public void IrisSupportVectorMachineClassifierTest()
{
DataSetLoader dataSetLoader = new DataSetLoader();
Console.WriteLine(" Reading DataSet.. ");
var irises = dataSetLoader.SelectIrises();
for (double i = 0; i < 1; i = i + 1)
{
Kernel kernel = new LinearKernel();
SVMClassifier animalSVMClassifier = new SVMClassifier(irises, kernel, 0.001, 10.0);
animalSVMClassifier.Train();
var irisesTest = dataSetLoader.SelectIrises();
var trueCounter = 0;
var counter = 0;
foreach (var item in irisesTest)
{
var outputValue = animalSVMClassifier.Classify(item.Item1);
if (outputValue == item.Item2)
{
trueCounter++;
}
Debug.WriteLine(string.Format("Value {0} - Predicted {1} = {2}",
item.Item2, outputValue, (outputValue == item.Item2) ? "true" : "false"));
counter++;
}
Debug.WriteLine(string.Format(" i = {0} Data {1} - True {2} Verhältnis: {3}", i,
counter.ToString(), trueCounter.ToString(), (Convert.ToDouble(trueCounter) / Convert.ToDouble(counter)).ToString()));
}
}
public void AnimalDualPerceptronTest()
{
DataSetLoader dataSetLoader = new DataSetLoader();
Console.WriteLine(" Reading DataSet.. ");
var animals = dataSetLoader.SelectAnimals();
for (double i = 0; i < 1; i = i + 1)
{
Kernel kernel = new LinearKernel();
DualPerceptronClassifier dualPerceptronClassifier = new DualPerceptronClassifier(animals, kernel);
dualPerceptronClassifier.Train();
var animalsTest = dataSetLoader.SelectAnimals();
var trueCounter = 0;
var counter = 0;
foreach (var item in animalsTest)
{
var outputValue = dualPerceptronClassifier.Classify(item.Item1);
if (outputValue == item.Item2)
{
trueCounter++;
}
Debug.WriteLine(string.Format("Value {0} - Predicted {1} = {2}",
item.Item2, outputValue, (outputValue == item.Item2) ? "true" : "false"));
counter++;
}
Debug.WriteLine(string.Format(" i = {0} Data {1} - True {2} Verhältnis: {3}", i,
counter.ToString(), trueCounter.ToString(), (Convert.ToDouble(trueCounter) / Convert.ToDouble(counter)).ToString()));
}
}
private void DualPerceptron(List <Tuple <double[], double> > data)
{
Kernel kernel = new LinearKernel();
foreach (var item in netMLObject.Options)
{
if (item == "linearkernel")
{
kernel = new LinearKernel();
}
else if (item == "gaussiankernel")
{
kernel = new GaussianKernel(1.0);
}
else if (item == "polynomialkernel")
{
kernel = new PolynomialKernel(1);
}
else if (item == "logitkernel")
{
kernel = new LogitKernel();
}
else if (item == "tanhkernel")
{
kernel = new TanhKernel();
}
}
classification = new DualPerceptronClassifier(data, kernel);
}
public void CreditDataClassifyMethod()
{
DataSetLoader dataSetLoader = new DataSetLoader();
var creditData = dataSetLoader.SelectCreditData();
var data = dataSetLoader.CalculatePercent(100, creditData);
DecisionTreeClassifier decisionTreeClassifier =
new DecisionTreeClassifier(data.Item1, new ShannonEntropySplitter());
NaiveBayesClassifier naiveBayes =
new NaiveBayesClassifier(data.Item1);
var list = new List <NetML.Classification>();
Kernel kernel = new LinearKernel();
SVMClassifier SVMClassifier =
new SVMClassifier(creditData, kernel, 0.001, 10.0);
var neuronalCreditData = dataSetLoader.SelectNeuronalNetworksCreditData();
NeuronalNetworkClassifier neuronalNetworkClassifier =
new NeuronalNetworkClassifier(neuronalCreditData, 20, 2, 20, 5000, 0.1);
list.Add(decisionTreeClassifier);
list.Add(naiveBayes);
list.Add(SVMClassifier);
//list.Add(neuronalNetworkClassifier);
