Gaussian Structure

Gaussian Kernel.

Namespace: Accord.Statistics.Kernels
Assembly: Accord.Statistics (in Accord.Statistics.dll) Version: 3.8.0

Syntax

[SerializableAttribute]
public struct Gaussian : IKernel, IKernel<double[]>, 
	IRadialBasisKernel, IDistance, IDistance<double[]>, IDistance<double[], double[]>, 
	IEstimable, IEstimable<double[]>, ICloneable, IReverseDistance, 
	IKernel<Sparse<double>>, IEstimable<Sparse<double>>, 
	IDistance<Sparse<double>>, IDistance<Sparse<double>, Sparse<double>>

<SerializableAttribute>
Public Structure Gaussian
	Implements IKernel, IKernel(Of Double()), 
	IRadialBasisKernel, IDistance, IDistance(Of Double()), IDistance(Of Double(), Double()), 
	IEstimable, IEstimable(Of Double()), ICloneable, IReverseDistance, 
	IKernel(Of Sparse(Of Double)), IEstimable(Of Sparse(Of Double)), 
	IDistance(Of Sparse(Of Double)), IDistance(Of Sparse(Of Double), Sparse(Of Double))

Request Example View Source

The Gaussian type exposes the following members.

Constructors

	Name	Description
	Gaussian	Constructs a new Gaussian Kernel with a given sigma value. To construct from a gamma value, use the FromGamma(Double) named constructor instead.

Top

Properties

	Name	Description
	Gamma	Gets or sets the gamma value for the kernel. When setting gamma, sigma gets updated accordingly (gamma = 0.5/sigma^2).
	Sigma	Gets or sets the sigma value for the kernel. When setting sigma, gamma gets updated accordingly (gamma = 0.5/sigma^2).
	SigmaSquared	Gets or sets the sigma² value for the kernel. When setting sigma², gamma gets updated accordingly (gamma = 0.5/sigma²).

Top

Methods

	Name	Description
	Clone	Creates a new object that is a copy of the current instance.
	Distance(Double, Double)	Computes the squared distance in feature space between two points given in input space.
	Distance(SparseDouble, SparseDouble)	Computes the squared distance in feature space between two points given in input space.
	Distances(Double, Int32)	Computes the set of all distances between all points in a random subset of the data.
	Distances(SparseDouble, Int32)	Computes the set of all distances between all points in a random subset of the data.
	DistancesTDistance, TInput(TInput, Int32, TDistance)	Computes the set of all distances between all points in a random subset of the data.
	Equals	Indicates whether this instance and a specified object are equal. (Inherited from ValueType.)
	Estimate(Double)	Estimate appropriate values for sigma given a data set.
	Estimate(SparseDouble)	Estimate appropriate values for sigma given a data set.
	Estimate(Double, DoubleRange)	Estimate appropriate values for sigma given a data set.
	Estimate(Double, Int32)	Estimates appropriate values for sigma given a data set.
	Estimate(SparseDouble, DoubleRange)	Estimate appropriate values for sigma given a data set.
	Estimate(SparseDouble, Int32)	Estimates appropriate values for sigma given a data set.
	Estimate(Double, Int32, DoubleRange)	Estimates appropriate values for sigma given a data set.
	Estimate(SparseDouble, Int32, DoubleRange)	Estimates appropriate values for sigma given a data set.
	EstimateT(T, Double)	Estimate appropriate values for sigma given a data set.
	EstimateT(T, Double, DoubleRange)	Estimate appropriate values for sigma given a data set.
	EstimateT(T, Double, Int32)	Estimates appropriate values for sigma given a data set.
	EstimateT(T, Double, Int32, DoubleRange)	Estimates appropriate values for sigma given a data set.
	EstimateTInput, TDistance(TInput, TDistance)	Estimate appropriate values for sigma given a data set.
	EstimateTInput, TDistance(TInput, Int32, TDistance)	Estimates appropriate values for sigma given a data set.
	EstimateTInput, TDistance(TInput, TDistance, DoubleRange)	Estimate appropriate values for sigma given a data set.
	EstimateTInput, TDistance(TInput, Int32, TDistance, DoubleRange)	Estimates appropriate values for sigma given a data set.
	FromGamma	Constructs a new Gaussian Kernel with a given gamma value. To construct from a sigma value, use the Gaussian(Double) constructor instead.
	Function(Double)	Gaussian Kernel function.
	Function(Double, Double)	Gaussian Kernel function.
	Function(SparseDouble, SparseDouble)	Gaussian Kernel function.
	GetHashCode	Returns the hash code for this instance. (Inherited from ValueType.)
	GetType	Gets the Type of the current instance. (Inherited from Object.)
	ReverseDistance(Double)	Computes the distance in input space given a distance computed in feature space.
	ReverseDistance(Double, Double)	Computes the squared distance in input space between two points given in feature space.
	ToString	Returns the fully qualified type name of this instance. (Inherited from ValueType.)

Top

Extension Methods

	Name	Description
	Distance	Computes the kernel distance for a kernel function even if it doesn't implement the IDistance interface. Can be used to check the proper implementation of the distance function. (Defined by Tools.)
	HasMethod	Checks whether an object implements a method with the given name. (Defined by ExtensionMethods.)
	IsEqual	Compares two objects for equality, performing an elementwise comparison if the elements are vectors or matrices. (Defined by Matrix.)
	To(Type)	Overloaded. Converts an object into another type, irrespective of whether the conversion can be done at compile time or not. This can be used to convert generic types to numeric types during runtime. (Defined by ExtensionMethods.)
	ToT	Overloaded. Converts an object into another type, irrespective of whether the conversion can be done at compile time or not. This can be used to convert generic types to numeric types during runtime. (Defined by ExtensionMethods.)

Top

Remarks

The Gaussian kernel requires tuning for the proper value of σ. Different approaches to this problem includes the use of brute force (i.e. using a grid-search algorithm) or a gradient ascent optimization.

References:

P. F. Evangelista, M. J. Embrechts, and B. K. Szymanski. Some Properties of the Gaussian Kernel for One Class Learning. Available on: http://www.cs.rpi.edu/~szymansk/papers/icann07.pdf

Reference

Accord.Statistics.Kernels Namespace