BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:UW-Madison-Physics-Events
BEGIN:VEVENT
SEQUENCE:1
UID:UW-Physics-Event-5929
DTSTART:20200701T160000Z
DTEND:20200701T170000Z
DTSTAMP:20260415T023621Z
LAST-MODIFIED:20200714T135403Z
LOCATION:Please register for this online event: http://physicsmeetsml.
 org
SUMMARY:Learning for Safety-Critical Control in Dynamical Systems\, Ph
 ysics meets ML\, Yisong Yue\, CalTech
DESCRIPTION:This talk describes ongoing research at Caltech on integra
 ting learning into the design of safety-critical controllers for dynam
 ical systems. To achieve control-theoretic safety guarantees while usi
 ng powerful function classes such as deep neural networks\, we must ca
 refully integrate conventional control principles with learning into u
 nified frameworks. I will present two paradigms: integration in dynami
 cs modeling and integration at the policy/controller design. A special
  emphasis will be placed on methods that both admit relevant safety gu
 arantees and are practical to deploy.
URL:https://www.physics.wisc.edu/events/?id=5929
END:VEVENT
END:VCALENDAR