Normalizing Flows and Diffusion Models for Images and Text

‍

Presenter:  Didrik Nielsen from DTU Compute

Abstract: Normalizing flows and diffusion models are two classes of deep probabilistic generative models that excel at modeling high-dimensional distributions. In recent years, a lot of progress has been made for these models, with the majority of work focusing on images.

In this talk, Nielsen will provide an overview of these model classes and their application to images. Additionally, he will briefly discuss our upcoming NeurIPS 2021 paper [1] where they extend these models to be applicable to categorical data such as text.

[1] Emiel Hoogeboom, Didrik Nielsen, Priyank Jaini, Patrick Forré, Max Welling Argmax Flows and Multinomial Diffusion: Learning Categorical Distributions https://arxiv.org/abs/2102.05379

‍