[mlpack-svn] r15932 - mlpack/trunk/src/mlpack/core/kernels

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

Author: rcurtin
Date: Fri Oct  4 12:31:04 2013
New Revision: 15932

Log:
Minor updates for normalized kernels.


Modified:
   mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel.cpp
   mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel.hpp
   mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel_impl.hpp
   mlpack/trunk/src/mlpack/core/kernels/gaussian_kernel.hpp
   mlpack/trunk/src/mlpack/core/kernels/laplacian_kernel.hpp
   mlpack/trunk/src/mlpack/core/kernels/triangular_kernel.hpp

Modified: mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel.cpp
==============================================================================
--- mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel.cpp	(original)
+++ mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel.cpp	Fri Oct  4 12:31:04 2013
@@ -26,8 +26,7 @@
 /**
  * Evaluate the kernel not for two points but for a numerical value.
  */
-double EpanechnikovKernel::Evaluate(const double t)
+double EpanechnikovKernel::Evaluate(const double distance) const
 {
-  double evaluatee = 1.0 - t * t * inverseBandwidthSquared;
-  return (evaluatee > 0.0) ? evaluatee : 0.0;
+  return std::max(0.0, 1 - std::pow(distance, 2.0) * inverseBandwidthSquared);
 }

Modified: mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel.hpp
==============================================================================
--- mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel.hpp	(original)
+++ mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel.hpp	Fri Oct  4 12:31:04 2013
@@ -41,7 +41,13 @@
    * @param b The other input vector.
    */
   template<typename Vec1Type, typename Vec2Type>
-  double Evaluate(const Vec1Type& a, const Vec2Type& b);
+  double Evaluate(const Vec1Type& a, const Vec2Type& b) const;
+
+  /**
+   * Evaluate the Epanechnikov kernel given that the distance between the two
+   * input points is known.
+   */
+  double Evaluate(const double distance) const;
 
   /**
    * Obtains the convolution integral [integral of K(||x-a||) K(||b-x||) dx]
@@ -62,11 +68,6 @@
    */
   double Normalizer(const size_t dimension);
 
-  /**
-   * Evaluate the kernel not for two points but for a numerical value.
-   */
-  double Evaluate(const double t);
-
  private:
   //! Bandwidth of the kernel.
   double bandwidth;

Modified: mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel_impl.hpp
==============================================================================
--- mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel_impl.hpp	(original)
+++ mlpack/trunk/src/mlpack/core/kernels/epanechnikov_kernel_impl.hpp	Fri Oct  4 12:31:04 2013
@@ -17,6 +17,7 @@
 
 template<typename Vec1Type, typename Vec2Type>
 inline double EpanechnikovKernel::Evaluate(const Vec1Type& a, const Vec2Type& b)
+    const
 {
   return std::max(0.0, 1.0 - metric::SquaredEuclideanDistance::Evaluate(a, b)
       * inverseBandwidthSquared);

Modified: mlpack/trunk/src/mlpack/core/kernels/gaussian_kernel.hpp
==============================================================================
--- mlpack/trunk/src/mlpack/core/kernels/gaussian_kernel.hpp	(original)
+++ mlpack/trunk/src/mlpack/core/kernels/gaussian_kernel.hpp	Fri Oct  4 12:31:04 2013
@@ -63,17 +63,18 @@
   }
 
   /**
-   * Evaluation of the Gaussian kernel using a double precision argument.
+   * Evaluation of the Gaussian kernel given the distance between two points.
    *
-   * @param t double value.
+   * @param t The distance between the two points the kernel is evaluated on.
    * @return K(t) using the bandwidth (@f$\mu at f$) specified in the
    *     constructor.
    */
   double Evaluate(double t) const
   {
     // The precalculation of gamma saves us a little computation time.
-    return exp(gamma * t * t);
+    return exp(gamma * std::pow(t, 2.0));
   }
+
   /**
    * Obtain the normalization constant of the Gaussian kernel.
    *
@@ -84,6 +85,7 @@
   {
     return pow(sqrt(2.0 * M_PI) * bandwidth, (double) dimension);
   }
+
   /**
    * Obtain a convolution integral of the Gaussian kernel.
    *

Modified: mlpack/trunk/src/mlpack/core/kernels/laplacian_kernel.hpp
==============================================================================
--- mlpack/trunk/src/mlpack/core/kernels/laplacian_kernel.hpp	(original)
+++ mlpack/trunk/src/mlpack/core/kernels/laplacian_kernel.hpp	Fri Oct  4 12:31:04 2013
@@ -59,13 +59,14 @@
   }
 
   /**
-   * Evaluation of the Laplacian kernel using a double precision argument.
+   * Evaluation of the Laplacian kernel given the distance between two points.
    *
-   * @param t double value.
+   * @param t The distance between the two points the kernel should be evaluated
+   *     on.
    * @return K(t) using the bandwidth (@f$\mu at f$) specified in the
    *     constructor.
    */
-  double Evaluate(double t) const
+  double Evaluate(const double t) const
   {
     // The precalculation of gamma saves us a little computation time.
     return exp(-t / bandwidth);

Modified: mlpack/trunk/src/mlpack/core/kernels/triangular_kernel.hpp
==============================================================================
--- mlpack/trunk/src/mlpack/core/kernels/triangular_kernel.hpp	(original)
+++ mlpack/trunk/src/mlpack/core/kernels/triangular_kernel.hpp	Fri Oct  4 12:31:04 2013
@@ -39,12 +39,23 @@
    * @param b Second vector.
    */
   template<typename Vec1Type, typename Vec2Type>
-  double Evaluate(const Vec1Type& a, const Vec2Type& b)
+  double Evaluate(const Vec1Type& a, const Vec2Type& b) const
   {
     return std::max(0.0, (1 - metric::EuclideanDistance::Evaluate(a, b) /
         bandwidth));
   }
 
+  /**
+   * Evaluate the triangular kernel given that the distance between the two
+   * points is known.
+   *
+   * @param distance The distance between the two points.
+   */
+  double Evaluate(const double distance) const
+  {
+    return std::max(0.0, (1 - distance) / bandwidth);
+  }
+
   //! Get the bandwidth of the kernel.
   double Bandwidth() const { return bandwidth; }
   //! Modify the bandwidth of the kernel.