Add --random_initialization for mlpack_hmm_train.

HISTORY.md

@@ -22,6 +22,9 @@
     mlpack_allknn and mlpack_allkfn programs will remain as copies until mlpack
     3.0.0.
 
+  * Add --random_initialization option to mlpack_hmm_train, for use when no
+    labels are provided.
+
 ### mlpack 2.0.1
 ###### 2016-02-04
   * Fix CMake to properly detect when MKL is being used with Armadillo.

src/mlpack/methods/hmm/hmm_train_main.cpp

@@ -24,7 +24,9 @@ PROGRAM_INFO("Hidden Markov Model (HMM) Training", "This program allows a "
     "\n\n"
     "The HMM is trained with the Baum-Welch algorithm if no labels are "
     "provided.  The tolerance of the Baum-Welch algorithm can be set with the "
-    "--tolerance option."
+    "--tolerance option.  In general it is a good idea to use random "
+    "initialization in this case, which can be specified with the "
+    "--random_initialization (-r) option."
     "\n\n"
     "Optionally, a pre-created HMM model can be used as a guess for the "
     "transition matrix and emission probabilities; this is specifiable with "
@@ -47,6 +49,8 @@ PARAM_STRING("output_model_file", "File to save trained HMM to.", "o",
     "output_hmm.xml");
 PARAM_INT("seed", "Random seed.  If 0, 'std::time(NULL)' is used.", "s", 0);
 PARAM_DOUBLE("tolerance", "Tolerance of the Baum-Welch algorithm.", "T", 1e-5);
+PARAM_FLAG("random_initialization", "Initialize emissions and transition "
+    "matrices with a uniform random distribution.", "r");
 
 using namespace mlpack;
 using namespace mlpack::hmm;
@@ -296,6 +300,21 @@ int main(int argc, char** argv)
       HMM<DiscreteDistribution> hmm(size_t(states),
           DiscreteDistribution(maxEmission), tolerance);
 
+      // Initialize with random starting point.
+      if (CLI::HasParam("random_initialization"))
+      {
+        hmm.Transition().randu();
+        for (size_t c = 0; c < hmm.Transition().n_cols; ++c)
+          hmm.Transition().col(c) /= arma::accu(hmm.Transition().col(c));
+
+        for (size_t e = 0; e < hmm.Emission().size(); ++e)
+        {
+          hmm.Emission()[e].Probabilities().randu();
+          hmm.Emission()[e].Probabilities() /=
+              arma::accu(hmm.Emission()[e].Probabilities());
+        }
+      }
+
       // Now train it.  Pass the already-loaded training data.
       Train::Apply(hmm, &trainSeq);
     }
@@ -314,6 +333,22 @@ int main(int argc, char** argv)
       HMM<GaussianDistribution> hmm(size_t(states),
           GaussianDistribution(dimensionality), tolerance);
 
+      // Initialize with random starting point.
+      if (CLI::HasParam("random_initialization"))
+      {
+        hmm.Transition().randu();
+        for (size_t c = 0; c < hmm.Transition().n_cols; ++c)
+          hmm.Transition().col(c) /= arma::accu(hmm.Transition().col(c));
+
+        for (size_t e = 0; e < hmm.Emission().size(); ++e)
+        {
+          hmm.Emission()[e].Mean().randu();
+          // Generate random covariance.
+          arma::mat r = arma::randu<arma::mat>(dimensionality, dimensionality);
+          hmm.Emission()[e].Covariance(r * r.t());
+        }
+      }
+
       // Now train it.
       Train::Apply(hmm, &trainSeq);
     }
@@ -336,6 +371,33 @@ int main(int argc, char** argv)
       HMM<GMM> hmm(size_t(states), GMM(size_t(gaussians), dimensionality),
           tolerance);
 
+      // Initialize with random starting point.
+      if (CLI::HasParam("random_initialization"))
+      {
+        hmm.Transition().randu();
+        for (size_t c = 0; c < hmm.Transition().n_cols; ++c)
+          hmm.Transition().col(c) /= arma::accu(hmm.Transition().col(c));
+
+        for (size_t e = 0; e < hmm.Emission().size(); ++e)
+        {
+          // Random weights.
+          hmm.Emission()[e].Weights().randu();
+          hmm.Emission()[e].Weights() /=
+              arma::accu(hmm.Emission()[e].Weights());
+
+          // Random means and covariances.
+          for (int g = 0; g < gaussians; ++g)
+          {
+            hmm.Emission()[e].Component(g).Mean().randu();
+
+            // Generate random covariance.
+            arma::mat r = arma::randu<arma::mat>(dimensionality,
+                dimensionality);
+            hmm.Emission()[e].Component(g).Covariance(r * r.t());
+          }
+        }
+      }
+
       // Issue a warning if the user didn't give labels.
       if (!CLI::HasParam("labels_file"))
         Log::Warn << "Unlabeled training of GMM HMMs is almost certainly not "