Classifier classifier = new Classifier();
classifier.Classify(list, creditData);
}
public void AnimalClassifyMethod()
{
DataSetLoader dataSetLoader = new DataSetLoader();
var animals = dataSetLoader.SelectAnimals();
var data = dataSetLoader.CalculatePercent(50, animals);
DecisionTreeClassifier decisionTreeClassifier =
new DecisionTreeClassifier(data.Item1, new ShannonEntropySplitter());
NaiveBayesClassifier naiveBayes =
new NaiveBayesClassifier(data.Item1);
var list = new List <NetML.Classification>();
Kernel kernel = new LinearKernel();
SVMClassifier animalSVMClassifier =
new SVMClassifier(animals, kernel, 0.001, 10.0);
var neuronalAnimals = dataSetLoader.SelectNeuronalNetworkAnimals();
NeuronalNetworkClassifier neuronalNetworkClassifier =
new NeuronalNetworkClassifier(neuronalAnimals, 16, 7, 16, 500, 0.1);
list.Add(decisionTreeClassifier);
list.Add(naiveBayes);
list.Add(animalSVMClassifier);
list.Add(neuronalNetworkClassifier);
Classifier classifier = new Classifier();
classifier.Classify(list, data.Item2);
}
public static void Main()
{
modshogun.init_shogun_with_defaults();
double width = 2.1;
double epsilon = 1e-5;
double C = 1.0;
int mkl_norm = 2;
double[,] traindata_real = Load.load_numbers("../data/fm_train_real.dat");
double[,] testdata_real = Load.load_numbers("../data/fm_test_real.dat");
double[] trainlab = Load.load_labels("../data/label_train_multiclass.dat");
CombinedKernel kernel = new CombinedKernel();
CombinedFeatures feats_train = new CombinedFeatures();
CombinedFeatures feats_test = new CombinedFeatures();
RealFeatures subkfeats1_train = new RealFeatures(traindata_real);
RealFeatures subkfeats1_test = new RealFeatures(testdata_real);
GaussianKernel subkernel = new GaussianKernel(10, width);
feats_train.append_feature_obj(subkfeats1_train);
feats_test.append_feature_obj(subkfeats1_test);
kernel.append_kernel(subkernel);
RealFeatures subkfeats2_train = new RealFeatures(traindata_real);
RealFeatures subkfeats2_test = new RealFeatures(testdata_real);
LinearKernel subkernel2 = new LinearKernel();
feats_train.append_feature_obj(subkfeats2_train);
feats_test.append_feature_obj(subkfeats2_test);
kernel.append_kernel(subkernel2);
RealFeatures subkfeats3_train = new RealFeatures(traindata_real);
RealFeatures subkfeats3_test = new RealFeatures(testdata_real);
PolyKernel subkernel3 = new PolyKernel(10, 2);
feats_train.append_feature_obj(subkfeats3_train);
feats_test.append_feature_obj(subkfeats3_test);
kernel.append_kernel(subkernel3);
kernel.init(feats_train, feats_train);
MulticlassLabels labels = new MulticlassLabels(trainlab);
MKLMulticlass mkl = new MKLMulticlass(C, kernel, labels);
mkl.set_epsilon(epsilon);
mkl.set_mkl_epsilon(epsilon);
mkl.set_mkl_norm(mkl_norm);
mkl.train();
kernel.init(feats_train, feats_test);
double[] outMatrix = LabelsFactory.to_multiclass(mkl.apply()).get_labels();
}
public CuRBFCSRKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfCsrFormatKernel";
cudaModuleName = "KernelsCSR.cubin";
}
public CudafyRBFSlicedEllpackKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
threadsPerRow = 4;
sliceSize = 64;
}
public CuRBFEllpackKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfEllpackFormatKernel";
cudaModuleName = "KernelsEllpack.cubin";
MakeDenseVectorOnGPU = false;
}
public CuRBFEllpackKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfEllpackFormatKernel";
cudaModuleName = "KernelsEllpack.cubin";
