BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Penn Engineering Events - ECPv6.16.3//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Penn Engineering Events
X-ORIGINAL-URL:https://seasevents.nmsdev7.com
X-WR-CALDESC:Events for Penn Engineering Events
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20240310T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20241103T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20250309T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20251102T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20260308T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20261101T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20250219T150000
DTEND;TZID=America/New_York:20250219T160000
DTSTAMP:20260602T160241
CREATED:20250211T152219Z
LAST-MODIFIED:20250211T152219Z
UID:13298-1739977200-1739980800@seasevents.nmsdev7.com
SUMMARY:Spring 2025 GRASP SFI: Qinghua Liu\, Microsoft Research\, “When Is Partially Observable Reinforcement Learning Not Scary?”
DESCRIPTION:This will be a hybrid event with in-person attendance in Levine 307 and virtual attendance on Zoom. \nABSTRACT\nPartial observability is ubiquitous in Reinforcement Learning (RL) applications\, where agents must make sequential decisions despite lacking complete information about the latent states of the controlled system. Partially observable RL is notoriously challenging in theory—well-known information-theoretic results show that learning partially observable Markov decision processes (POMDPs) requires an exponential number of samples in the worst case. However\, this does not rule out the existence of interesting subclasses of POMDPs that encompass a diverse set of practical applications while remaining tractable. \nIn this talk\, we identify a rich family of tractable POMDPs\, which we call weakly revealing POMDPs. This family excludes pathological cases where observations provide so little information that learning becomes infeasible. We prove that for weakly revealing POMDPs\, a simple algorithm combining optimism and Maximum Likelihood Estimation (MLE) is sufficient to guarantee polynomial sample complexity. Finally\, we discuss the practical implications of this theory\, including strategies for collecting samples in partially observable tasks and the limitations of purely model-free algorithms.
URL:https://seasevents.nmsdev7.com/event/spring-2025-grasp-sfi-qinghua-liu/
LOCATION:Levine 307\, 3330 Walnut Street\, Philadelphia\, PA\, 19104\, United States
CATEGORIES:Seminar
ORGANIZER;CN="General Robotics%2C Automation%2C Sensing and Perception (GRASP) Lab":MAILTO:grasplab@seas.upenn.edu
END:VEVENT
END:VCALENDAR