Added featureExtraction option

maca_common/include/ProgramParameters.hpp

@@ -70,6 +70,7 @@ struct ProgramParameters
   static bool printOutputEntropy;
   static std::string tapeToMask;
   static float maskRate;
+  static bool featureExtraction;
 
   private :
 

maca_common/src/ProgramParameters.cpp

@@ -64,4 +64,5 @@ bool ProgramParameters::printOutputEntropy;
 int ProgramParameters::dictCapacity;
 std::string ProgramParameters::tapeToMask;
 float ProgramParameters::maskRate;
+bool ProgramParameters::featureExtraction;
 

trainer/src/Trainer.cpp

@@ -52,7 +52,7 @@ void Trainer::computeScoreOnDev()
     else
     {
       // Print current iter advancement in percentage
-      if (ProgramParameters::interactive)
+      if (ProgramParameters::interactive && !ProgramParameters::featureExtraction)
       {
         int totalSize = ProgramParameters::devTapeSize;
         int steps = devConfig->getHead();
@@ -200,7 +200,7 @@ void Trainer::train()
         }
 
         // Print current iter advancement in percentage
-        if (ProgramParameters::interactive)
+        if (ProgramParameters::interactive && !ProgramParameters::featureExtraction)
         {
           int totalSize = ProgramParameters::iterationSize == -1 ? ProgramParameters::tapeSize : ProgramParameters::iterationSize;
           int steps = ProgramParameters::iterationSize == -1 ? trainConfig.getHead() : nbSteps;
@@ -211,12 +211,15 @@ void Trainer::train()
           }
         }
 
-        auto weightedActions = tm.getCurrentClassifier()->weightActions(trainConfig);
         std::string pAction = "";
         std::string oAction = "";
-
         bool pActionIsZeroCost = false;
 
+        Classifier::WeightedActions weightedActions;
+        if (!ProgramParameters::featureExtraction)
+        {
+          weightedActions = tm.getCurrentClassifier()->weightActions(trainConfig);
+
           for (auto & it : weightedActions)
             if (it.first)
             {
@@ -232,6 +235,11 @@ void Trainer::train()
 
           if (pAction == oAction)
             pActionIsZeroCost = true;
+        }
+        else
+        {
+          oAction = tm.getCurrentClassifier()->getZeroCostActions(trainConfig)[0];
+        }
 
         if (oAction.empty())
           oAction = tm.getCurrentClassifier()->getDefaultAction();
@@ -252,6 +260,7 @@ void Trainer::train()
           exit(1);
         }
 
+        if (!ProgramParameters::featureExtraction)
           tm.getCurrentClassifier()->trainOnExample(trainConfig, tm.getCurrentClassifier()->getActionIndex(oAction));
 
         TI.addTrainExample(tm.getCurrentClassifier()->name);
@@ -262,6 +271,13 @@ void Trainer::train()
 
         std::string actionName = "";
 
+        //ici
+        if (ProgramParameters::featureExtraction)
+        {
+          auto features = tm.getCurrentClassifier()->getFeatureModel()->getFeatureDescription(trainConfig).featureValues();
+          fprintf(stdout, "%s\t%s\n", oAction.c_str(), features.c_str());
+        }
+
         if (TI.getEpoch() >= k && choiceWithProbability(ProgramParameters::dynamicProbability))
         {
           actionName = pAction;

trainer/src/macaon_train.cpp

@@ -85,7 +85,8 @@ po::options_description getOptionsDescription()
       "The name of the Tape for which some of the elements will be masked.")
     ("maskRate", po::value<float>()->default_value(0.0),
       "The rate of elements of the Tape that will be masked.")
-    ("printTime", "Print time on stderr")
+    ("printTime", "Print time on stderr.")
+    ("featureExtraction", "Use macaon only a feature extractor, print corpus to stdout.")
     ("shuffle", po::value<bool>()->default_value(true),
       "Shuffle examples after each iteration");
 
@@ -268,6 +269,7 @@ int main(int argc, char * argv[])
   ProgramParameters::debug = vm.count("debug") == 0 ? false : true;
   ProgramParameters::printEntropy = vm.count("printEntropy") == 0 ? false : true;
   ProgramParameters::printTime = vm.count("printTime") == 0 ? false : true;
+  ProgramParameters::featureExtraction = vm.count("featureExtraction") == 0 ? false : true;
   ProgramParameters::trainName = vm["train"].as<std::string>();
   ProgramParameters::devName = vm["dev"].as<std::string>();
   ProgramParameters::lang = vm["lang"].as<std::string>();

transition_machine/include/Classifier.hpp

@@ -197,6 +197,8 @@ class Classifier
   ///
   /// @return The number of actions.
   unsigned int getNbActions();
+  /// @brief Get a pointer to the FeatureModel.
+  FeatureModel * getFeatureModel();
 };
 
 #endif

transition_machine/include/FeatureModel.hpp

@@ -64,6 +64,10 @@ class FeatureModel
     ///
     /// @return The string representing this FeatureDescription
     std::string toString();
+    /// @brief Return a string representing the values of the features
+    ///
+    /// @return The string representing the values of the features
+    std::string featureValues();
   };
 
   private :

transition_machine/src/Classifier.cpp

@@ -328,3 +328,8 @@ unsigned int Classifier::getNbActions()
   return as->size();
 }
 
+FeatureModel * Classifier::getFeatureModel()
+{
+  return fm.get();
+}
+

transition_machine/src/FeatureModel.cpp

@@ -133,3 +133,17 @@ FeatureModel::FeatureValue::FeatureValue()
 {
 }
 
+std::string FeatureModel::FeatureDescription::featureValues()
+{
+  std::string res;
+
+  for (auto & feature : values)
+    for (auto & value : feature.values)
+      res += value + "\t";
+
+  if (!res.empty())
+    res.pop_back();
+
+  return res;
+}
+