Added bias usage capability for crammer-singer liblinear

shogun-toolbox · Mar 8, 2012 · db96f7c · db96f7c
1 parent cdd0218
commit db96f7c
Show file tree

Hide file tree

Showing 3 changed files with 47 additions and 6 deletions.
diff --git a/src/shogun/classifier/svm/MulticlassLibLinear.cpp b/src/shogun/classifier/svm/MulticlassLibLinear.cpp
@@ -10,6 +10,7 @@
 
 
 #include <shogun/classifier/svm/MulticlassLibLinear.h>
+#include <shogun/classifier/svm/SVM_linear.h>
 #include <shogun/mathematics/Math.h>
 
 using namespace shogun;
@@ -24,13 +25,13 @@ bool CMulticlassLibLinear::train_machine(CFeatures* data)
 
 	problem mc_problem;
 	mc_problem.l = num_vectors;
-	mc_problem.n = m_features->get_dim_feature_space();
+	mc_problem.n = m_features->get_dim_feature_space()+1;
 	mc_problem.y = SG_MALLOC(int32_t, mc_problem.l);
 	for (int32_t i=0; i<num_vectors; i++)
 		mc_problem.y[i] = labels->get_int_label(i);
 
 	mc_problem.x = m_features;
-	mc_problem.use_bias = true;
+	mc_problem.use_bias = m_use_bias;
 
 	float64_t* w = SG_MALLOC(float64_t, mc_problem.n*num_classes);
 	float64_t* C = SG_MALLOC(float64_t, num_vectors);
@@ -45,10 +46,12 @@ bool CMulticlassLibLinear::train_machine(CFeatures* data)
 	for (int32_t i=0; i<num_classes; i++)
 	{
 		CLinearMachine* machine = new CLinearMachine();
-		SGVector<float64_t> cw(mc_problem.n);
-		for (int32_t j=0; j<mc_problem.n; j++)
+		float64_t* cw = SG_MALLOC(float64_t, mc_problem.n);
+		for (int32_t j=0; j<mc_problem.n-1; j++)
 			cw[j] = w[j*num_classes+i];
-		machine->set_w(cw);
+		machine->set_w(SGVector<float64_t>(cw,mc_problem.n-1));
+		CMath::display_vector(cw,mc_problem.n);
+		machine->set_bias(w[(mc_problem.n-1)*num_classes+i]);
 
 		m_machines[i] = machine;
 	}

diff --git a/src/shogun/classifier/svm/MulticlassLibLinear.h b/src/shogun/classifier/svm/MulticlassLibLinear.h
@@ -13,7 +13,6 @@
 
 #include <shogun/lib/common.h>
 #include <shogun/features/DotFeatures.h>
-#include <shogun/classifier/svm/SVM_linear.h>
 #include <shogun/machine/multiclass/LinearMulticlassMachine.h>
 
 namespace shogun
@@ -38,6 +37,7 @@ class CMulticlassLibLinear : public CLinearMulticlassMachine
 		{
 			set_epsilon(1e-2);
 			set_max_iter(10000);
+			set_use_bias(false);
 		}
 
 		/** destructor */
@@ -77,6 +77,21 @@ class CMulticlassLibLinear : public CLinearMulticlassMachine
 		 */
 		inline float64_t get_epsilon() const { return m_epsilon; }
 
+		/** set use bias
+		 * @param use_bias use_bias value
+		 */
+		inline void set_use_bias(bool use_bias)
+		{
+			m_use_bias = use_bias;
+		}
+		/** get use bias 
+		 * @return use_bias value
+		 */
+		inline bool get_use_bias() const
+		{
+			return m_use_bias;
+		}
+
 		/** set max iter
 		 * @param max_iter max iter value
 		 */
@@ -106,6 +121,9 @@ class CMulticlassLibLinear : public CLinearMulticlassMachine
 
 		/** max number of iterations */
 		int32_t m_max_iter;
+
+		/** use bias */
+		bool m_use_bias;
 };
 }
 #endif
diff --git a/src/shogun/classifier/svm/SVM_linear.cpp b/src/shogun/classifier/svm/SVM_linear.cpp
@@ -429,6 +429,8 @@ void Solver_MCSVM_CS::Solve(double *w)
 			alpha_index[i*nr_class+m] = m;
 
 		QD[i] = prob->x->dot(i, prob->x,i);
+		if (prob->use_bias)
+			QD[i] += 1.0;
 
 		active_size_i[i] = nr_class;
 		y_index[i] = prob->y[i];
@@ -468,6 +470,15 @@ void Solver_MCSVM_CS::Solve(double *w)
 						G[m] += w_i[alpha_index_i[m]]*(feature_value);
 					got_feature = prob->x->get_next_feature(feature_index,feature_value,feature_iter);
 				}
+				// experimental
+				// ***
+				if (prob->use_bias)
+				{
+					double *w_i = &w[(w_size-1)*nr_class];
+					for(m=0;m<active_size_i[i];m++)
+						G[m] += w_i[alpha_index_i[m]];
+				}
+				// ***
 				prob->x->free_feature_iterator(feature_iter);
 
 				double minG = CMath::INFTY;
@@ -547,6 +558,15 @@ void Solver_MCSVM_CS::Solve(double *w)
 						w_i[d_ind[m]] += d_val[m]*feature_value;
 					got_feature = prob->x->get_next_feature(feature_index,feature_value,feature_iter);
 				}
+				// experimental
+				// ***
+				if (prob->use_bias)
+				{
+					double *w_i = &w[(w_size-1)*nr_class];
+					for(m=0;m<nz_d;m++)
+						w_i[d_ind[m]] += d_val[m];
+				}
+				// ***
 				prob->x->free_feature_iterator(feature_iter);
 			}
 		}