BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME: CAM Colloquium - Nisha Chandramoorthy\, University of Chicago
  "Learning ergodic dynamics from data"
X-WR-TIMEZONE:Eastern Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260515T051145Z
UID:tag:localist.com\,2008:EventInstance_49462793203054
DTSTART:20250502T194500Z
DTEND:20250502T204500Z
DESCRIPTION:Title: Learning ergodic dynamics from data\n\nAbstract: We are 
 often given time series data from which we would like to learn their gover
 ning laws. How can we learn the dynamical system that reproduces not just 
 the short term but also the long term patterns in the data? In this talk\,
  we are interested in answering this question for chaotic time series gene
 rated by an unknown ergodic dynamical system. Under certain conditions\, w
 e prove that regression of the one-step dynamics is sufficient to emulate 
 ergodic (long-term) behavior provided we also learn the first-order deriva
 tive (Jacobian) of the one-step dynamics. In the second half\, we study th
 e problem of sampling from a target probability distribution in the settin
 g of simulation-based Bayesian inference. Here\, either the target is know
 n up to a normalization constant\, or its score function — gradient of l
 og density — can be evaluated anywhere. We propose to learn the zero of 
 a score residual operator that is the difference between the target score 
 and the score of a pushforward density of a known source distribution thro
 ugh a transport map. The desired transport map is an invertible function t
 hat takes samples from an easy-to-sample source density to produce samples
  according to the target density. We compare such a score operator Newton-
 Raphson method to existing approaches for sampling using measure transport
 . Finally\, we tackle the generative modeling setting in which we discuss 
 how the dynamical systems approach to measure transport can yield new insi
 ghts. The first part of the talk is joint work with Jeongjin Park (Georgia
  Tech) and the second with Youssef Marzouk (MIT) and Adriaan de Clercq (UC
 hicago).\n\nBio: Nisha Chandramoorthy is an Assistant Professor in the Com
 mittee on Computational and Applied Mathematics in the Department of Stati
 stics at the University of Chicago. Recently\, she has been working on und
 erstanding machine learning models and developing efficient algorithms for
  learning and sampling by taking a dynamical systems approach. She receive
 d her PhD from MIT in 2021\, and after that was a postdoctoral researcher 
 at MIT from 2021-2023 and a faculty member at Georgia Tech from 2023-2024.
GEO:42.443451;-76.481506
LOCATION:Frank H. T. Rhodes Hall\, 655
SUMMARY: CAM Colloquium - Nisha Chandramoorthy\, University of Chicago "Lea
 rning ergodic dynamics from data"
URL;VALUE=URI:https://events.cornell.edu/event/cam-colloquium-nisha-chandra
 moorthy-university-of-chicago-learning-ergodic-dynamics-from-data
CATEGORIES:Colloquium
END:VEVENT
END:VCALENDAR