com.rapidminer.operator.learner.functions.kernel.jmysvm.kernel
Class Kernel

java.lang.Object
  com.rapidminer.operator.learner.functions.kernel.jmysvm.kernel.Kernel

All Implemented Interfaces:

java.io.Serializable

Direct Known Subclasses:

KernelAnova, KernelDot, KernelEpanechnikov, KernelGaussianCombination, KernelMultiquadric, KernelNeural, KernelPolynomial, KernelRadial

public abstract class Kernel
extends java.lang.Object
implements java.io.Serializable

Abstract base class for all kernels.

Author:

Stefan Rueping, Ingo Mierswa

See Also:

Serialized Form

Field Summary

protected int	cache_MB Size of cache in MB
protected int	dim dimension of the examples
protected int	examples_total number of examples after shrinking
protected Cache	kernel_cache Kernel cache
protected int	kernel_cache_size Number of elements in cache
protected SVMExamples	the_examples Container for the examples, parameters etc.

Constructor Summary

Kernel()
Class constructor

Method Summary

double[]	calculate_K_row(double[] result, int i)
abstract double	calculate_K(int[] x_index, double[] x_att, int[] y_index, double[] y_att) Calculates kernel value of vectors x and y
double	calculate_K(int i, int j) Calculate K(i,j)
double	calculate_K(SVMExample x, SVMExample y)
double[]	get_row(int i) Gets a kernel row
int	getCacheSize()
abstract java.lang.String	getDistanceFormula(double[] x, java.lang.String[] attributeNames)
void	init_kernel_cache(int size) Inits the kernel cache.
void	init(SVMExamples examples, int cacheSizeMB) Init the kernel
double	innerproduct(int[] x_index, double[] x_att, int[] y_index, double[] y_att) calculate inner product
double	norm2(int[] x_index, double[] x_att, int[] y_index, double[] y_att) calculate ||x-y||^2
void	set_examples_size(int new_examples_total) Sets the number of examples to new value
void	swap(int pos1, int pos2) swap two training examples
java.lang.String	toString() Output as String

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

the_examples

protected SVMExamples the_examples

Container for the examples, parameters etc.

dim

protected int dim

dimension of the examples

kernel_cache

protected transient Cache kernel_cache

Kernel cache

kernel_cache_size

protected int kernel_cache_size

Number of elements in cache

cache_MB

protected int cache_MB

Size of cache in MB

examples_total

protected int examples_total

number of examples after shrinking

Constructor Detail

Kernel

public Kernel()

Class constructor

Method Detail

toString

public java.lang.String toString()

Output as String

Overrides:

toString in class java.lang.Object

init

public void init(SVMExamples examples,
                 int cacheSizeMB)

Init the kernel

Parameters:

examples - Container for the examples.

calculate_K

public abstract double calculate_K(int[] x_index,
                                   double[] x_att,
                                   int[] y_index,
                                   double[] y_att)

Calculates kernel value of vectors x and y

getDistanceFormula

public abstract java.lang.String getDistanceFormula(double[] x,
                                                    java.lang.String[] attributeNames)

innerproduct

public double innerproduct(int[] x_index,
                           double[] x_att,
                           int[] y_index,
                           double[] y_att)

calculate inner product

norm2

public double norm2(int[] x_index,
                    double[] x_att,
                    int[] y_index,
                    double[] y_att)

calculate ||x-y||^2

get_row

public double[] get_row(int i)

Gets a kernel row

init_kernel_cache

public void init_kernel_cache(int size)

Inits the kernel cache.

Parameters:

size - of the cache in MB

getCacheSize

public int getCacheSize()

set_examples_size

public void set_examples_size(int new_examples_total)

Sets the number of examples to new value

calculate_K

public double calculate_K(int i,
                          int j)

Calculate K(i,j)

calculate_K

public double calculate_K(SVMExample x,
                          SVMExample y)

calculate_K_row

public double[] calculate_K_row(double[] result,
                                int i)

swap

public void swap(int pos1,
                 int pos2)

swap two training examples

Parameters:

pos1 -

pos2 -