nf.SetMaximumFractionDigits(5);
foreach (IDatum <L, F> datum in dataset)
{
ICounter <L> scores = classifier.ScoresOf(datum);
pw2.Println(Counters.ToString(scores, nf));
}
pw2.Close();
}
if (useSigmoid)
{
if (verbose)
{
System.Console.Out.Write("fitting sigmoid...");
}
classifier.SetPlatt(FitSigmoid(classifier, dataset));
if (verbose)
{
System.Console.Out.WriteLine("done");
}
}
return(classifier);
}
catch (Exception e)
{
throw new Exception(e);
}
}
