BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Representation-based Reinforcement Learning and Control for Dynami
 cal Systems - Na Li (Harvard University)
DTSTART:20251110T140000Z
DTEND:20251110T144000Z
UID:TALK238429@talks.cam.ac.uk
DESCRIPTION:The explosive growth of machine learning and data-driven metho
 dologies have revolutionized numerous fields. Yet\, the translation of the
 se successes to the domain of dynamical physical systems remains a signifi
 cant challenge. Closing the loop from data to actions in these systems fac
 es many difficulties\, stemming from the need for sample efficiency and co
 mputational feasibility\, along with many other requirement such as verifi
 ability\, robustness\, and safety. In this talk\, we present a framework t
 hat bridges this gap by introducing novel representations for developing n
 onlinear stochastic control and reinforcement learning algorithms. Our app
 roach enables efficient\, safe\, robust\, and scalable decisionmaking with
  provable guarantees. We further demonstrate how these representations hel
 p close the simto-real gap\, enhance data efficiency in imitation learning
 \, and enable scalable computation of localized policies for large-scale n
 onlinear networked systems. Lastly\, we will briefly present our latest wo
 rk on using diffusion models to represent control policies and how to onli
 ne train diffusion policies\, along with their applications to manipulatio
 n tasks.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR