BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Eventi DIAG
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20221030T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20220327T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RDATE:20230326T020000
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.24920.field_data.0@www.ugov-ricerca.uniroma1.it
DTSTAMP:20260407T183757Z
CREATED:20220914T164456Z
DESCRIPTION:AbstractReinforcement learning (RL) is a powerful machine learn
 ing paradigm that studies the interaction between a single agent with an u
 nknown environment. The way this interaction works is as follows: at any g
 iven time the agent finds itself in a given state and has a number of acti
 ons to choose from\, it then chooses an action and as a consequence the en
 vironment provides a reward and a new state to which the agent transitions
  to. This interaction may go on forever or for a limited amount of time. T
 he goal of the agent is to learn a policy (a policy is a function that det
 ermines what actions are chosen in each of the possible states) so as to m
 aximize the long term cumulative rewards. A plethora of applications fit i
 nto the RL framework\, however\, in many cases of interest\, a team of age
 nts will need to interact with the environment and with each other to achi
 eve a common goal. This is the object study of collaborative multi-agent R
 L (MARL). Addressing the problem of collaborative MARL introduces many cha
 llenges\, one of which is that the number of possible actions that the tea
 m can choose from grows exponentially with the number of agents. Due to th
 is exponential growth of the joint action set\, learning a 'joint team pol
 icy' using conventional single-agent RL algorithms becomes unfeasible. The
 refore\, it is necessary to rely on learning factored policies instead. In
  this talk\, I will further clarify the challenge of learning factored pol
 icies in cooperative MARL\, explain why it is an important problem to stud
 y and I will introduce the Logical Team Q-learning algorithm\, which is on
 e possible solution to this problem. Zoom video available at: https://unir
 oma1.zoom.us/j/83592522402?pwd=Yzhuak4zQnYvNGthQlZCeGdjWTBkUT09 Short Bio:
 Lucas Cassano received his Electronics Engineer degree from Buenos Aires I
 nstitute of Technology in 2013 and then joined Satellogic as a full-time e
 ngineer developing and implementing star tracker algorithms for micro-sate
 llites. Afterwards he went to UCLA where he received  his M.S. and Ph.D. d
 egrees in 2015 and 2020\, respectively. After obtaining his PhD he joined 
 EPFL a postdoctoral researcher. The main focus of his academic research ha
 s been cooperative multi-agent reinforcement learning. Currently\, he hold
 s an Applied Scientist position at Amazon where he works on ranking.
DTSTART;TZID=Europe/Paris:20220921T140000
DTEND;TZID=Europe/Paris:20220921T140000
LAST-MODIFIED:20220914T180022Z
LOCATION:Aula Magna
SUMMARY:Logical Team Q-learning: An approach towards factored policies in c
 ooperative MARL - Lucas Cassano - Amazon
URL;TYPE=URI:http://www.ugov-ricerca.uniroma1.it/node/24920
END:VEVENT
END:VCALENDAR