BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260910T131620EDT-5268U8Gd2O@132.216.98.100
DTSTAMP:20260910T171620Z
DESCRIPTION:Virtual Informal Systems Seminar (VISS) Centre for Intelligent 
 Machines (CIM) and Groupe d'Etudes et de Recherche en Analyse des Decision
 s (GERAD)\n\nDileep Kalathil\n\nAbstract:\n	Reinforcement Learning (RL) is 
 the class of machine learning that addresses the problem of learning to co
 ntrol unknown dynamical systems. RL has achieved remarkable success recent
 ly in applications like playing games and robotics. However\, most of thes
 e successes are limited to very structured or simulated environments. When
  applied to real-world systems\, RL algorithms face two fundamental source
 s of fragility. Firstly\, the real-world system parameters can be very dif
 ferent from that of the nominal values used for training RL algorithms. Se
 condly\, the control policy for any real-world system is required to maint
 ain some necessary safety criteria to avoid undesirable outcomes. Most dee
 p RL algorithms overlook these fundamental challenges which often results 
 in learned polices that can performs poorly in the real-world setting. We 
 address these issues in two steps. First\, we propose a robust reinforceme
 nt learning algorithm to train policies that account for the possible para
 meter mismatches between the simulation system and real-world system. Seco
 nd\, we develop a safe reinforcement learning algorithm to learn policies 
 such that the frequency of visiting undesirable states and expensive actio
 ns satisfies the safety constraints.\n	\n	Bio:\n	Dileep Kalathil is an Assist
 ant Professor in the Department of Electrical and Computer Engineering at 
 Texas A&M University. His main research area is reinforcement learning\, w
 ith applications in cyber-physical systems\, intelligent transportation sy
 stems and power systems. In particular\, his research addresses three fund
 amental problems in RL: (i) How to develop data efficient RL algorithms? (
 ii) How to develop safe and robust RL algorithms? and (iii) How to develop
  scalable multi-agent RL algorithms? Before joining TAMU\, he was a postdo
 ctoral researcher in the EECS department at UC Berkeley. He received his P
 hD from University of Southern California (USC) in 2014 where he won the b
 est PhD Dissertation Prize in the Department of Electrical Engineering. He
  received an M. Tech. from IIT Madras where he won the award for the best 
 academic performance in the Electrical Engineering Department.\n
DTSTART:20201030T153000Z
DTEND:20201030T163000Z
LOCATION:CA\, ZOOM
SUMMARY:Reinforcement Learning with Robustness and Safety Guarantees
URL:https://www.mcgill.ca/cim/channels/event/reinforcement-learning-robustn
 ess-and-safety-guarantees-325752
END:VEVENT
END:VCALENDAR