BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260611T020806EDT-1358usP8Zl@132.216.98.100
DTSTAMP:20260611T060806Z
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\nZoom Link\n	Meeting ID: 910 7928 6959\n\nPasscode: VISS\n	\n	Spe
 aker: Rabih Salhab\, Postdoctoral Associate\, Institute for Data\, Systems
 \, and Society (IDSS)\, MIT\n	\n	Abstract:\n\nI will present two recent work
 s on social learning under behavioral assumptions. The first is Social Lea
 rning with Sparse Belief Samples. In this work\, we introduce a non-Bayesi
 an model of learning over a social network where a group of agents with in
 sufficient and heterogeneous sources of information share their experience
 s to learn an underlying state of the world. Inspired by a recent body of 
 research in cognitive science on human decision making\, we presume two be
 havioral assumptions. Motivated by the coarseness of communication\, our f
 irst assumption posits that agents only share samples taken from their bel
 ief distribution over the set of states\, to which we refer as their actio
 ns. This situation is to be contrasted with that of sharing the full belie
 f\, i.e. probability distribution over the entire set of states. The secon
 d assumption is limited cognitive power\, based on which individuals incor
 porate their neighbors' actions into their beliefs following a simple DeGr
 oot-like social learning rule which suffers from redundancy neglect and im
 perfect recall of the past history. We show that so long as all the indivi
 duals trust their neighbors' actions more than their private signals\, the
 y may end up mislearning the state with positive probability. Learning\, o
 n the other hand\, requires that the population includes a group of self-c
 onfident experts in different states. This means that for each state\, the
 re is an agent whose signaling function for her state of expertise is dist
 inguishable from the convex hull of the remaining signaling functions\, an
 d that her private signals sufficiently weigh in her social learning rule.
  This is a joint work with Amir Ajorlou and Ali Jadbabaie.\n	\n	The second w
 ork is Social Learning with Unreliable Agents and Self-reinforcing Stochas
 tic Dynamics. We consider a group of agents that have fixed unobservable b
 inary ``beliefs''. An individual's belief models for example their politic
 al support (Democrat or Republican). At each time period\, agents broadcas
 t binary opinions on a social network. We assume that individuals may lie 
 and declare opinions different from their true beliefs to conform with the
 ir neighbors. This raises the natural question as to whether one can estim
 ate the agents' true beliefs from observations of declared opinions. We an
 alyze this question in the special case of complete graph. We show that\, 
 as long as the population does not include large majorities\, estimation o
 f aggregate true belief and individual true beliefs is possible. On the ot
 her hand\, large majorities force minorities to lie as time goes to infini
 ty\, which makes asymptotic estimation impossible.\n\n \n\nThis is a joint
  work with Anuran Makur\, Ali Jadbabaie\, and Elchanan Mossel.\n	\n	Biograph
 y:\n\nI'm currently a Postdoctoral Associate at the MIT Institute for Data
 \, Systems\, and Society (IDSS) hosted by Prof. Ali Jadbabaie. From 2018 t
 o 2019\, I was a Postdoctoral Fellow (IVADO) at HEC Montreal hosted by Pro
 f. Georges Zaccour. I finished my Ph.D. degree in Electrical Engineering i
 n the Department of Electrical Engineering\, Ecole Polytechnique de Montre
 al\, Canada\, under the supervision of Prof. Roland Malhamé and Jerome Le 
 Ny in April 2018. I received the B.S. degree in Electrical Engineering fro
 m Ecole Superieure d'Ingenieurs de Beyrouth (E.S.I.B)\, Lebanon\, in 2008.
  From 2008 to 2013\, I was an Electrical Engineer with Dar al Handasah Sha
 ir and Partners\, Lebanon.\n
DTSTART:20210115T190000Z
DTEND:20210115T190000Z
LOCATION:CA\, ZOOM
SUMMARY:Social Learning under Behavioral Assumptions
URL:https://www.mcgill.ca/cim/channels/event/social-learning-under-behavior
 al-assumptions-327423
END:VEVENT
END:VCALENDAR