Fall 2021

Generative Models for Sampling on Supports Inspired by Observed Data

Thursday, Sep. 30, 2021 4:00 pm4:40 pm PDT

Add to Calendar


Susan Holmes (Stanford)


Calvin Lab Auditorium and Zoom

Statisticians often use Markov chains to generate points that have a similar support to observed data. This poses several challenges: - the data themselves are discrete and are dense on the underlying manifolds and complex structures. - the data are not uniformly distributed on the manifolds. I will discuss these issues and some solutions provided that can be useful in applied problems.