more refactoring of API examples and required c++ adjustments (#4275)

examples/meta/src/base_api/put_get_add.sg

@@ -26,11 +26,11 @@ matrix[1,1] = 0.4
 #RealMatrix matrix2 = feats.get_real_matrix("feature_matrix")
 #feats.put("feature_matrix", matrix2)
 
-EuclideanDistance distance()
-knn.put("distance", distance)
+Distance dist = distance("EuclideanDistance")
+knn.put("distance", dist)
 
-SGObject distance2 = knn.get("distance")
-knn.put("distance", distance2)
+SGObject dist2 = knn.get("distance")
+knn.put("distance", dist2)
 
 LibSVM svm()
 svm.put("kernel", k)

examples/meta/src/binary/averaged_perceptron.sg

@@ -6,8 +6,8 @@ CSVFile f_labels_test("../../data/classifier_binary_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-BinaryLabels labels_train(f_labels_train)
-BinaryLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![set_parameters]

examples/meta/src/binary/linear_support_vector_machine.sg

@@ -6,8 +6,8 @@ CSVFile f_labels_test("../../data/classifier_binary_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-BinaryLabels labels_train(f_labels_train)
-BinaryLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![set_parameters]

examples/meta/src/binary/perceptron.sg

@@ -6,8 +6,8 @@ CSVFile f_labels_test("../../data/classifier_binary_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-BinaryLabels labels_train(f_labels_train)
-BinaryLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![create_instance]

examples/meta/src/multiclass/ecoc_random.sg

@@ -8,12 +8,12 @@ CSVFile f_labels_test("../../data/classifier_4class_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-MulticlassLabels labels_train(f_labels_train)
-MulticlassLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![create_classifier]
-LibLinear classifier()
+Machine classifier = machine("LibLinear")
 #![create_classifier]
 
 #![choose_strategy]

examples/meta/src/multiclass/linear.sg

@@ -6,12 +6,12 @@ CSVFile f_labels_test("../../data/classifier_4class_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-MulticlassLabels labels_train(f_labels_train)
-MulticlassLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![create_classifier]
-LibLinear classifier()
+Machine classifier = machine("LibLinear")
 #![create_classifier]
 
 #![choose_strategy]

examples/meta/src/multiclass/linear_discriminant_analysis.sg

@@ -6,8 +6,8 @@ CSVFile f_labels_test("../../data/classifier_4class_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-MulticlassLabels labels_train(f_labels_train)
-MulticlassLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![create_instance]

examples/meta/src/multiclass/logistic_regression.sg

@@ -6,8 +6,8 @@ CSVFile f_labels_test("../../data/classifier_4class_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-MulticlassLabels labels_train(f_labels_train)
-MulticlassLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 

examples/meta/src/multiclass/quadratic_discriminant_analysis.sg

@@ -6,8 +6,8 @@ CSVFile f_labels_test("../../data/classifier_4class_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-MulticlassLabels labels_train(f_labels_train)
-MulticlassLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![create_instance]

examples/meta/src/multiclass/shareboost.sg

@@ -6,8 +6,8 @@ CSVFile f_labels_test("../../data/classifier_4class_2d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-MulticlassLabels labels_train(f_labels_train)
-MulticlassLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![create_instance]

examples/meta/src/regression/least_angle_regression.sg

@@ -6,8 +6,8 @@ CSVFile f_labels_test("../../data/regression_1d_linear_labels_test.dat")
 #![create_features]
 Features features_train = features(f_feats_train)
 Features features_test = features(f_feats_test)
-RegressionLabels labels_train(f_labels_train)
-RegressionLabels labels_test(f_labels_test)
+Labels labels_train = labels(f_labels_train)
+Labels labels_test = labels(f_labels_test)
 #![create_features]
 
 #![preprocess_features]
@@ -28,7 +28,7 @@ Machine lars = machine("LeastAngleRegression", labels=labels_train, lasso=False,
 
 #![train_and_apply]
 lars.train(features_train)
-RegressionLabels labels_predict = lars.apply_regression(features_test)
+Labels labels_predict = lars.apply(features_test)
 
 #[!extract_w]
 RealVector weights = lars.get_real_vector("w")
@@ -40,5 +40,5 @@ real mse = eval.evaluate(labels_predict, labels_test)
 #![evaluate_error]
 
