GeneticAlgorithm now uses Nth best MLP, where N is a program parameter

maca_common/include/ProgramParameters.hpp

@@ -60,6 +60,7 @@ struct ProgramParameters
   static bool onlyPrefixes;
   static std::map<std::string,std::string> featureModelByClassifier;
   static int nbErrorsToShow;
+  static int nbIndividuals;
 
   private :
 

maca_common/src/ProgramParameters.cpp

@@ -54,3 +54,5 @@ int ProgramParameters::batchSize;
 std::string ProgramParameters::loss;
 std::map<std::string,std::string> ProgramParameters::featureModelByClassifier;
 int ProgramParameters::nbErrorsToShow;
+int ProgramParameters::nbIndividuals;
+

neural_network/src/GeneticAlgorithm.cpp

@@ -39,7 +39,26 @@ void GeneticAlgorithm::init(int nbInputs, const std::string & topology, int nbOu
 
 std::vector<float> GeneticAlgorithm::predict(FeatureModel::FeatureDescription & fd)
 {
-  return generation[0]->mlp.predict(fd);
+  int toAsk = ProgramParameters::nbIndividuals;
+
+  if (toAsk < 0 || toAsk > (int)generation.size())
+  {
+    fprintf(stderr, "ERROR (%s) : trying to save \'%d\' individuals out of a population of \'%lu\'. Aborting.\n", ERRINFO, toAsk, generation.size());
+    exit(1);
+  }
+
+  auto prediction = generation[0]->mlp.predict(fd);
+  for (int i = 1; i < toAsk; i++)
+  {
+    auto otherPrediction = generation[i]->mlp.predict(fd);
+    for (unsigned int j = 0; j < prediction.size(); j++)
+      prediction[j] += otherPrediction[j];
+  }
+
+  for (unsigned int j = 0; j < prediction.size(); j++)
+    prediction[j] /= toAsk;
+
+  return prediction;
 }
 
 float GeneticAlgorithm::update(FeatureModel::FeatureDescription & fd, int gold)
@@ -115,11 +134,24 @@ float GeneticAlgorithm::update(FeatureModel::FeatureDescription & fd, int gold)
 
 void GeneticAlgorithm::save(const std::string & filename)
 {
+  int toSave = ProgramParameters::nbIndividuals;
+
+  if (toSave < 0 || toSave > (int)generation.size())
+  {
+    fprintf(stderr, "ERROR (%s) : trying to save \'%d\' individuals out of a population of \'%lu\'. Aborting.\n", ERRINFO, toSave, generation.size());
+    exit(1);
+  }
+
   File * file = new File(filename, "w");
-  fprintf(file->getDescriptor(), "%u\n", generation[0]->id);
+  for (int i = 0; i < toSave; i++)
+    fprintf(file->getDescriptor(), "%u\n", generation[i]->id);
   delete file;
-  generation[0]->mlp.saveStruct(filename);
-  generation[0]->mlp.saveParameters(filename);
+  
+  for (int i = 0; i < toSave; i++)
+  {
+    generation[i]->mlp.saveStruct(filename);
+    generation[i]->mlp.saveParameters(filename);
+  }
 }
 
 void GeneticAlgorithm::printTopology(FILE * output)
@@ -137,18 +169,28 @@ void GeneticAlgorithm::printTopology(FILE * output)
 
 void GeneticAlgorithm::load(const std::string & filename)
 {
+  std::vector<int> ids;
   File * file = new File(filename, "r");
-  unsigned int bestId;
-  if (fscanf(file->getDescriptor(), "%u\n", &bestId) != 1)
+  unsigned int id;
+  while (fscanf(file->getDescriptor(), "%u\n", &id) == 1)
+    ids.emplace_back(id);
+  delete file;
+
+  if (ids.empty())
   {
-    fprintf(stderr, "ERROR (%s) : expected best id when reading file \'%s\'. Aborting.\n", ERRINFO, filename.c_str());
+    fprintf(stderr, "ERROR (%s) : Missing MLP\''s ids in file \'%s\'. Aborting.\n", ERRINFO, filename.c_str());
     exit(1);
   }
-  delete file;
-  generation.emplace_back(new Individual(bestId));
 
-  generation[0]->mlp.loadStruct(model, filename, 0);
-  generation[0]->mlp.loadParameters(model, filename);
+  ProgramParameters::nbIndividuals = ids.size();
+
+  for (auto & id : ids)
+  {
+    generation.emplace_back(new Individual(id));
+
+    generation.back()->mlp.loadStruct(model, filename, generation.size()-1);
+    generation.back()->mlp.loadParameters(model, filename);
+  }
 }
 
 GeneticAlgorithm::Individual::Individual(dynet::ParameterCollection & model, int nbInputs, const std::string & topology, int nbOutputs) : mlp("MLP_" + std::to_string(idCount))
@@ -244,7 +286,7 @@ void GeneticAlgorithm::Individual::mutate(float probability)
 
       for (unsigned int k = 0; k < nbValues; k++)
         if (choiceWithProbability(probability))
-          thisValues[k] = getRandomValueInRange(1);
+          thisValues[k] = getRandomValueInRange(3);
     }
 }
 

trainer/src/macaon_train.cpp

@@ -99,7 +99,12 @@ po::options_description getOptionsDescription()
     ("bias", po::value<float>()->default_value(1e-8),
       "bias parameter for the Amsgtad or Adam  or Adagrad optimizer");
 
-  desc.add(req).add(opt).add(oracle).add(ams);
+  po::options_description ga("Genetic algorithm related options");
+  ga.add_options()
+    ("nbIndividuals", po::value<int>()->default_value(5),
+      "Number of individuals that will be used to take decisions");
+
+  desc.add(req).add(opt).add(oracle).add(ams).add(ga);
 
   return desc;
 }
@@ -273,6 +278,7 @@ int main(int argc, char * argv[])
   ProgramParameters::beta1 = vm["b1"].as<float>();
   ProgramParameters::beta2 = vm["b2"].as<float>();
   ProgramParameters::bias = vm["bias"].as<float>();
+  ProgramParameters::nbIndividuals = vm["nbIndividuals"].as<int>();
   ProgramParameters::optimizer = vm["optimizer"].as<std::string>();
   ProgramParameters::dicts = vm["dicts"].as<std::string>();
   ProgramParameters::loss = vm["loss"].as<std::string>();