[mlpack-svn] r14906 - mlpack/trunk/src/mlpack/methods/hmm

[fastlab-svn at coffeetalk-1.cc.gatech.edu]

Author: rcurtin
Date: 2013-04-15 17:16:37 -0400 (Mon, 15 Apr 2013)
New Revision: 14906

Modified:
   mlpack/trunk/src/mlpack/methods/hmm/hmm.hpp
   mlpack/trunk/src/mlpack/methods/hmm/hmm_impl.hpp
Log:
Allow parameterization of the tolerance of the Baum-Welch algorithm.


Modified: mlpack/trunk/src/mlpack/methods/hmm/hmm.hpp
===================================================================
--- mlpack/trunk/src/mlpack/methods/hmm/hmm.hpp	2013-04-15 21:06:58 UTC (rev 14905)
+++ mlpack/trunk/src/mlpack/methods/hmm/hmm.hpp	2013-04-15 21:16:37 UTC (rev 14906)
@@ -84,11 +84,17 @@
    * observations is taken from the emissions variable, so it is important that
    * the given default emission distribution is set with the correct
    * dimensionality.  Alternately, set the dimensionality with Dimensionality().
+   * Optionally, the tolerance for convergence of the Baum-Welch algorithm can
+   * be set.
    *
    * @param states Number of states.
    * @param emissions Default distribution for emissions.
+   * @param tolerance Tolerance for convergence of training algorithm
+   *      (Baum-Welch).
    */
-  HMM(const size_t states, const Distribution emissions);
+  HMM(const size_t states,
+      const Distribution emissions,
+      const double tolerance = 1e-5);
 
   /**
    * Create the Hidden Markov Model with the given transition matrix and the
@@ -103,10 +109,17 @@
    * The emission matrix should be such that E(i, j) is the probability of
    * emission i while in state j.  The columns of the matrix should sum to 1.
    *
+   * Optionally, the tolerance for convergence of the Baum-Welch algorithm can
+   * be set.
+   *
    * @param transition Transition matrix.
    * @param emission Emission distributions.
+   * @param tolerance Tolerance for convergence of training algorithm
+   *      (Baum-Welch).
    */
-  HMM(const arma::mat& transition, const std::vector<Distribution>& emission);
+  HMM(const arma::mat& transition,
+      const std::vector<Distribution>& emission,
+      const double tolerance = 1e-5);
 
   /**
    * Train the model using the Baum-Welch algorithm, with only the given
@@ -123,6 +136,11 @@
    * with labeled data first, and then continue to train the model using this
    * overload of Train() with unlabeled data.
    *
+   * The tolerance of the Baum-Welch algorithm can be set either in the
+   * constructor or with the Tolerance() method.  When the change in
+   * log-likelihood of the model between iterations is less than the tolerance,
+   * the Baum-Welch algorithm terminates.
+   *
    * @note
    * Train() can be called multiple times with different sequences; each time it
    * is called, it uses the current parameters of the HMM as a starting point
@@ -247,6 +265,11 @@
   //! Set the dimensionality of observations.
   size_t& Dimensionality() { return dimensionality; }
 
+  //! Get the tolerance of the Baum-Welch algorithm.
+  double Tolerance() const { return tolerance; }
+  //! Modify the tolerance of the Baum-Welch algorithm.
+  double& Tolerance() { return tolerance; }
+
  private:
   // Helper functions.
 
@@ -287,6 +310,9 @@
 
   //! Dimensionality of observations.
   size_t dimensionality;
+
+  //! Tolerance of Baum-Welch algorithm.
+  double tolerance;
 };
 
 }; // namespace hmm

Modified: mlpack/trunk/src/mlpack/methods/hmm/hmm_impl.hpp
===================================================================
--- mlpack/trunk/src/mlpack/methods/hmm/hmm_impl.hpp	2013-04-15 21:06:58 UTC (rev 14905)
+++ mlpack/trunk/src/mlpack/methods/hmm/hmm_impl.hpp	2013-04-15 21:16:37 UTC (rev 14906)
@@ -19,10 +19,13 @@
  * given number of emission states.
  */
 template<typename Distribution>
-HMM<Distribution>::HMM(const size_t states, const Distribution emissions) :
+HMM<Distribution>::HMM(const size_t states,
+                       const Distribution emissions,
+                       const double tolerance) :
     transition(arma::ones<arma::mat>(states, states) / (double) states),
     emission(states, /* default distribution */ emissions),
-    dimensionality(emissions.Dimensionality())
+    dimensionality(emissions.Dimensionality()),
+    tolerance(tolerance)
 { /* nothing to do */ }
 
 /**
@@ -31,9 +34,11 @@
  */
 template<typename Distribution>
 HMM<Distribution>::HMM(const arma::mat& transition,
-                       const std::vector<Distribution>& emission) :
+                       const std::vector<Distribution>& emission,
+                       const double tolerance) :
     transition(transition),
-    emission(emission)
+    emission(emission),
+    tolerance(tolerance)
 {
   // Set the dimensionality, if we can.
   if (emission.size() > 0)
@@ -160,7 +165,7 @@
     Log::Debug << "Iteration " << iter << ": log-likelihood " << loglik
         << std::endl;
 
-    if (fabs(oldLoglik - loglik) < 1e-5)
+    if (std::abs(oldLoglik - loglik) < tolerance)
     {
       Log::Debug << "Converged after " << iter << " iterations." << std::endl;
       break;