Examples of RDFWLSubTreeKernel

org.data2semantics.proppred.kernels.rdfgraphkernels.RDFWLSubTreeKernel
This class implements a WL kernel directly on an RDF graph. The difference with a normal WL kernel is that subgraphs are not explicitly extracted. However we use the idea of subgraph implicitly by tracking for each vertex/edge the distance from an instance vertex. For one thing, this leads to the fact that 1 black list is applied to the entire RDF graph, instead of 1 (small) blacklist per graph. @author Gerben

Examples of org.data2semantics.proppred.kernels.rdfgraphkernels.RDFWLSubTreeKernel

    for (int i : depths) {      
      for (int it : iterations) {
        resTable.newRow("");


        LibLINEARParameters linParms = new LibLINEARParameters(LibLINEARParameters.SVC_DUAL, cs);
        KernelExperiment<RDFFeatureVectorKernel> exp = new RDFLinearKernelExperiment(new RDFWLSubTreeKernel(it, i, inference, true), seeds, linParms, dataset, instances, target, blackList, evalFuncs);


        System.out.println("Running WL RDF: " + i + " " + it);
        exp.run();


        for (Result res : exp.getResults()) {

View Full Code Here

Examples of org.data2semantics.proppred.kernels.rdfgraphkernels.RDFWLSubTreeKernel

  
    // create a list of doubles as train target
    List<Double> target = EvaluationUtils.createTarget(labels);


    RDFPairKernel kernel = new RDFPairKernel();
    RDFWLSubTreeKernel k1 = new RDFWLSubTreeKernel(4,2,true,true);
    RDFWLSubTreeKernel k2 = new RDFWLSubTreeKernel(4,2,true,true);
    
    // Compute the kernel
    double[][] matrix = kernel.compute(dataset, instances, blacklist, k1, k2);
    
    double[] cs = {0.001, 0.01, 0.1, 1, 10, 100, 1000};

View Full Code Here

Examples of org.data2semantics.proppred.kernels.rdfgraphkernels.RDFWLSubTreeKernel

    
    // We remove classes with fewer than 5 instances
    EvaluationUtils.removeSmallClasses(instances, labels, 5);
    
    // Create the RDFFeatureVectorKernel that we are going to use
    RDFFeatureVectorKernel kernel = new RDFWLSubTreeKernel(6,3,true,true);
    
    // Compute feature vectors
    SparseVector[] featureVectors = kernel.computeFeatureVectors(dataset, instances, blackList);
    
    // Create a list of doubles as target, with our labelMap, so that we can use it later on (i.e. get the reverseMap)
    Map<Value, Double> labelMap = new HashMap<Value, Double>();
    List<Double> target = EvaluationUtils.createTarget(labels, labelMap);

View Full Code Here

0 1 2

TOP

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.