BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Heteroskedastic PCA: Algorithm\, Optimality\, and Applications - T
 ony Cai\, University of Pennsylvania
DTSTART:20181011T150000Z
DTEND:20181011T160000Z
UID:TALK109642@talks.cam.ac.uk
CONTACT:Dr Sergio Bacallado
DESCRIPTION:Principal component analysis (PCA) is a ubiquitous method in s
 tatistics\, machine learning and applied mathematics. PCA has been well st
 udied and used mostly in the homoskedastic noise case. \n\nIn this talk\, 
 we consider PCA in the setting where the noise is heteroskedastic\, which 
 arises naturally from a range of applications. We proposed an algorithm ca
 lled DIALECT for heteroskedastic PCA and establish its optimality. A key t
 echnical step is a deterministic robust perturbation analysis\, which can 
 be of independent interest. We will also discuss some applications in the 
 analysis of high-dimensional data\, including heteroskedastic matrix SVD\,
  community detection in bipartite stochastic block model\, and noisy matri
 x completion.
LOCATION:MR13
END:VEVENT
END:VCALENDAR