fix a bug for linear coef

Author: QinbinLi

include/thundersvm/dataset.h

@@ -80,6 +80,7 @@ class DataSet {
 
     const vector<int> original_index(int y_i, int y_j) const;
 
+	const bool is_zero_based() const;
 private:
     vector<float_type> y_;
     node2d instances_;
@@ -89,5 +90,6 @@ class DataSet {
     vector<int> count_; //the number of instances of each class
     vector<int> label_;
     vector<int> perm_;
+	bool zero_based = 0; //is zero_based format dataset?
 };
 #endif //THUNDERSVM_DATASET_H

include/thundersvm/model/svmmodel.h

@@ -105,7 +105,7 @@ class SvmModel {
     //return prob_predict
     const vector<float> &get_prob_predict() const;
 
-    void compute_linear_coef_single_model(size_t n_feature);
+    void compute_linear_coef_single_model(size_t n_feature, const bool zero_based);
     //get the params, for scikit load params
     void get_param(char* kernel_type, int* degree, float* gamma, float* coef0, int* probability);
 

src/thundersvm/dataset.cpp

@@ -77,6 +77,7 @@ void DataSet::load_from_file(string file_name) {
                     float v;
                     CHECK_EQ(sscanf(tuple.c_str(), "%d:%f", &i, &v), 2) << "read error, using [index]:[value] format";
                     instances_thread[tid].back().emplace_back(i, v);
+					if(i == 0 && zero_based == 0) zero_based = 1;
                     if (i > local_feature[tid]) local_feature[tid] = i;
                 };
 
@@ -158,7 +159,7 @@ void DataSet::load_from_dense(int row_size, int features, float* data, float* la
         if(label != NULL)
             y_.push_back(label[i]);
         instances_.emplace_back();
-        for(int j = 0; j < features; j++){
+        for(int j = 1; j <= features; j++){
             ind = j;
             v = data[off];
             off++;
@@ -285,4 +286,6 @@ const vector<float_type> &DataSet::y() const {
     return y_;
 }
 
-
+const bool DataSet::is_zero_based() const{
+	return zero_based;
+}

src/thundersvm/model/nusvr.cpp

@@ -36,7 +36,7 @@ void NuSVR::train(const DataSet &dataset, SvmParam param) {
     save_svr_coef(alpha_2, dataset.instances());
 
     if(param.kernel_type == SvmParam::LINEAR){
-        compute_linear_coef_single_model(dataset.n_features());
+        compute_linear_coef_single_model(dataset.n_features(), dataset.is_zero_based());
     }
 }
 

src/thundersvm/model/oneclass_svc.cpp

@@ -51,7 +51,7 @@ void OneClassSVC::train(const DataSet &dataset, SvmParam param) {
     coef.copy_from(coef_vec.data(), coef_vec.size());
 
     if(param.kernel_type == SvmParam::LINEAR){
-        compute_linear_coef_single_model(dataset.n_features());
+        compute_linear_coef_single_model(dataset.n_features(), dataset.is_zero_based());
     }
 }
 

src/thundersvm/model/svc.cpp

@@ -109,17 +109,24 @@ void SVC::train(const DataSet &dataset, SvmParam param) {
     ///TODO: Use coef instead of alpha_data to compute linear_coef_data
     if(param.kernel_type == SvmParam::LINEAR){
         int k = 0;
-        linear_coef.resize(n_binary_models * dataset_.n_features());
+        if(dataset_.is_zero_based())
+			linear_coef.resize(n_binary_models * (dataset_.n_features()+1));
+		else	
+			linear_coef.resize(n_binary_models * dataset_.n_features());
         float_type *linear_coef_data = linear_coef.host_data();
         for (int i = 0; i < n_classes; i++){
             for (int j = i + 1; j < n_classes; j++){
                 const float_type *alpha_data = alpha[k].host_data();
                 DataSet::node2d ins = dataset_.instances(i, j);//get instances of class i and j
                 for(int iid = 0; iid < ins.size(); iid++) {
                     for (int fid = 0; fid < ins[iid].size(); fid++) {
-                        if(alpha_data[iid] != 0)
-                            linear_coef_data[k * dataset_.n_features() + ins[iid][fid].index - 1] += alpha_data[iid] * ins[iid][fid].value;
-                    }
+                        if(alpha_data[iid] != 0){
+							if(dataset_.is_zero_based())
+                            	linear_coef_data[k * dataset_.n_features() + ins[iid][fid].index] += alpha_data[iid] * ins[iid][fid].value;
+                            else
+								linear_coef_data[k * dataset_.n_features() + ins[iid][fid].index - 1] += alpha_data[iid] * ins[iid][fid].value;
+                    	}
+					}
                 }
                 k++;
             }

src/thundersvm/model/svmmodel.cpp

@@ -381,17 +381,23 @@ void SvmModel::get_param(char* kernel_type, int* degree, float* gamma, float* co
     *probability = param.probability;
 }
 
-void SvmModel::compute_linear_coef_single_model(size_t n_feature){
-    linear_coef.resize(n_feature);
+void SvmModel::compute_linear_coef_single_model(size_t n_feature, const bool zero_based){
+	if(zero_based)
+		linear_coef.resize(n_feature+1);
+	else
+    	linear_coef.resize(n_feature);
     float_type* linear_coef_data = linear_coef.host_data();
     float_type* coef_data = coef.host_data();
     for(int i = 0; i < n_total_sv; i++){
         for(int j = 0; j < sv[i].size(); j++){
-            linear_coef_data[sv[i][j].index - 1] += coef_data[i] * sv[i][j].value;
+            if(zero_based)
+				linear_coef_data[sv[i][j].index] += coef_data[i] * sv[i][j].value;
+			else
+				linear_coef_data[sv[i][j].index - 1] += coef_data[i] * sv[i][j].value;
         }
     }
 }
 
 int SvmModel::get_sv_max_index() const{
     return sv_max_index;
-}
+}

src/thundersvm/model/svr.cpp

@@ -36,7 +36,7 @@ void SVR::train(const DataSet &dataset, SvmParam param) {
     save_svr_coef(alpha_2, dataset.instances());
 
     if(param.kernel_type == SvmParam::LINEAR){
-        compute_linear_coef_single_model(dataset.n_features());
+        compute_linear_coef_single_model(dataset.n_features(), dataset.is_zero_based());
     }
 }