Reseted the pseudo random number generator

bob/bio/gmm/algorithm/GMM.py

@@ -112,6 +112,10 @@ class GMM (Algorithm):
 
     # Trains using the KMeansTrainer
     logger.info("  -> Training K-Means")
+
+    # Reseting the pseudo random number generator so we can have the same initialization for serial and parallel execution. 
+    del self.rng
+    self.rng = bob.core.random.mt19937(self.init_seed)
     bob.learn.em.train(self.kmeans_trainer, kmeans, array, self.kmeans_training_iterations, self.training_threshold, rng=self.rng)
 
     variances, weights = kmeans.get_variances_and_weights_for_each_cluster(array)
@@ -125,6 +129,9 @@ class GMM (Algorithm):
 
     # Trains the GMM
     logger.info("  -> Training GMM")
+    # Reseting the pseudo random number generator so we can have the same initialization for serial and parallel execution. 
+    del self.rng
+    self.rng = bob.core.random.mt19937(self.init_seed)
     bob.learn.em.train(self.ubm_trainer, self.ubm, array, self.gmm_training_iterations, self.training_threshold, rng=self.rng)
 
 
@@ -198,7 +205,7 @@ class GMM (Algorithm):
     logger.debug(" .... Enrolling with %d feature vectors", array.shape[0])
 
     gmm = bob.learn.em.GMMMachine(self.ubm)
-    gmm.set_variance_thresholds(self.variance_threshold)
+    gmm.set_variance_thresholds(self.variance_threshold)    
     bob.learn.em.train(self.enroll_trainer, gmm, array, self.gmm_enroll_iterations, self.training_threshold, rng=self.rng)
     return gmm
 

bob/bio/gmm/algorithm/ISV.py

@@ -64,6 +64,9 @@ class ISV (GMM):
     logger.info("  -> Training ISV enroller")
     self.isvbase = bob.learn.em.ISVBase(self.ubm, self.subspace_dimension_of_u)
     # train ISV model
+    # Reseting the pseudo random number generator so we can have the same initialization for serial and parallel execution. 
+    del self.rng
+    self.rng = bob.core.random.mt19937(self.init_seed)
     bob.learn.em.train(self.isv_trainer, self.isvbase, data, self.isv_training_iterations, rng=self.rng)
 
 

bob/bio/gmm/algorithm/IVector.py

@@ -94,6 +94,10 @@ class IVector (GMM):
     logger.info("  -> Training IVector enroller")
     self.tv = bob.learn.em.IVectorMachine(self.ubm, self.subspace_dimension_of_t, self.variance_threshold)
 
+    # Reseting the pseudo random number generator so we can have the same initialization for serial and parallel execution. 
+    del self.rng
+    self.rng = bob.core.random.mt19937(self.init_seed)
+
     # train IVector model
     bob.learn.em.train(self.ivector_trainer, self.tv, training_stats, self.tv_training_iterations, rng=self.rng)
 
@@ -125,6 +129,11 @@ class IVector (GMM):
     variance_flooring = 1e-5
     training_features = [numpy.vstack(client) for client in training_features]
     input_dim = training_features[0].shape[1]
+
+    # Reseting the pseudo random number generator so we can have the same initialization for serial and parallel execution. 
+    del self.rng
+    self.rng = bob.core.random.mt19937(self.init_seed)
+    
     self.plda_base = bob.learn.em.PLDABase(input_dim, self.plda_dim_F, self.plda_dim_G, variance_flooring)
     bob.learn.em.train(self.plda_trainer, self.plda_base, training_features, self.plda_training_iterations, rng=self.rng)
 

bob/bio/gmm/tools/gmm.py

@@ -32,6 +32,10 @@ def kmeans_initialize(algorithm, extractor, limit_data = None, force = False, al
     # Perform KMeans initialization
     kmeans_machine = bob.learn.em.KMeansMachine(algorithm.gaussians, data.shape[1])
     # Creates the KMeansTrainer and call the initialization procedure
+    # Reseting the pseudo random number generator so we can have the same initialization for serial and parallel execution. 
+    del algorithm.rng
+    algorithm.rng = bob.core.random.mt19937(algorithm.init_seed)
+    
     algorithm.kmeans_trainer.initialize(kmeans_machine, data, algorithm.rng)
     bob.io.base.create_directories_safe(os.path.dirname(output_file))
     kmeans_machine.save(bob.io.base.HDF5File(output_file, 'w'))