MakeDenseVectorOnGPU = false;
}
public CuRBFCSRKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfCsrFormatKernel";
cudaModuleName = "KernelsCSR.cubin";
}
public static void Main()
{
modshogun.init_shogun_with_defaults();
double width = 2.1;
double epsilon = 1e-5;
double C = 1.0;
int mkl_norm = 2;
double[,] traindata_real = Load.load_numbers("../data/fm_train_real.dat");
double[,] testdata_real = Load.load_numbers("../data/fm_test_real.dat");
double[] trainlab = Load.load_labels("../data/label_train_multiclass.dat");
CombinedKernel kernel = new CombinedKernel();
CombinedFeatures feats_train = new CombinedFeatures();
CombinedFeatures feats_test = new CombinedFeatures();
RealFeatures subkfeats1_train = new RealFeatures(traindata_real);
RealFeatures subkfeats1_test = new RealFeatures(testdata_real);
GaussianKernel subkernel = new GaussianKernel(10, width);
feats_train.append_feature_obj(subkfeats1_train);
feats_test.append_feature_obj(subkfeats1_test);
kernel.append_kernel(subkernel);
RealFeatures subkfeats2_train = new RealFeatures(traindata_real);
RealFeatures subkfeats2_test = new RealFeatures(testdata_real);
LinearKernel subkernel2 = new LinearKernel();
feats_train.append_feature_obj(subkfeats2_train);
feats_test.append_feature_obj(subkfeats2_test);
kernel.append_kernel(subkernel2);
RealFeatures subkfeats3_train = new RealFeatures(traindata_real);
RealFeatures subkfeats3_test = new RealFeatures(testdata_real);
PolyKernel subkernel3 = new PolyKernel(10, 2);
feats_train.append_feature_obj(subkfeats3_train);
feats_test.append_feature_obj(subkfeats3_test);
kernel.append_kernel(subkernel3);
kernel.init(feats_train, feats_train);
MulticlassLabels labels = new MulticlassLabels(trainlab);
MKLMulticlass mkl = new MKLMulticlass(C, kernel, labels);
mkl.set_epsilon(epsilon);
mkl.set_mkl_epsilon(epsilon);
mkl.set_mkl_norm(mkl_norm);
mkl.train();
kernel.init(feats_train, feats_test);
double[] outMatrix = MulticlassLabels.obtain_from_generic(mkl.apply()).get_labels();
modshogun.exit_shogun();
}
private void SupportVectorMachine(List <Tuple <double[], double> > data)
{
Kernel kernel = new LinearKernel();
double n = 0.0;
double C = 0.0;
bool nAndCSet = false;
foreach (var item in netMLObject.Options)
{
if (item == "linearkernel")
{
kernel = new LinearKernel();
}
else if (item == "gaussiankernel")
{
kernel = new GaussianKernel(1.0);
}
else if (item == "polynomialkernel")
{
kernel = new PolynomialKernel(1);
}
else if (item == "logitkernel")
{
kernel = new LogitKernel();
}
else if (item == "tanhkernel")
{
kernel = new TanhKernel();
}
}
foreach (var value in netMLObject.DoubleValues)
{
if (value.Key == "n")
{
n = value.Value;
nAndCSet = true;
}
else if (value.Key == "c")
{
C = value.Value;
nAndCSet = true;
}
}
if (nAndCSet)
{
classification = new SVMClassifier(data, kernel, n, C);
}
else
{
classification = new SVMClassifier(data, kernel);
}
}
public static void Main()
{
modshogun.init_shogun_with_defaults();
double scale = 1.2;
double[,] traindata_real = Load.load_numbers("../data/fm_train_real.dat");
double[,] testdata_real = Load.load_numbers("../data/fm_test_real.dat");
RealFeatures feats_train = new RealFeatures(traindata_real);
RealFeatures feats_test = new RealFeatures(testdata_real);
LinearKernel kernel = new LinearKernel(feats_train, feats_test);
kernel.set_normalizer(new AvgDiagKernelNormalizer(scale));
