it.uniroma2.sag.kelp.kernel.standard

Class LinearKernelCombination

java.lang.Object

it.uniroma2.sag.kelp.kernel.Kernel

it.uniroma2.sag.kelp.kernel.KernelCombination

it.uniroma2.sag.kelp.kernel.standard.LinearKernelCombination

public class LinearKernelCombination
extends KernelCombination

Weighted Linear Combination of Kernels
Given a kernel some kernel K1...Km, with weights c1,...cn, the combination formula is:
SUM(Ki*ci)

Author:

Simone Filice

Field Summary

Fields inherited from class it.uniroma2.sag.kelp.kernel.KernelCombination

toCombine

Constructor Summary

Constructors

Constructor and Description
LinearKernelCombination()

Method Summary

Methods

Modifier and Type	Method and Description
void	addKernel(float weight, Kernel kernel) Adds a kernel with a corresponding weight to the linear combination of kernels
List<Float>	getWeights()
protected float	kernelComputation(Example exA, Example exB) Returns the kernel similarity between the given examples.
void	normalizeWeights() Scales the weights in order to make their sum being equal to 1
void	setWeights(List<Float> weights)

Methods inherited from class it.uniroma2.sag.kelp.kernel.KernelCombination

getToCombine, setToCombine

Methods inherited from class it.uniroma2.sag.kelp.kernel.Kernel

disableCache, getKernelCache, getKernelComputations, getNumberOfHits, getNumberOfMisses, getSquaredNormCache, innerProduct, load, reset, save, setKernelCache, setSquaredNormCache, squaredNorm, squaredNormOfTheDifference

Methods inherited from class java.lang.Object

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

Constructor Detail

LinearKernelCombination

public LinearKernelCombination()

Method Detail

getWeights

public List<Float> getWeights()

Returns:

the weights

setWeights

public void setWeights(List<Float> weights)

Parameters:

weights - the weights to set

addKernel

public void addKernel(float weight,
             Kernel kernel)

Adds a kernel with a corresponding weight to the linear combination of kernels

Parameters:

weight - the weight of the kernel to be added

kernel - the kernel to be added

normalizeWeights

public void normalizeWeights()

Scales the weights in order to make their sum being equal to 1

kernelComputation

protected float kernelComputation(Example exA,
                      Example exB)

Description copied from class: Kernel

Returns the kernel similarity between the given examples. This corresponds to the inner product in the implicit Reproducing Kernel Hilbert Space (RKHS)

Specified by:

kernelComputation in class Kernel

Parameters:

exA - the first example

exB - the second example

Returns:

the kernel operation between exA and exB