Examples of cc.mallet.types.InstanceList

cc.mallet.types.InstanceList
A list of machine learning instances, typically used for training or testing of a machine learning algorithm.
All of the instances in the list will have been passed through the same {@link cc.mallet.pipe.Pipe}, and thus must also share the same data and target Alphabets. InstanceList keeps a reference to the pipe and the two alphabets.
The most common way of adding instances to an InstanceList is through the add(PipeInputIterator) method. PipeInputIterators are a way of mapping general data sources into instances suitable for processing through a pipe. As each {@link cc.mallet.types.Instance} is pulled from the PipeInputIterator, the InstanceListcopies the instance and runs the copy through its pipe (with resultant destructive modifications) before saving the modified instance on its list. This is the usual way in which instances are transformed by pipes.
InstanceList also contains methods for randomly generating lists of feature vectors; splitting lists into non-overlapping subsets (useful for test/train splits), and iterators for cross validation. @see Instance @see Pipe @author Andrew McCallum mccallum@cs.umass.edu

                          clustering.getNumClusters(),
                          newLabels);
  }
  
  public static Clustering mergeInstancesWithSameLabel (Clustering clustering) {
    InstanceList list = clustering.getInstances();
    for (int i = 0; i < list.size(); i++) {
      Instance ii = list.get(i);
      int li = clustering.getLabel(i);
      for (int j = i + 1; j < list.size(); j++) {
        Instance ij = list.get(j);
        int lj = clustering.getLabel(j);
        if (li != lj && ii.getLabeling().equals(ij.getLabeling()))
          clustering = ClusterUtils.mergeClusters(clustering, li, lj);
      }
    }

View Full Code Here

    
    // TRAIN
    Clustering training = sampleClustering(alphabet);    
    Pipe clusterPipe = new OverlappingFeaturePipe();
    System.err.println("Training with " + training);
    InstanceList trainList = new InstanceList(clusterPipe);
    trainList.addThruPipe(new ClusterSampleIterator(training, random, 0.5, 100));
    System.err.println("Created " + trainList.size() + " instances.");
    Classifier me = new MaxEntTrainer().train(trainList);
    ClassifyingNeighborEvaluator eval =
      new ClassifyingNeighborEvaluator(me, "YES");
                                           
    Trial trial = new Trial(me, trainList);
    System.err.println(new ConfusionMatrix(trial));
    InfoGain ig = new InfoGain(trainList);
    ig.print();


//     Clusterer clusterer = new GreedyAgglomerative(training.getInstances().getPipe(),
//                                                   eval, 0.5);
    Clusterer clusterer = new GreedyAgglomerativeByDensity(training.getInstances().getPipe(),
                                                           eval, 0.5, false,
                                                           new java.util.Random(1));


    // TEST
    Clustering testing = sampleClustering(alphabet);    
    InstanceList testList = testing.getInstances();
    Clustering predictedClusters = clusterer.cluster(testList);      


    // EVALUATE
    System.err.println("\n\nEvaluating System: " + clusterer);
    ClusteringEvaluators evaluators = new ClusteringEvaluators(new ClusteringEvaluator[]{

View Full Code Here

   * Sample a InstanceList and its true clustering.
   * @param alph
   * @return
   */
  private Clustering sampleClustering (Alphabet alph) {
    InstanceList instances =
      new InstanceList(random,
                       alph,
                       new String[]{"foo", "bar"},
                       30).subList(0, 20);
    Clustering singletons = ClusterUtils.createSingletonClustering(instances);
    // Merge instances that both have feature0
    for (int i = 0; i < instances.size(); i++) {
      FeatureVector fvi = (FeatureVector)instances.get(i).getData();
      for (int j = i + 1; j < instances.size(); j++) {
        FeatureVector fvj = (FeatureVector)instances.get(j).getData();
        if (fvi.contains("feature0") && fvj.contains("feature0")) {
          singletons = ClusterUtils.mergeClusters(singletons,
                                                  singletons.getLabel(i),
                                                  singletons.getLabel(j));
        } else if (!(fvi.contains("feature0") || fvj.contains("feature0"))

