Package gov.sandia.cognition.learning.function.kernel

Provides kernel functions.

See:
          Description

Interface Summary
Kernel<InputType> The Kernel interface the functionality required from an object that implements a kernel function.
KernelContainer<InputType> Defines an object that contains a Kernel.
 

Class Summary
DefaultKernelContainer<InputType> The DefaultKernelContainer class implements an object that contains a kernel inside.
DefaultKernelsContainer<InputType> The DefaultKernelsContainer class implements a container of kernels.
ExponentialKernel<InputType> The ExponentialKernel class implements a kernel that applies the exponential function to the result of another kernel.
KernelDistanceMetric<InputType> The KernelDistanceMetric class implements a distance metric that utilizes an underlying Kernel for computing the distance.
KernelUtil A utility class for dealing with kernels.
LinearKernel The LinearKernel class implements the most basic kernel: it just does the actual inner product between two vectors.
NormalizedKernel<InputType> The NormalizedKernel class implements an Kernel that returns a normalized value between 0.0 and 1.0 by normalizing the results of a given kernel.
PolynomialKernel The PolynomialKernel class implements a kernel for two given vectors that is the polynomial function:
(x dot y + c)^d
d is the degree of the polynomial, which must be a positive integer.
ProductKernel<InputType> The ProductKernel class implements a kernel that takes the product of applying multiple kernels to the same pair of inputs.
RadialBasisKernel The RadialBasisKernel implements the standard radial basis kernel, which is:
exp( -||x - y||^2 / (2 * sigma^2) )
where sigma is the parameter that controls the bandwidth of the kernel.
ScalarFunctionKernel<InputType> The ScalarFunctionKernel class implements a kernel that applies a scalar function two the two inputs to the kernel and then returns their product.
SigmoidKernel The SigmoidKernel class implements a sigmoid kernel based on the hyperbolic tangent.
SumKernel<InputType> The SumKernel class implements a kernel that adds together the result of applying multiple kernels to the same pair of inputs.
VectorFunctionKernel The VectorFunctionKernel implements a kernel that makes use of a vector function plus a kernel that operates on vectors.
WeightedKernel<InputType> The WeightedKernel class implements a kernel that takes another kernel, evaluates it, and then the result is rescaled by a given weight.
ZeroKernel The ZeroKernel always returns zero.
 

Package gov.sandia.cognition.learning.function.kernel Description

Provides kernel functions. Kernel functions are components used in kernel-based learning algorithms. They represent a type of similarity function that obeys certain properties, known as the Mercer conditions. They can be understood as representing an inner product (dot product) for between some vector representation of the two data points.

Since:
2.0
Author:
Justin Basilico