stefanradev93 / BayesFlow

A Python library for amortized Bayesian workflows using generative neural networks.
https://bayesflow.org/
MIT License
286 stars 45 forks source link

context-aware bootstrap sampling for point estimation #151

Closed han-ol closed 4 months ago

han-ol commented 4 months ago

The AmortizedPointEstimatior now has a method for uncertainty quantification of point estimators with parametric bootstrap.

bootstrap_sample(forward_dict, n_bootstrap, simulator, configurator, to_numpy=True, **kwargs) returns estimator samples of the bootstrap distribution. Because simulator and configurator in general depend on context variables, the full output forward_dict of the generative model needs to be passed.

han-ol commented 4 months ago

Related to https://github.com/stefanradev93/BayesFlow/issues/121