Stanford CS236: Deep Generative Models I 2023 I Lecture 8 - Normalizing Flows

Transcript

The plan for today is to continue talking about normalizing flow models. So recall that in the last lecture, we've introduced this idea of building a latent variable model that will allow us to evaluate likelihoods exactly, so without having to rely on variational inference. And so it's going to be similar to a variational autoencoder, in the sense... Read More