kernel.init(feats_train, feats_train);
double[,] km_train = kernel.get_kernel_matrix();
kernel.init(feats_train, feats_test);
double[,] km_test = kernel.get_kernel_matrix();
// Parse and Display km_train
Console.Write("km_train:\n");
int numRows = km_train.GetLength(0);
int numCols = km_train.GetLength(1);
for (int i = 0; i < numRows; i++)
{
for (int j = 0; j < numCols; j++)
{
Console.Write(km_train[i, j] + " ");
}
Console.Write("\n");
}
// Parse and Display km_test
Console.Write("\nkm_test:\n");
numRows = km_test.GetLength(0);
numCols = km_test.GetLength(1);
for (int i = 0; i < numRows; i++)
{
for (int j = 0; j < numCols; j++)
{
Console.Write(km_test[i, j] + " ");
}
Console.Write("\n");
}
modshogun.exit_shogun();
}
public void TestSummmationReadWrite()
{
double log_length = System.Math.Log(1.234);
double log_sig_sd = System.Math.Log(2.345);
double log_nse_sd = System.Math.Log(3.456);
double[] log_var = { System.Math.Log(0.987), System.Math.Log(0.876), System.Math.Log(0.765) };
SummationKernel kf = new SummationKernel(new SquaredExponential(log_length, log_sig_sd));
kf += new LinearKernel(log_var);
kf += new WhiteNoise(log_nse_sd);
TestReadWrite(kf);
}
static void Main(string[] argv)
{
modshogun.init_shogun_with_defaults();
double width = 2.1;
double epsilon = 1e-5;
double C = 1.0;
int mkl_norm = 2;
DoubleMatrix traindata_real = Load.load_numbers("../data/fm_train_real.dat");
DoubleMatrix testdata_real = Load.load_numbers("../data/fm_test_real.dat");
DoubleMatrix trainlab = Load.load_labels("../data/label_train_twoclass.dat");
CombinedKernel kernel = new CombinedKernel();
CombinedFeatures feats_train = new CombinedFeatures();
CombinedFeatures feats_test = new CombinedFeatures();
RealFeatures subkfeats_train = new RealFeatures(traindata_real);
RealFeatures subkfeats_test = new RealFeatures(testdata_real);
GaussianKernel subkernel = new GaussianKernel(10, width);
feats_train.append_feature_obj(subkfeats_train);
feats_test.append_feature_obj(subkfeats_test);
kernel.append_kernel(subkernel);
LinearKernel subkernel2 = new LinearKernel();
feats_train.append_feature_obj(subkfeats_train);
feats_test.append_feature_obj(subkfeats_test);
kernel.append_kernel(subkernel2);
PolyKernel subkernel3 = new PolyKernel(10, 2);
feats_train.append_feature_obj(subkfeats_train);
feats_test.append_feature_obj(subkfeats_test);
kernel.append_kernel(subkernel3);
kernel.init(feats_train, feats_train);
Labels labels = new Labels(trainlab);
MKLMultiClass mkl = new MKLMultiClass(C, kernel, labels);
mkl.set_epsilon(epsilon);
mkl.set_mkl_epsilon(epsilon);
mkl.set_mkl_norm(mkl_norm);
mkl.train();
kernel.init(feats_train, feats_test);
DoubleMatrix @out = mkl.apply().get_labels();
modshogun.exit_shogun();
}
/// <summary>
/// Initialises a SVMGenerator object
/// </summary>
public SVMGenerator()
{
Bias = 0d;
C = 1d;
Epsilon = 0.001;
MaxIterations = 10;
KernelFunction = new LinearKernel();
NormalizeFeatures = true;
if (SelectionFunction == null)
{
SelectionFunction = new WorkingSetSelection3();
}
}
public CuRBFERTILPKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfERTILP";
//cudaProductKernelName = "rbfEllRTILP_old";
cudaModuleName = "KernelsEllpack.cubin";
MakeDenseVectorOnGPU = false;
ThreadsPerRow = 4;
Prefetch = 2;
}
public static void Main()
{
modshogun.init_shogun_with_defaults();
double scale = 1.2;
double[,] traindata_real = Load.load_numbers("../data/fm_train_word.dat");
