BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Colloquium: Physical Agents that Learn from Experience
DTSTART:20260319T101500
DTEND:20260319T111500
DTSTAMP:20260415T075814Z
UID:62ccb606c22d9de3690269d78ef07b6d8434e0a8e6adb2c2aa015681
CATEGORIES:Conferences - Seminars
DESCRIPTION:Par : Andrew Wagenmaker - UC Berkeley\nIC Faculty candidate\n\
 nAbstract\nWhile the last decade has witnessed tremendous progress in the 
 development of AI\, this progress has been concentrated largely in “virt
 ual” realms\, and successful applications of AI to the physical world re
 main limited. Closing this gap and enabling AI in the physical world prese
 nts a fundamental challenge: the source of large-scale data that enabled A
 I progress in the virtual world—the internet—simply does not exist for
  the physical world. \n\nIn this talk\, I will investigate how we can ove
 rcome this lack of large-scale data by enabling AI agents to learn from ex
 perience as they interact with the physical world. In particular\, I will 
 consider how we can collect the experience—explore—that allows for lea
 rning and improvement\, and how the limited sources of data that are often
  available to us in the physical world—human demonstrations and dynamics
  models—can enable this. Focusing on robotic control\, I will first show
  how generative robot policies trained on human demonstrations can be util
 ized to achieve highly focused exploration and enable fast online improvem
 ent\, and how we can pretrain generative policies on human demonstrations 
 that can themselves collect the experience necessary to learn and improve.
  I will then consider how dynamics models\, even coarse models that are in
 sufficient for obtaining effective task-solving policies\, can enable effi
 cient exploration\, and how the resulting exploration allows for learning 
 performant task-solving robotic behaviors. Across these examples\, I will 
 argue that the insights gained through rigorous analysis are key to uncove
 ring the algorithmic approaches that enable learning from experience\, and
  ultimately bringing AI to the physical world.\n\nBio\nAndrew Wagenmaker i
 s a postdoctoral scholar in Electrical Engineering and Computer Sciences a
 t UC Berkeley working with Sergey Levine. Previously\, he completed a PhD 
 in Computer Science at the University of Washington\, where he was advised
  by Kevin Jamieson. Andrew’s research focuses on learning in dynamic\, i
 nteractive settings\, spanning fundamental algorithm development to practi
 cal approaches for real-world learning and decision-making\, particularly 
 toward enabling efficient learning in robotic systems. His work has been r
 ecognized by a Best Paper nomination at the Conference on Robot Learning\,
  and he is a recipient of the NSF Graduate Research Fellowship.\n\nMore in
 formation\n 
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420 https://epfl.zoom.us/
 j/63384377896
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR