# Bayes theorem

The "marginalization" part is also sometimes called the "evidence".

In machine learning, the marginal log-likelihood is the likelihood of the data after integrating over the latent space.