View source on GitHub |
Get the KL-divergence KL(distribution_a || distribution_b).
tfp.substrates.numpy.distributions.kl_divergence(
distribution_a, distribution_b, allow_nan_stats=True, name=None
)
If there is no KL method registered specifically for type(distribution_a)
and type(distribution_b)
, then the class hierarchies of these types are
searched.
If one KL method is registered between any pairs of classes in these two parent hierarchies, it is used.
If more than one such registered method exists, the method whose registered classes have the shortest sum MRO paths to the input types is used.
If more than one such shortest path exists, the first method
identified in the search is used (favoring a shorter MRO distance to
type(distribution_a)
).
Returns | |
---|---|
A Tensor with the batchwise KL-divergence between distribution_a
and distribution_b .
|
Raises | |
---|---|
NotImplementedError
|
If no KL method is defined for distribution types
of distribution_a and distribution_b .
|
Built-in KL(distribution_a || distribution_b) registrations:
distribution_a || distribution_b
====================================================================================
Bernoulli || Bernoulli
Beta || Beta
Categorical || Categorical
Cauchy || Cauchy
Chi || Chi
Chi2 || Chi2
Chi2 || Gamma
+ Exponential
ContinuousBernoulli || ContinuousBernoulli
Dirichlet || Dirichlet
ExpGamma || ExpGamma
ExpInverseGamma || ExpInverseGamma
Gamma || Chi2
Exponential +
Gamma || Gamma
Exponential + + Exponential
GaussianProcess || MultivariateNormalLinearOperator
GaussianProcessRegressionModel + + MultivariateNormalDiag
VariationalGaussianProcess + + MultivariateNormalDiagPlusLowRank
+ MultivariateNormalFullCovariance
+ MultivariateNormalTriL
Gumbel || Gumbel
HalfNormal || HalfNormal
Laplace || Laplace
LogNormal || LogNormal
LogitNormal || LogitNormal
MatrixNormalLinearOperator || MatrixNormalLinearOperator
Moyal || Moyal
MultivariateNormalLinearOperator || GaussianProcess
MultivariateNormalDiag + + GaussianProcessRegressionModel
MultivariateNormalDiagPlusLowRank + + VariationalGaussianProcess
MultivariateNormalFullCovariance +
MultivariateNormalTriL +
MultivariateNormalLinearOperator || MultivariateNormalLinearOperator
MultivariateNormalDiag + + MultivariateNormalDiag
MultivariateNormalDiagPlusLowRank + + MultivariateNormalDiagPlusLowRank
MultivariateNormalFullCovariance + + MultivariateNormalFullCovariance
MultivariateNormalTriL + + MultivariateNormalTriL
Normal || Normal
OneHotCategorical || OneHotCategorical
OrderedLogistic || OrderedLogistic
Pareto || Pareto
PowerSpherical || SphericalUniform
PowerSpherical || VonMisesFisher
ProbitBernoulli || ProbitBernoulli
StoppingRatioLogistic || StoppingRatioLogistic
Uniform || Uniform
VonMises || VonMises
VonMisesFisher || SphericalUniform
Weibull || Gamma
+ Exponential
Weibull || Weibull
_BaseDeterministic || Distribution
Deterministic + + AutoCompositeTensorDistribution
VectorDeterministic + + Autoregressive
+ BatchBroadcast
+ BatchConcat
+ BatchReshape
+ Bates
+ 116 more
_Blockwise || _Blockwise
Blockwise + + Blockwise
_Independent || _Independent
Independent + + Independent
_JointDistributionSequential || _JointDistributionSequential
JointDistributionNamed + + JointDistributionNamed
JointDistributionNamedAutoBatched + + JointDistributionNamedAutoBatched
JointDistributionSequential + + JointDistributionSequential
JointDistributionSequentialAutoBatched + + JointDistributionSequentialAutoBatched
_Masked || _Masked
Masked + + Masked
_NonCompositeTensorCast || _NonCompositeTensorCast
_Sample || _Sample
Sample + + Sample
_TransformedDistribution || _TransformedDistribution
Chi + + Chi
ExpInverseGamma + + ExpInverseGamma
GeneralizedExtremeValue + + GeneralizedExtremeValue
Gumbel + + Gumbel
JohnsonSU + + JohnsonSU
Kumaraswamy + + Kumaraswamy
15 more + + 15 more