BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:www.tcs.tifr.res.in/event/1714
DTSTAMP:20260511T053246Z
SUMMARY:Rare Events in Stochastic Multi-Armed Bandits
DESCRIPTION:Speaker: Anirban Bhattacharjee (TIFR)\n\nAbstract: \nWe examine
  stochastic multi-armed bandit (MAB) problems in rare event regimes with e
 mphasis on Best Arm Identification (BAI) and also touch upon regret minimi
 zation. When arm rewards occur infrequently but are high in magnitude\, we
  develop algorithms for BAI which are based on Poisson approximation and d
 rastically reduce computational effort at the cost of negligible increase 
 in sample complexity. For identifying the safest system among a given set 
 of safety-critical systems with rare failures\, we consider simulation mod
 els of the systems\, and simulate from them following BAI methods. Reasona
 bly accurate approximations of the lower bound on sample complexity reveal
  that sample complexity depends on the failure rate of the second-best sys
 tem\, as opposed to the best. Further\, standard regret minimization algor
 ithms are shown to perform poorly in rare-event regimes where rewards are 
 high-value but rarely seen\, necessitating scaled modifications that ensur
 e optimality.\n
URL:https://www.tcs.tifr.res.in/web/events/1714
DTSTART;TZID=Asia/Kolkata:20260514T160000
DTEND;TZID=Asia/Kolkata:20260514T170000
LOCATION:A-201 and Zoom
END:VEVENT
END:VCALENDAR