double[,] testdata_real = Load.load_numbers("../data/fm_test_word.dat");
short[,] traindata_word = new short[traindata_real.GetLength(0), traindata_real.GetLength(1)];
for (int i = 0; i < traindata_real.GetLength(0); i++)
{
for (int j = 0; j < traindata_real.GetLength(1); j++)
{
traindata_word[i, j] = (short)traindata_real[i, j];
}
}
short[,] testdata_word = new short[testdata_real.GetLength(0), testdata_real.GetLength(1)];
for (int i = 0; i < testdata_real.GetLength(0); i++)
{
for (int j = 0; j < testdata_real.GetLength(1); j++)
{
testdata_word[i, j] = (short)testdata_real[i, j];
}
}
WordFeatures feats_train = new WordFeatures(traindata_word);
WordFeatures feats_test = new WordFeatures(testdata_word);
LinearKernel kernel = new LinearKernel(feats_train, feats_test);
kernel.set_normalizer(new AvgDiagKernelNormalizer(scale));
kernel.init(feats_train, feats_train);
double[,] km_train = kernel.get_kernel_matrix();
kernel.init(feats_train, feats_test);
double[,] km_test = kernel.get_kernel_matrix();
foreach (double item in km_train)
{
Console.Write(item);
}
foreach (double item in km_test)
{
Console.Write(item);
}
modshogun.exit_shogun();
}
public CuRBFERTILPKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfERTILP";
//cudaProductKernelName = "rbfEllRTILP_old";
cudaModuleName = "KernelsEllpack.cubin";
MakeDenseVectorOnGPU = false;
ThreadsPerRow = 4;
Prefetch =2;
}
public CuRBFSlEllKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfSlicedEllpackKernel";
cudaModuleName = "KernelsSlicedEllpack.cubin";
threadsPerRow = 4;
sliceSize = 64;
//threadsPerRow = 2;
//sliceSize = 4;
}
public CuRBFSlEllKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfSlicedEllpackKernel";
cudaModuleName = "KernelsSlicedEllpack.cubin";
threadsPerRow = 4;
sliceSize = 64;
//threadsPerRow = 2;
//sliceSize = 4;
}
public CuRBFEllILPKernelCol2(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfEllpackILPcol2";
//cudaProductKernelName = "rbfEllpackILPcol2_Prefetch2";
cudaModuleName = "KernelsEllpackCol2.cubin";
MakeDenseVectorOnGPU = false;
preFetch = 2;
//threadsPerBlock = 128;
}
public CuRBFEllILPKernelCol2(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfEllpackILPcol2";
//cudaProductKernelName = "rbfEllpackILPcol2_Prefetch2";
cudaModuleName = "KernelsEllpackCol2.cubin";
MakeDenseVectorOnGPU = false;
preFetch = 2;
//threadsPerBlock = 128;
}
public static void Main() {
modshogun.init_shogun_with_defaults();
double scale = 1.2;
double[,] traindata_real = Load.load_numbers("../data/fm_train_real.dat");
double[,] testdata_real = Load.load_numbers("../data/fm_test_real.dat");
RealFeatures feats_train = new RealFeatures(traindata_real);
RealFeatures feats_test = new RealFeatures(testdata_real);
LinearKernel kernel = new LinearKernel(feats_train, feats_test);
kernel.set_normalizer(new AvgDiagKernelNormalizer(scale));
kernel.init(feats_train, feats_train);
double[,] km_train = kernel.get_kernel_matrix();
kernel.init(feats_train, feats_test);
double[,] km_test = kernel.get_kernel_matrix();
// Parse and Display km_train
Console.Write("km_train:\n");
int numRows = km_train.GetLength(0);
int numCols = km_train.GetLength(1);
for(int i = 0; i < numRows; i++){
for(int j = 0; j < numCols; j++){
Console.Write(km_train[i,j] +" ");
}
Console.Write("\n");
}
// Parse and Display km_test
Console.Write("\nkm_test:\n");
numRows = km_test.GetLength(0);
