BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:A new metric on kernel matrices with applications to matrix means 
 - Suvrit Sra\, Max-Planck Institute for Biological Cybernetics in Tübinge
 n (Germany)
DTSTART:20120927T090000Z
DTEND:20120927T094500Z
UID:TALK40050@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:I will talk about some recent but fundamental work related to 
 distance metrics on the manifold of kernel matrices\, including a bit abou
 t the original application in nearest neighbor search that motivated our w
 ork.\n\nSymmetric positive definite (spd) matrices are remarkably pervasiv
 e\, especially in machine learning\, statistics\, and optimization. We con
 sider the fundamental task of measuring distances between two spd matrices
 \; a task that is  nontrivial whenever an application uses distance functi
 ons that must respect the non-Euclidean geometry of spd matrices. Unfortun
 ately\, typical non-Euclidean distance measures such as the Riemannian met
 ric are computationally demanding and also complicated to use. To ameliora
 te these difficulties\, we introduce a new metric on spd matrices: this me
 tric not only respects non-Euclidean geometry\, it also offers faster comp
 utation than the Riemannian metric while being less complicated to use. We
  support our claims theoretically via a series of theorems that relate our
  metric to the Riemannian metric\, and experimentally by studying the prob
 lem of computing matrix geometric means. Amazingly\, though nonconvex\, we
  show it to be efficiently solvable to global optimality.\n\nBrief Bio: \n
 \nSuvrit Sra is a Senior Research Scientist at the Max Planck Institute fo
 r Intelligent Systems (formerly Biological Cybernetics) in Tübingen\, Ger
 many.  He obtained his M.S. and Ph.D. in Computer Science from the Univers
 ity of Texas at Austin in 2007\, and a B.E. (Hons.) in Computer Science fr
 om BITS\, Pilani (India) in 1999.   His primary research focus is on large
 -scale optimization (both convex and nonconvex) with application to proble
 ms in machine learning\, statistics\, and scientific computing. He has a s
 trong interest in several flavors of analysis\, most notably matrix analys
 is. \n\nHis research has won awards at leading international conferences o
 n machine learning and data mining (ICML\, ECML\, SIAM DM). His work on "t
 he Metric Nearness Problem" was selected to receive the SIAM Outstanding P
 aper Prize (2011) (awarded triennially). He regularly organizes the Neural
  Information Processing Systems (NIPS) workshops on "Optimization for Mach
 ine Learning\," and has recently co-edited a book with the same title (MIT
  Press\, 2011).
LOCATION:Small lecture theatre\, Microsoft Research Ltd\, 7 J J Thomson Av
 enue (Off Madingley Road)\, Cambridge
END:VEVENT
END:VCALENDAR