Module: tfp.substrates.numpy.math.psd_kernels

Positive-semidefinite kernels package.

Classes

class AutoCompositeTensorPsdKernel: Abstract base class for positive semi-definite kernel functions.

class ChangePoint: Changepoint Kernel.

class Constant: Kernel that just outputs positive constant values.

class ExpSinSquared: Exponentiated Sine Squared Kernel.

class ExponentialCurve: Exponential Curve Kernel.

class ExponentiatedQuadratic: The ExponentiatedQuadratic kernel.

class FeatureScaled: Kernel that first rescales all feature dimensions.

class FeatureTransformed: Input transformed kernel.

class GammaExponential: The GammaExponential kernel.

class GeneralizedMatern: Generalized Matern Kernel.

class KumaraswamyTransformed: Transform inputs by Kumaraswamy bijector.

class Linear: Linear Kernel.

class MaternFiveHalves: Matern 5/2 Kernel.

class MaternOneHalf: Matern Kernel with parameter 1/2.

class MaternThreeHalves: Matern Kernel with parameter 3/2.

class Parabolic: The Parabolic kernel.

class PointwiseExponential: Pointwise exponential of a positive semi-definite kernel.

class Polynomial: Polynomial Kernel.

class PositiveSemidefiniteKernel: Abstract base class for positive semi-definite kernel functions.

class RationalQuadratic: RationalQuadratic Kernel.

class SchurComplement: The SchurComplement kernel.

class SpectralMixture: The SpectralMixture kernel.