probly.transformation.bayesian

probly.transformation.bayesian(base, use_base_weights=False, posterior_std=0.05, prior_mean=0.0, prior_std=1.0)[source]

Create a Bayesian predictor from a base predictor based on [BCKW15].

Parameters:

  • base (T) – The base model to be used for the Bayesian neural network.

  • use_base_weights (bool) – bool, If True, the weights of the base model are used as the prior mean.

  • posterior_std (float) – float, The initial posterior standard deviation.

  • prior_mean (float) – float, The prior mean.

  • prior_std (float) – float, The prior standard deviation.

Returns:

The Bayesian predictor.

Return type:

T