View Full Code Here

    return clusterings;
  }
  
  public void testEvaluators ()
  {
    InstanceList instances = new InstanceList(new Randoms(1), 100, 2).subList(0,12);
    System.err.println(instances.size() + " instances");
    Clustering truth = generateTruth(instances);
    System.err.println("truth=" + truth);


    Clustering[] predicted = generatePredicted(instances);
    ClusteringEvaluator pweval = new PairF1Evaluator();

View Full Code Here

    public Instance pipe (Instance carrier) {
      boolean mergeFirst = false;
      
      AgglomerativeNeighbor neighbor = (AgglomerativeNeighbor)carrier.getData();
      Clustering original = neighbor.getOriginal();
      InstanceList list = original.getInstances();      
      int[] mergedIndices = neighbor.getNewCluster();
      boolean match = true;
      for (int i = 0; i < mergedIndices.length; i++) {
        for (int j = i + 1; j < mergedIndices.length; j++) {
          if ((original.getLabel(mergedIndices[i]) !=
               original.getLabel(mergedIndices[j])) || mergeFirst) {
            FeatureVector fvi = (FeatureVector)list.get(mergedIndices[i]).getData();
            FeatureVector fvj = (FeatureVector)list.get(mergedIndices[j]).getData();
            if (!(fvi.contains("feature0") && fvj.contains("feature0"))) {
              match = false;
              break;              
            }
          }

View Full Code Here


    /** Returns the <code>Instance</code> at the specified index. If
     * this Instance is not in memory, swap a block of instances back
     * into memory. */
    public Instance get (int index) {
        InstanceList page = getPageForIndex (index, false);
        return page.get (index % this.instancesPerPage);
    }

View Full Code Here


    /** Replaces the <code>Instance</code> at position
     * <code>index</code> with a new one. Note that this is the only
     * sanctioned way of changing an Instance. */
    public Instance set (int index, Instance instance) {
        InstanceList page = getPageForIndex (index, true);
        return page.set (index % this.instancesPerPage, instance);
    }

View Full Code Here

        this.collectGarbage = b;
    
    }


    public InstanceList shallowClone () {
        InstanceList ret = this.cloneEmpty ();
        for (int i = 0; i < this.size (); i++) {
            ret.add (get (i));
        }
        return ret;
    }

View Full Code Here

     * @throws IOException
     * @throws ClassNotFoundException
     */
    private InstanceList deserializePage(ObjectInputStream in)
    throws IOException, ClassNotFoundException {
        InstanceList page = new InstanceList(noopPipe);
        int size = in.readInt();
        
        for (int i = 0; i < size; i++) {
            Object data = deserializeObject (in);
            Object target = deserializeObject (in);
            Object name = in.readObject ();
            Object source = in.readObject ();
            double weight = in.readDouble ();
            page.add (new Instance (data, target, name, source), weight);
        }
        
        return page;
    }

View Full Code Here

   {
     Pipe pipe = TestMEMM.makeSpacePredictionPipe ();
     String[] data0 = { TestCRF.data[0] };
     String[] data1 = { TestCRF.data[1] };


     InstanceList training = new InstanceList (pipe);
     training.addThruPipe (new ArrayIterator (data0));
     InstanceList testing = new InstanceList (pipe);
     testing.addThruPipe (new ArrayIterator (data1));


     CRF crf = new CRF (pipe, null);
     crf.addFullyConnectedStatesForLabels ();
     CRFTrainerByLabelLikelihood crft = new CRFTrainerByLabelLikelihood (crf);
     crft.trainIncremental (training);

View Full Code Here

0 1 2 3 4 5 6 7 8 9

TOP

Related Classes of cc.mallet.types.InstanceList

cc.mallet.classify.C45$Node

cc.mallet.classify.DecisionTree$Node

cc.mallet.classify.MaxEntPRTrainer

cc.mallet.classify.NaiveBayesEMTrainer

cc.mallet.classify.tui.Vectors2FeatureConstraints

cc.mallet.cluster.Clustering

cc.mallet.cluster.evaluate.BCubedEvaluator

cc.mallet.cluster.evaluate.tests.TestClusteringEvaluators

cc.mallet.cluster.examples.FirstOrderClusterExample

cc.mallet.cluster.examples.FirstOrderClusterExample$OverlappingFeaturePipe

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.