 # integration testing variables
-RealVector output = labels_test.get_labels()
+RealVector output = labels_predict.get_real_vector("labels")
 

examples/undocumented/libshogun/classifier_lda.cpp

@@ -44,12 +44,11 @@ void test()
 	lda->train();
 
 	// Classify and display output
-	CMulticlassLabels* output=CLabelsFactory::to_multiclass(lda->apply());
+	auto output = multiclass_labels(lda->apply());
 	SG_REF(output);
 	SGVector<float64_t>::display_vector(output->get_labels().vector, output->get_num_labels());
 
 	// Free memory
-	SG_UNREF(output);
 	SG_UNREF(lda);
 #endif
 }

src/shogun/classifier/AveragedPerceptron.cpp

@@ -43,7 +43,6 @@ void CAveragedPerceptron::init()
 bool CAveragedPerceptron::train_machine(CFeatures* data)
 {
 	ASSERT(m_labels)
-	ASSERT(m_labels->get_label_type() == LT_BINARY)
 
 	if (data)
 	{
@@ -54,7 +53,7 @@ bool CAveragedPerceptron::train_machine(CFeatures* data)
 	ASSERT(features)
 	bool converged=false;
 	int32_t iter=0;
-	SGVector<int32_t> train_labels=((CBinaryLabels*) m_labels)->get_int_labels();
+	SGVector<int32_t> train_labels = binary_labels(m_labels)->get_int_labels();
 	int32_t num_feat=features->get_dim_feature_space();
 	int32_t num_vec=features->get_num_vectors();
 

src/shogun/classifier/Perceptron.cpp

@@ -47,7 +47,6 @@ CPerceptron::~CPerceptron()
 bool CPerceptron::train_machine(CFeatures* data)
 {
 	ASSERT(m_labels)
-	ASSERT(m_labels->get_label_type() == LT_BINARY)
 
 	if (data)
 	{
@@ -59,7 +58,7 @@ bool CPerceptron::train_machine(CFeatures* data)
 	ASSERT(features)
 	bool converged=false;
 	int32_t iter=0;
-	SGVector<int32_t> train_labels=((CBinaryLabels*) m_labels)->get_int_labels();
+	SGVector<int32_t> train_labels = binary_labels(m_labels)->get_int_labels();
 	int32_t num_feat=features->get_dim_feature_space();
 	int32_t num_vec=features->get_num_vectors();
 

src/shogun/classifier/svm/LibLinear.cpp

@@ -74,7 +74,6 @@ bool CLibLinear::train_machine(CFeatures* data)
 {
 
 	ASSERT(m_labels)
-	ASSERT(m_labels->get_label_type() == LT_BINARY)
 	init_linear_term();
 
 	if (data)
@@ -142,9 +141,10 @@ bool CLibLinear::train_machine(CFeatures* data)
 	double Cp = get_C1();
 	double Cn = get_C2();
 
+	auto labels = binary_labels(m_labels);
 	for (int32_t i = 0; i < prob.l; i++)
 	{
-		prob.y[i] = ((CBinaryLabels*)m_labels)->get_int_label(i);
+		prob.y[i] = labels->get_int_label(i);
 		if (prob.y[i] == +1)
 			Cs[i] = get_C1();
 		else if (prob.y[i] == -1)

src/shogun/evaluation/MulticlassAccuracy.cpp

@@ -15,15 +15,16 @@ float64_t CMulticlassAccuracy::evaluate(CLabels* predicted, CLabels* ground_trut
 {
 	ASSERT(predicted && ground_truth)
 	ASSERT(predicted->get_num_labels() == ground_truth->get_num_labels())
-	ASSERT(predicted->get_label_type() == LT_MULTICLASS)
-	ASSERT(ground_truth->get_label_type() == LT_MULTICLASS)
 	int32_t length = predicted->get_num_labels();
+	auto predicted_mc = multiclass_labels(predicted);
+	auto ground_truth_mc = multiclass_labels(ground_truth);
 	int32_t correct = 0;
 	if (m_ignore_rejects)
 	{
 		for (int32_t i=0; i<length; i++)
 		{
-			if (((CMulticlassLabels*) predicted)->get_int_label(i)==((CMulticlassLabels*) ground_truth)->get_int_label(i))
+			if (predicted_mc->get_int_label(i) ==
+			    ground_truth_mc->get_int_label(i))
 				correct++;
 		}
 		return ((float64_t)correct)/length;
@@ -33,11 +34,11 @@ float64_t CMulticlassAccuracy::evaluate(CLabels* predicted, CLabels* ground_trut
 		int32_t total = length;
 		for (int32_t i=0; i<length; i++)
 		{
-			int32_t predicted_label = ((CMulticlassLabels*) predicted)->get_int_label(i);
+			int32_t predicted_label = predicted_mc->get_int_label(i);
 
