Merge branch 'master' of github.com:shogun-toolbox/shogun

shogun-toolbox · Jun 20, 2011 · 880d418 · 880d418
2 parents a909a9f + ab158e1
commit 880d418
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 8 deletions.
diff --git a/examples/undocumented/python_modular/graphical/krr_sinc.py b/examples/undocumented/python_modular/graphical/krr_sinc.py
@@ -20,10 +20,10 @@
 krr.train()
 
 plot(X, Y, '.', label='train data')
-plot(X[0], krr.classify().get_labels(), hold=True, label='train output')
+plot(X[0], krr.apply().get_labels(), hold=True, label='train output')
 
 XE, YE=util.compute_output_plot_isolines_sine(krr, gk, feat)
-YE200=krr.classify_example(200)
+YE200=krr.apply(200)
 
 plot(XE[0], YE, hold=True, label='test output')
 plot([XE[0,200]], [YE200], '+', hold=True)

diff --git a/examples/undocumented/python_modular/graphical/svr_sinc.py b/examples/undocumented/python_modular/graphical/svr_sinc.py
@@ -19,7 +19,7 @@
 svr.train()
 
 plot(X, Y, '.', label='train data')
-plot(X[0], svr.classify().get_labels(), hold=True, label='train output')
+plot(X[0], svr.apply().get_labels(), hold=True, label='train output')
 
 XE, YE=util.compute_output_plot_isolines_sine(svr, gk, feat)
 plot(XE[0], YE, hold=True, label='test output')

diff --git a/examples/undocumented/python_modular/graphical/util.py b/examples/undocumented/python_modular/graphical/util.py
@@ -79,7 +79,7 @@ def compute_output_plot_isolines(classifier, kernel=None, train=None, sparse=Fal
 	else:
 		classifier.set_features(test)
 
-	labels=classifier.classify().get_labels()
+	labels=classifier.apply().get_labels()
 	z=labels.reshape((size, size))
 
 	return x, y, z
@@ -98,6 +98,6 @@ def compute_output_plot_isolines_sine(classifier, kernel, train):
 	x.sort()
 	test=RealFeatures(x)
 	kernel.init(train, test)
-	y=classifier.classify().get_labels()
+	y=classifier.apply().get_labels()
 
 	return x, y
diff --git a/src/libshogun/preprocessor/ClassicMDS.cpp b/src/libshogun/preprocessor/ClassicMDS.cpp
@@ -91,15 +91,19 @@ SGMatrix<float64_t> CClassicMDS::embed_by_distance(CDistance* distance)
 
 	// replace feature matrix with (top) eigenvectors associated with largest
 	// positive eigenvalues (ignores negative eigenvalues)
-	float64_t* replace_feature_matrix = new float64_t[N*m_target_dim];
 	for (i=0; i<m_target_dim; i++)
 	{
 		if (eigenvalues_vector[N-i-1]<0)
 		{
-			m_target_dim = i;
-			break;
+					SG_WARNING("Can't embed into %dd space, embedded into %dd",m_target_dim,i);
+					m_target_dim = i;
+					break;
 		}
+	}
 
+	float64_t* replace_feature_matrix = new float64_t[N*m_target_dim];
+	for (i=0; i<m_target_dim; i++)
+	{
 		for (j=0; j<N; j++)
 			replace_feature_matrix[j*m_target_dim+i] = Ds_matrix[(N-i-1)*N+j]*CMath::sqrt(eigenvalues_vector[N-i-1]);
 	}