BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Adaptive Importance Sampling for accelerating the minimization of 
 tail risks - Karthyek Rajhaa Annaswamy Murthy (Singapore University of Tec
 hnology and Design)
DTSTART:20240423T084500Z
DTEND:20240423T093000Z
UID:TALK214114@talks.cam.ac.uk
DESCRIPTION:The ability to estimate and control extreme tail risks\, besid
 es being an integral part of quantitative risk&nbsp\;management\, is centr
 al to running operations requiring high service levels and cyber-physical 
 systems&nbsp\;with high-reliability specifications. &nbsp\;Despite this si
 gnificance\, scalable algorithmic approaches have&nbsp\;remained elusive: 
 This is due to the rarity with which relevant risky samples get observed\,
  and the&nbsp\;critical role experts need to play in devising variance red
 uction techniques based on instance-specific large&nbsp\;deviations. Our g
 oal in this talk is to describe a variance reducing importance sampling te
 chnique which can be flexibly combined with stochastic approximation to yi
 eld significantly faster algorithms for minimizing tail risks.&nbsp\;We ai
 m to bring out &nbsp\;how the central challenge of selecting a good import
 ance sampler\, which in turn depends on the knowledge of the solution we a
 re seeking\, can be tackled with a novel black-box importance sampler.
LOCATION:External
END:VEVENT
END:VCALENDAR