-			if (predicted_label==((CMulticlassLabels*) predicted)->REJECTION_LABEL)
+			if (predicted_label == predicted_mc->REJECTION_LABEL)
 				total--;
-			else if (predicted_label==((CMulticlassLabels*) ground_truth)->get_int_label(i))
+			else if (predicted_label == ground_truth_mc->get_int_label(i))
 				correct++;
 		}
 		m_rejects_num = length-total;

src/shogun/machine/MulticlassMachine.cpp

@@ -247,7 +247,8 @@ bool CMulticlassMachine::train_machine(CFeatures* data)
 	SG_REF(train_labels);
 	m_machine->set_labels(train_labels);
 
-	m_multiclass_strategy->train_start(CLabelsFactory::to_multiclass(m_labels), train_labels);
+	m_multiclass_strategy->train_start(
+	    multiclass_labels(m_labels), train_labels);
 	while (m_multiclass_strategy->train_has_more())
 	{
 		SGVector<index_t> subset=m_multiclass_strategy->train_prepare_next();

src/shogun/multiclass/ShareBoost.cpp

@@ -139,7 +139,8 @@ void CShareBoost::compute_pred(const float64_t *W)
 
 void CShareBoost::compute_rho()
 {
-	CMulticlassLabels *lab = dynamic_cast<CMulticlassLabels *>(m_labels);
+	auto lab = multiclass_labels(m_labels);
+
 	for (int32_t i=0; i < m_rho.num_rows; ++i)
 	{ // i loop classes
 		for (int32_t j=0; j < m_rho.num_cols; ++j)
@@ -163,6 +164,7 @@ void CShareBoost::compute_rho()
 int32_t CShareBoost::choose_feature()
 {
 	SGVector<float64_t> l1norm(m_fea.num_rows);
+	auto lab = multiclass_labels(m_labels);
 	for (int32_t j=0; j < m_fea.num_rows; ++j)
 	{
 		if (std::find(&m_activeset[0], &m_activeset[m_activeset.vlen], j) !=
@@ -173,7 +175,6 @@ int32_t CShareBoost::choose_feature()
 		else
 		{
 			l1norm[j] = 0;
-			CMulticlassLabels *lab = dynamic_cast<CMulticlassLabels *>(m_labels);
 			for (int32_t k=0; k < m_multiclass_strategy->get_num_classes(); ++k)
 			{
 				float64_t abssum = 0;

src/shogun/multiclass/ShareBoostOptimizer.cpp

@@ -51,7 +51,7 @@ float64_t ShareBoostOptimizer::lbfgs_evaluate(void *userdata, const float64_t *W
 	int32_t k = optimizer->m_sb->m_multiclass_strategy->get_num_classes();
 
 	SGMatrix<float64_t> fea = optimizer->m_sb->m_fea;
-	CMulticlassLabels *lab = dynamic_cast<CMulticlassLabels *>(optimizer->m_sb->m_labels);
+	auto lab = multiclass_labels(optimizer->m_sb->m_labels);
 
 	// compute gradient
 	for (int32_t i=0; i < m; ++i)

src/shogun/multiclass/tree/C45ClassifierTree.cpp

@@ -530,14 +530,16 @@ void CC45ClassifierTree::prune_tree_from_current_node(CDenseFeatures<float64_t>*
 		SG_UNREF(right_child);
 	}
 
-	SG_UNREF(children);
+	SG_UNREF(children)
 
 	CMulticlassLabels* predicted_unpruned=apply_multiclass_from_current_node(feats, current);
+	SG_REF(predicted_unpruned);
 	SGVector<float64_t> pruned_labels=SGVector<float64_t>(feature_matrix.num_cols);
 	for (int32_t i=0; i<feature_matrix.num_cols; i++)
 		pruned_labels[i]=current->data.class_label;
 
 	CMulticlassLabels* predicted_pruned=new CMulticlassLabels(pruned_labels);
+	SG_REF(predicted_pruned);
 
