I n part 3 of this book, you learn about Bayesian DL models. You’ll see that Bayesian models become especially important when you encounter novel situations. Bayesian models are a special form of probabilistic models that add additional uncertainty.
In part 2 of this book, you learned how to set up non-Bayesian probabilistic NN models. These probabilistic models allowed you to describe the uncertainty inherent in data. You always need to deal with the inherent uncertainty in data if there’s some randomness, meaning the observed outcome can’t be determined completely by the input. This uncertainty is called aleatoric uncertainty.