ml::Kernels::DifferentiableRBFKernel< DiffRBF > Class Template Reference

Differentiable radial basis function kernel. More...

#include <Kernels.hpp>

Inheritance diagram for ml::Kernels::DifferentiableRBFKernel< DiffRBF >:

Collaboration diagram for ml::Kernels::DifferentiableRBFKernel< DiffRBF >:

Public Member Functions
void	gradient (const Eigen::Ref< const Eigen::VectorXd > x1, const Eigen::Ref< const Eigen::VectorXd > x2, Eigen::Ref< Eigen::VectorXd > dydx1) const override
	Gradient of the kernel \( K(\vec{x}_1, \vec{x}_2) \) over the first feature vector. The gradient over the second vector can be calculated by swapping them, due to the symmetry of the kernel function. More...
Public Member Functions inherited from ml::Kernels::Kernel
virtual	~Kernel ()
	Virtual destructor.
virtual unsigned int	dim () const =0
	Dimension of the feature space.
Public Member Functions inherited from ml::Kernels::RBFKernel< DifferentiableRadialBasisFunction >
	RBFKernel (std::unique_ptr< const DifferentiableRadialBasisFunction > &&rbf, unsigned int dim)
	Constructor. More...
	RBFKernel (const RBFKernel &other)=delete
	RBFKernel (RBFKernel &&other)=default
RBFKernel &	operator= (const RBFKernel &other)=delete
RBFKernel &	operator= (RBFKernel &&other)=default
double	value (const Eigen::Ref< const Eigen::VectorXd > x1, const Eigen::Ref< const Eigen::VectorXd > x2) const override
	Value of the kernel \( K(\vec{x}_1, \vec{x}_2) \). More...
unsigned int	dim () const override
	Dimension of the feature space.

Additional Inherited Members
Protected Member Functions inherited from ml::Kernels::Kernel
void	validate_arguments (const Eigen::Ref< const Eigen::VectorXd > x1, const Eigen::Ref< const Eigen::VectorXd > x2) const
Protected Attributes inherited from ml::Kernels::RBFKernel< DifferentiableRadialBasisFunction >
std::unique_ptr< const DifferentiableRadialBasisFunction >	rbf_

Detailed Description

template<class DiffRBF = DifferentiableRadialBasisFunction>
class ml::Kernels::DifferentiableRBFKernel< DiffRBF >

Differentiable radial basis function kernel.

Template Parameters

DiffRBF

DifferentiableRadialBasisFunction or its child class.

Member Function Documentation

gradient()

template<class DiffRBF = DifferentiableRadialBasisFunction>

void ml::Kernels::DifferentiableRBFKernel< DiffRBF >::gradient ( const Eigen::Ref< const Eigen::VectorXd >  x1, const Eigen::Ref< const Eigen::VectorXd >  x2, Eigen::Ref< Eigen::VectorXd >  dydx1  ) const

inlineoverridevirtual

Gradient of the kernel \( K(\vec{x}_1, \vec{x}_2) \) over the first feature vector. The gradient over the second vector can be calculated by swapping them, due to the symmetry of the kernel function.

Parameters

[in]	x1	First feature vector.
[in]	x2	Second feature vector.
[out]	Gradient	over x1. Must have the same size as x1 and x2.

Exceptions

std::invalid_argument

If x1.size() != #dim(), x2.size() != #dim() or dydx1.size() != #dim().

Implements ml::Kernels::DifferentiableKernel.

---

The documentation for this class was generated from the following file:

• ML/Kernels.hpp