inferpy.inference package

Subpackages

• inferpy.inference.variational package
  • Subpackages
    • inferpy.inference.variational.loss_functions package
      • Submodules
      • inferpy.inference.variational.loss_functions.elbo module
      • Module contents
  • Submodules
  • inferpy.inference.variational.svi module
  • inferpy.inference.variational.vi module
  • Module contents

Submodules

inferpy.inference.inference module

class inferpy.inference.inference.Inference

Bases: object

This class implements the functionality of any Inference class.

compile(pmodel, data_size, extra_loss_tensor=None)

get_interceptable_condition_variables()

posterior(target_names=None, data={})

posterior_predictive(target_names=None, data={})

update(sample_dict)

inferpy.inference.mcmc module

class inferpy.inference.mcmc.MCMC(step_size=0.01, num_leapfrog_steps=5, num_burnin_steps=1000, num_results=500)

Bases: inferpy.inference.inference.Inference

compile(pmodel, data_size, extra_loss_tensor=None)

posterior(target_names=None, data={})

posterior_predictive(target_names=None, data={})

update(data)

Module contents

Any inference class must implement a run method, which receives a sample_dict object, and returns a dict of posterior objects (random distributions, list of samples, etc.)

class inferpy.inference.MCMC(step_size=0.01, num_leapfrog_steps=5, num_burnin_steps=1000, num_results=500)

Bases: inferpy.inference.inference.Inference

compile(pmodel, data_size, extra_loss_tensor=None)

posterior(target_names=None, data={})

posterior_predictive(target_names=None, data={})

update(data)

class inferpy.inference.SVI(*args, batch_size=100, **kwargs)

Bases: inferpy.inference.variational.vi.VI

compile(pmodel, data_size, extra_loss_tensor=None)

create_input_data_tensor(data_loader)

update(data)

class inferpy.inference.VI(qmodel, loss='ELBO', optimizer='AdamOptimizer', epochs=1000)

Bases: inferpy.inference.inference.Inference

compile(pmodel, data_size, extra_loss_tensor=None)

get_interceptable_condition_variables()

property losses

posterior(target_names=None, data={})

posterior_predictive(target_names=None, data={})

update(data)