 	CMulticlassAccuracy* accuracy=new CMulticlassAccuracy();
 	float64_t unpruned_accuracy=accuracy->evaluate(predicted_unpruned, gnd_truth);

src/shogun/multiclass/tree/ID3ClassifierTree.cpp

@@ -335,11 +335,13 @@ void CID3ClassifierTree::prune_tree_machine(CDenseFeatures<float64_t>* feats,
 	SG_UNREF(children);
 
 	CMulticlassLabels* predicted_unpruned = apply_multiclass_from_current_node(feats, current);
+	SG_REF(predicted_unpruned);
 	SGVector<float64_t> pruned_labels = SGVector<float64_t>(feature_matrix.num_cols);
 	for (int32_t i=0; i<feature_matrix.num_cols; i++)
 		pruned_labels[i] = current->data.class_label;
 
 	CMulticlassLabels* predicted_pruned = new CMulticlassLabels(pruned_labels);
+	SG_REF(predicted_pruned);
 
 	CMulticlassAccuracy* accuracy = new CMulticlassAccuracy();
 	float64_t unpruned_accuracy = accuracy->evaluate(predicted_unpruned, gnd_truth);

src/shogun/regression/LeastAngleRegression.cpp

@@ -104,9 +104,6 @@ void CLeastAngleRegression::plane_rot(ST x0, ST x1,
 
 bool CLeastAngleRegression::train_machine(CFeatures* data)
 {
-	REQUIRE(m_labels->get_label_type() == LT_REGRESSION, "Provided labels (%s) are of type (%d) - they should be regression labels (%d) instead.\n"
-		, m_labels->get_name(), m_labels->get_label_type(), LT_REGRESSION, m_labels->get_label_type())
-
 	if (!data)
 	{
 		REQUIRE(features, "No features provided.\n")
@@ -155,7 +152,7 @@ bool CLeastAngleRegression::train_machine_templated(CDenseFeatures<ST> * data)
 	m_is_active.resize(n_fea);
 	fill(m_is_active.begin(), m_is_active.end(), false);
 
-	SGVector<ST> y = ((CRegressionLabels*) m_labels)->template get_labels_t<ST>();
+	SGVector<ST> y = regression_labels(m_labels)->template get_labels_t<ST>();
 	typename SGVector<ST>::EigenVectorXtMap map_y(y.vector, y.size());
 
 	// transpose(X) is more convenient to work with since we care

tests/unit/multiclass/MulticlassOCAS_unittest.cc

@@ -28,6 +28,7 @@ TEST(MulticlassOCASTest,train)
   mocas->train();
 
   CMulticlassLabels* pred = (CMulticlassLabels*)mocas->apply(test_feats);
+  SG_REF(pred);
   CMulticlassAccuracy evaluate = CMulticlassAccuracy();
   float64_t result = evaluate.evaluate(pred, ground_truth);
   EXPECT_GT(result, 0.99);

tests/unit/multiclass/tree/C45ClassifierTree_unittest.cc

@@ -496,6 +496,8 @@ TEST(C45ClassifierTree, tree_prune_categorical_attributes)
 	CDenseFeatures<float64_t>* train_features=new CDenseFeatures<float64_t>(data);
 	CMulticlassLabels* train_lab=new CMulticlassLabels(train_labels);
 	CMulticlassLabels* validation_lab=new CMulticlassLabels(validation_labels);
+	SG_REF(train_lab);
+	SG_REF(validation_lab);
 
 	CC45ClassifierTree* c45tree=new CC45ClassifierTree();
 	c45tree->set_labels(train_lab);
@@ -581,8 +583,10 @@ TEST(C45ClassifierTree, tree_prune_continuous_attributes)
 
 	CDenseFeatures<float64_t>* train_features=new CDenseFeatures<float64_t>(data);
 	CMulticlassLabels* train_lab=new CMulticlassLabels(train_labels);
+	SG_REF(train_lab);
 	CDenseFeatures<float64_t>* validation_features=new CDenseFeatures<float64_t>(validation_data);
 	CMulticlassLabels* validation_lab=new CMulticlassLabels(validation_labels);
+	SG_REF(validation_lab);
 
 	CC45ClassifierTree* c45tree=new CC45ClassifierTree();
 	c45tree->set_labels(train_lab);