BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Causal Bandits
DTSTART:20221101T103000
DTEND:20221101T123000
DTSTAMP:20260407T120324Z
UID:7cbc885141c4c2f05b6bd1e36d2a01889e94c91cd4a0a8d59e081a15
CATEGORIES:Conferences - Seminars
DESCRIPTION:Mikhail Konobeev\nEDIC candidacy exam\nExam president: Prof. M
 atthias Grossglauser\nThesis advisor: Prof. Negar Kiyavash\nCo-examiner: P
 rof. Michael Kapralov\n\nAbstract\nWe consider the problem of maximizing a
  designated reward variable in probabilistic causal model via intervention
 s on other variables. Our goal is to eventually develop methods that do no
 t require the knowledge of the underlying causal graph. We present three p
 apers that are the benchmarks for our future work: one that defines the ca
 usal bandit problem\, another that solves it using only the knowledge of t
 he essential graph\, and the last that characterizes the set of actions an
  agent has to explore. We present a randomized algorithm for finding the p
 arent of the reward node and distinguish three classes of graphs with the 
 asymptotic number of interventions of the proposed algorithm ranging from 
 logarithmic to sublinear to linear. We also discuss different other settin
 gs in which we would like to achieve our goal.\n\nBackground papers\n1. C
 ausal bandits: Learning good interventions via causal inference\n2. Causa
 l bandits with unknown graph structure\n3. Structural causal bandits: w
 here to intervene?\n 
LOCATION:BC 133 https://plan.epfl.ch/?room==BC%20133
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR