BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Robust Statistics\, Revisited
DTSTART:20170615T113000
DTEND:20170615T121500
DTSTAMP:20260406T170449Z
UID:b5037175be16f533adaf1e883234a7b769d4d8c8462cb2964b81ac00
CATEGORIES:Conferences - Seminars
DESCRIPTION:Ankur Moitra\, MIT\nStarting from the seminal works of Tukey (
 1960) and Huber (1964)\, the field of robust statistics asks: Are there es
 timators that provably work in the presence of noise? The trouble is that 
 all known provably robust estimators are also hard to compute in high-dime
 nsions.\nHere\, we study a basic problem in robust statistics\, posed in v
 arious forms in the above works. Given corrupted samples from a high-dimen
 sional Gaussian\, are there efficient algorithms to accurately estimate it
 s parameters? We give the first algorithms that are able to tolerate a con
 stant fraction of corruptions that is independent of the dimension. Additi
 onally\, we give several more applications of our techniques to product di
 stributions and various mixture models.\nThis is based on joint work with 
 Ilias Diakonikolas\, Jerry Li\, Gautam Kamath\, Daniel Kane and Alistair S
 tewart.
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR