Added the ability to learn non-negative weight vectors to the structural_assignment_trainer

object.

dlib/svm/structural_assignment_trainer.h

@@ -55,6 +55,16 @@ namespace dlib
             return num_threads;
         }
 
+        bool learns_nonnegative_weights (
+        ) const { return learn_nonnegative_weights; }
+       
+        void set_learns_nonnegative_weights (
+            bool value
+        )
+        {
+            learn_nonnegative_weights = value;
+        }
+
         void set_epsilon (
             double eps_
         )
@@ -183,7 +193,13 @@ namespace dlib
 
             matrix<double,0,1> weights; 
 
-            solver(prob, weights);
+            unsigned long num_nonnegative = 0;
+            if (learn_nonnegative_weights)
+            {
+                num_nonnegative = fe.num_features();
+            }
+
+            solver(prob, weights, num_nonnegative);
 
             return assignment_function<feature_extractor>(weights,fe,force_assignment);
 
@@ -192,6 +208,7 @@ namespace dlib
 
     private:
 
+        bool learn_nonnegative_weights;
         bool force_assignment;
         double C;
         oca solver;
@@ -208,6 +225,7 @@ namespace dlib
             eps = 0.1;
             num_threads = 2;
             max_cache_size = 40;
+            learn_nonnegative_weights = false;
         }
 
         feature_extractor fe;

dlib/svm/structural_assignment_trainer_abstract.h

@@ -52,6 +52,7 @@ namespace dlib
                 - #get_max_cache_size() == 40
                 - #get_feature_extractor() == a default initialized feature_extractor
                 - #forces_assignment() == false
+                - #learns_nonnegative_weights() == false
         !*/
 
         explicit structural_assignment_trainer (
@@ -66,6 +67,7 @@ namespace dlib
                 - #get_max_cache_size() == 40
                 - #get_feature_extractor() == fe 
                 - #forces_assignment() == false
+                - #learns_nonnegative_weights() == false
         !*/
 
         const feature_extractor& get_feature_extractor (
@@ -162,6 +164,23 @@ namespace dlib
                 - returns a copy of the optimizer used to solve the structural SVM problem.  
         !*/
 
+        bool learns_nonnegative_weights (
+        ) const;
+        /*!
+            ensures
+                - The output of training is a weight vector that defines the behavior of an
+                  assignment_function object.  If learns_nonnegative_weights() == true then
+                  the resulting weight vector will always have non-negative entries.  
+        !*/
+       
+        void set_learns_nonnegative_weights (
+            bool value
+        );
+        /*!
+            ensures
+                - #learns_nonnegative_weights() == value
+        !*/
+
         void set_c (
             double C 
         );