tf.contrib.distributions

 

upper level

 

Modules

 

bijectors:

Bijector Ops.

 

Classes

 

Autoregressive:

Autoregressive distributions.

Bernoulli:

Bernoulli distribution.

BernoulliWithSigmoidProbs:

Bernoulli with probs = nn.sigmoid(logits).

Beta:

Beta distribution.

BetaWithSoftplusConcentration:

Beta with softplus transform of concentration1 and concentration0.

Binomial:

Binomial distribution.

Categorical:

Categorical distribution.

Cauchy:

The Cauchy distribution with location loc and scale scale.

Chi2:

Chi2 distribution.

Chi2WithAbsDf:

Chi2 with parameter transform df = floor(abs(df)).

ConditionalDistribution:

Distribution that supports intrinsic parameters (local latents).

ConditionalTransformedDistribution:

A TransformedDistribution that allows intrinsic conditioning.

Deterministic:

Scalar Deterministic distribution on the real line.

Dirichlet:

Dirichlet distribution.

DirichletMultinomial:

Dirichlet-Multinomial compound distribution.

Distribution:

A generic probability distribution base class.

ExpRelaxedOneHotCategorical:

ExpRelaxedOneHotCategorical distribution with temperature and logits.

Exponential:

Exponential distribution.

ExponentialWithSoftplusRate:

Exponential with softplus transform on rate.

Gamma:

Gamma distribution.

GammaWithSoftplusConcentrationRate:

Gamma with softplus of concentration and rate.

Geometric:

Geometric distribution.

HalfNormal:

The Half Normal distribution with scale scale.

Independent:

Independent distribution from batch of distributions.

InverseGamma:

InverseGamma distribution.

InverseGammaWithSoftplusConcentrationRate:

InverseGamma with softplus of concentration and rate.

Laplace:

The Laplace distribution with location loc and scale parameters.

LaplaceWithSoftplusScale:

Laplace with softplus applied to scale.

Logistic:

The Logistic distribution with location loc and scale parameters.

Mixture:

Mixture distribution.

MixtureSameFamily:

Mixture (same-family) distribution.

Multinomial:

Multinomial distribution.

MultivariateNormalDiag:

The multivariate normal distribution on R^k.

MultivariateNormalDiagPlusLowRank:

The multivariate normal distribution on R^k.

MultivariateNormalDiagWithSoftplusScale:

MultivariateNormalDiag with diag_stddev = softplus(diag_stddev).

MultivariateNormalFullCovariance:

The multivariate normal distribution on R^k.

MultivariateNormalTriL:

The multivariate normal distribution on R^k.

NegativeBinomial:

NegativeBinomial distribution.

Normal:

The Normal distribution with location loc and scale parameters.

NormalWithSoftplusScale:

Normal with softplus applied to scale.

OneHotCategorical:

OneHotCategorical distribution.

Poisson:

Poisson distribution.

PoissonLogNormalQuadratureCompound:

PoissonLogNormalQuadratureCompound distribution.

QuantizedDistribution:

Distribution representing the quantization Y = ceiling(X).

RegisterKL:

Decorator to register a KL divergence implementation function.

RelaxedBernoulli:

RelaxedBernoulli distribution with temperature and logits parameters.

RelaxedOneHotCategorical:

RelaxedOneHotCategorical distribution with temperature and logits.

ReparameterizationType:

Instances of this class represent how sampling is reparameterized.

SinhArcsinh:

The SinhArcsinh transformation of a distribution on (-inf, inf).

StudentT:

Student’s t-distribution.

StudentTWithAbsDfSoftplusScale:

StudentT with df = floor(abs(df)) and scale = softplus(scale).

TransformedDistribution:

A Transformed Distribution.

Uniform:

Uniform distribution with low and high parameters.

VectorDeterministic:

Vector Deterministic distribution on R^k.

VectorDiffeomixture:

VectorDiffeomixture distribution.

VectorExponentialDiag:

The vectorization of the Exponential distribution on R^k.

VectorLaplaceDiag:

The vectorization of the Laplace distribution on R^k.

VectorSinhArcsinhDiag:

The (diagonal) SinhArcsinh transformation of a distribution on R^k.

WishartCholesky:

The matrix Wishart distribution on positive definite matrices.

WishartFull:

The matrix Wishart distribution on positive definite matrices.

 

Functions

 

assign_log_moving_mean_exp(…):

Compute the log of the exponentially weighted moving mean of the exp.

assign_moving_mean_variance(…):

Compute exponentially weighted moving {mean,variance} of a streaming value.

auto_correlation(…):

Auto correlation along one axis.

estimator_head_distribution_regression(…):

Creates a Head for regression under a generic distribution.

fill_triangular(…):

Creates a (batch of) triangular matrix from a vector of inputs.

kl_divergence(…):

Get the KL-divergence KL(distribution_a || distribution_b).

matrix_diag_transform(…):

Transform diagonal of [batch-]matrix, leave rest of matrix unchanged.

moving_mean_variance(…):

Compute exponentially weighted moving {mean,variance} of a streaming value.

normal_conjugates_known_scale_posterior(…):

Posterior Normal distribution with conjugate prior on the mean.

normal_conjugates_known_scale_predictive(…):

Posterior predictive Normal distribution w. conjugate prior on the mean.

percentile(…):

Compute the q-th percentile of x.

quadrature_scheme_lognormal_gauss_hermite(…):

Use Gauss-Hermite quadrature to form quadrature on positive-reals.

quadrature_scheme_lognormal_quantiles(…):

Use LogNormal quantiles to form quadrature on positive-reals.

quadrature_scheme_softmaxnormal_gauss_hermite(…):

Use Gauss-Hermite quadrature to form quadrature on K – 1 simplex.

quadrature_scheme_softmaxnormal_quantiles(…):

Use SoftmaxNormal quantiles to form quadrature on K – 1 simplex.

reduce_weighted_logsumexp(…):

Computes log(abs(sum(weight * exp(elements across tensor dimensions)))).

softplus_inverse(…):

Computes the inverse softplus, i.e., x = softplus_inverse(softplus(x)).

tridiag(…):

Creates a matrix with values set above, below, and on the diagonal.