numCols = km_test.GetLength(1);
for(int i = 0; i < numRows; i++){
for(int j = 0; j < numCols; j++){
Console.Write(km_test[i,j] +" ");
}
Console.Write("\n");
}
modshogun.exit_shogun();
}
public void TestSumKDerivs()
{
double[] log_length = { System.Math.Log(0.543), System.Math.Log(0.432), System.Math.Log(0.321) };
double log_sig_sd = System.Math.Log(2.345);
double[] log_var = { System.Math.Log(0.987), System.Math.Log(0.876), System.Math.Log(0.765) };
double log_nse_sd = System.Math.Log(3.456);
SummationKernel kf = new SummationKernel(new ARD(log_length, log_sig_sd));
kf += new LinearKernel(log_var);
kf += new WhiteNoise(log_nse_sd);
double[] x1 = { 0.1, 0.2, 0.3 };
double[] x2 = { 0.9, 0.7, 0.5 };
Vector x1Vec = Vector.FromArray(x1);
Vector x2Vec = Vector.FromArray(x2);
TestDerivatives(kf, x1Vec, x1Vec);
TestDerivatives(kf, x1Vec, x2Vec);
}
public CuRBFSERTILPKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfSERTILP";
//cudaProductKernelName = "rbfSlicedEllpackKernel_shared";
cudaModuleName = "KernelsSlicedEllpack.cubin";
threadsPerRow = 4;
sliceSize = 64;
preFechSize = 2;
//threadsPerRow = 2;
//sliceSize = 4;
}
public CuRBFSERTILPKernel(float gamma)
{
linKernel = new LinearKernel();
Gamma = gamma;
cudaProductKernelName = "rbfSERTILP";
//cudaProductKernelName = "rbfSlicedEllpackKernel_shared";
cudaModuleName = "KernelsSlicedEllpack.cubin";
threadsPerRow = 4;
sliceSize = 64;
preFechSize = 2;
//threadsPerRow = 2;
//sliceSize = 4;
}
public void TestLinearKernelDerivs()
{
double[] logVariances = { -0.123, 0.456, 1.789 };
LinearKernel lk = new LinearKernel(logVariances);
double[] x1 = { 0.1, 0.2, 0.3 };
double[] x2 = { 0.9, 0.7, 0.5 };
Vector v1 = Vector.FromArray(x1);
Vector v2 = Vector.FromArray(x2);
List <Vector> xlist = new List <Vector>(2);
xlist.Add(v1);
xlist.Add(v2);
lk.InitialiseFromData(xlist);
Vector x1Vec = Vector.FromArray(x1);
Vector x2Vec = Vector.FromArray(x2);
TestDerivatives(lk, x1Vec, x1Vec);
TestDerivatives(lk, x1Vec, x2Vec);
}
public static void Main()
{
modshogun.init_shogun_with_defaults();
double scale = 1.2;
double[,] traindata_real = Load.load_numbers("../data/fm_train_word.dat");
double[,] testdata_real = Load.load_numbers("../data/fm_test_word.dat");
short[,] traindata_word = new short[traindata_real.GetLength(0), traindata_real.GetLength(1)];
for (int i = 0; i < traindata_real.GetLength(0); i++){
for (int j = 0; j < traindata_real.GetLength(1); j++)
traindata_word[i, j] = (short)traindata_real[i, j];
}
short[,] testdata_word = new short[testdata_real.GetLength(0), testdata_real.GetLength(1)];
for (int i = 0; i < testdata_real.GetLength(0); i++){
for (int j = 0; j < testdata_real.GetLength(1); j++)
testdata_word[i, j] = (short)testdata_real[i, j];
}
WordFeatures feats_train = new WordFeatures(traindata_word);
WordFeatures feats_test = new WordFeatures(testdata_word);
LinearKernel kernel = new LinearKernel(feats_train, feats_test);
kernel.set_normalizer(new AvgDiagKernelNormalizer(scale));
kernel.init(feats_train, feats_train);
double[,] km_train = kernel.get_kernel_matrix();
kernel.init(feats_train, feats_test);
double[,] km_test = kernel.get_kernel_matrix();
foreach(double item in km_train) {
Console.Write(item);
}
foreach(double item in km_test) {
Console.Write(item);
}
modshogun.exit_shogun();
}
static void Main(string[] argv)
{
modshogun.init_shogun_with_defaults();
double scale = 1.2;
