Abstract: I will overview two approaches to parametrize and generate new samples from highly nonGaussian distributions: normalizing flows and variational autoencoders. Time perimitting, I will briefly discuss some applications to the early Universe and field theory.
Generative NonGaussianity: Normalizing Flows and Variational Autoencoders
Jonathan Braden (CITA) // February 15, 2023
Abstract: I will overview two approaches to parametrize and generate new samples from highly nonGaussian distributions: normalizing flows and variational autoencoders. Time perimitting, I will briefly discuss some applications to the early Universe and field theory.