BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Math. of Deep Learning Seminar: Marylou Gabrie
DTSTART;VALUE=DATE-TIME:20211207T160000Z
DTEND;VALUE=DATE-TIME:20211207T170000Z
DTSTAMP;VALUE=DATE-TIME:20230127T114500Z
UID:indico-event-2738@indico.flatironinstitute.org
DESCRIPTION:Title: Assisting Sampling with Learning: Adaptive Monte Carlo
with Normalizing Flows\n\nAbstract: In many applications in computationa
l sciences and statistical inference\, one seeks to compute expectations o
n complex high-dimensional distributions. These problems are often plagued
by multi-modality/metastability\; slow relaxation between unconnected mod
es leads to slow convergence of estimators of such expectations. In this t
alk\, I will present a strategy to enhance sampling with deep generative m
odels called Normalizing Flows. We will see when simultaneously sampling a
nd training can incur a drastic acceleration of MCMC convergence and discu
ss current limits in learning for sampling.\n\nThis is joint work with Gra
nt Rotskoff (Stanford) and Eric Vanden-Eijnden (Courant Institute\, NYU).\
n\n \n\n \n\nIf you would like to attend\, please email crampersad@flati
roninstitute.org for the Zoom link.\n\nhttps://indico.flatironinstitute.o
rg/event/2738/
URL:https://indico.flatironinstitute.org/event/2738/
END:VEVENT
END:VCALENDAR