DoubleMatrix traindata_real = Load.load_numbers("../data/fm_train_word.dat");
DoubleMatrix testdata_real = Load.load_numbers("../data/fm_test_word.dat");
WordFeatures feats_train = new WordFeatures(traindata_real);
WordFeatures feats_test = new WordFeatures(testdata_real);
LinearKernel kernel = new LinearKernel(feats_train, feats_test);
kernel.set_normalizer(new AvgDiagKernelNormalizer(scale));
kernel.init(feats_train, feats_train);
DoubleMatrix km_train = kernel.get_kernel_matrix();
kernel.init(feats_train, feats_test);
DoubleMatrix km_test = kernel.get_kernel_matrix();
Console.WriteLine(km_train.ToString());
Console.WriteLine(km_test.ToString());
modshogun.exit_shogun();
}
//const string cudaKernelName = "linearCsrFormatKernel";
public CuLinearKernel()
{
linKernel = new LinearKernel();
cudaProductKernelName = "linearCsrFormatKernel";
}
public RBFDualEvaluator(float gamma)
{
linKernel = new LinearKernel();
this.gamma = gamma;
}
internal static HandleRef getCPtr(LinearKernel obj) {
return (obj == null) ? new HandleRef(null, IntPtr.Zero) : obj.swigCPtr;
}
internal static HandleRef getCPtr(LinearKernel obj)
{
return((obj == null) ? new HandleRef(null, IntPtr.Zero) : obj.swigCPtr);
}
private static IEnumerable <Assignment> AngularCluster(IList <Matrix <double> > features, ulong numberClusters)
{
if (features == null)
{
throw new ArgumentNullException(nameof(features));
}
var size = features.Count;
if (size == 0)
{
throw new ArgumentException("The dataset can't be empty", nameof(features));
}
var featureSize = features[0].Size;
for (var index = 0; index < size; ++index)
{
if (features[index].Size != featureSize)
{
throw new ArgumentException("All feature vectors must have the same length.", nameof(features));
}
}
// find the centroid of feats
Matrix <double> tmp;
var m = Matrix <double> .CreateTemplateParameterizeMatrix(0, 1);
for (var index = 0; index < size; ++index)
{
tmp = m + features[index];
m.Dispose();
m = tmp;
}
tmp = m;
m = tmp / size;
tmp.Dispose();
// Now center feats and then project onto the unit sphere. The reason for projecting
// onto the unit sphere is so pick_initial_centers() works in a sensible way.
for (var index = 0; index < size; ++index)
{
tmp = features[index] - m;
features[index].Dispose();
features[index] = tmp;
var length = Dlib.Length(features[index]);
if (Math.Abs(length) > double.Epsilon)
{
tmp = features[index] / length;
features[index].Dispose();
features[index] = tmp;
}
}
// now do angular clustering of the points
var linearKernel = new LinearKernel <Matrix <double> >(0, 1);
var tempCenters = Dlib.PickInitialCenters((int)numberClusters, features, linearKernel, 0.05).ToArray();
var centers = Dlib.FindClustersUsingAngularKMeans(features, tempCenters).ToArray();
foreach (var center in tempCenters)
{
center.Dispose();
}
linearKernel.Dispose();
// and then report the resulting assignments
var assignments = new List <Assignment>(size);
for (var index = 0; index < size; ++index)
{
var temp = new Assignment();
temp.C = Dlib.NearestCenter(centers, features[index]);
using (var temp2 = features[index] - centers[temp.C])
temp.Distance = Dlib.Length(temp2);
temp.Index = (ulong)index;
assignments.Add(temp);
}
return(assignments);
}
//const string cudaKernelName = "linearCsrFormatKernel";
public CuLinearKernel()
{
linKernel = new LinearKernel();
cudaProductKernelName = "linearCsrFormatKernel";
}
