BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Classification on the Grassmann Manifold: Performance Limits of Co
 mpressive Classifiers - Dr. Miguel Rodrigues\, University College London
DTSTART:20140227T140000Z
DTEND:20140227T150000Z
UID:TALK50511@talks.cam.ac.uk
CONTACT:Prof. Ramji Venkataramanan
DESCRIPTION:The reliable classification of high-dimensional signals from l
 ow-dimensional measurements is an increasingly crucial task in the age of 
 the data deluge. This talk demonstrates that tools and intuitions from Sha
 nnon theory enable the derivation of fundamental limits on the performance
  of such classification systems\, by focusing on the classification of hig
 h-dimensional (rank-deficient) Gaussian signals from noisy\, low-dimension
 al signal projections.\n\nLeveraging the syntactic equivalence of discrimi
 nation between Gaussian classes and communication over vector wireless cha
 nnels\, bounds on classifier performance will be presented that are asympt
 otic in two regimes. First\, the notion of classification capacity is intr
 oduced\, which characterizes the number of classes that can be discriminat
 ed reliably as the signal dimensionality approaches infinity\; tight bound
 s on the classification capacity associated with Gaussian classes are pres
 ented. Second\, the notion of diversity-discrimination tradeoff is also in
 troduced\, which\, by analogy with the diversity-multiplexing tradeoff of 
 vector channels\, characterizes the tradeoff between the misclassification
  probability and the number of discernible classes as the signal-to-noise 
 ratio goes to infinity\; again tight bounds on this tradeoff are also prov
 en.\n\nThese results reveal that the “easiest” classes to discriminate
  correspond to (affine) subspaces drawn from an appropriate Grassmann mani
 fold\; they further reveal a precise relationship between signal and measu
 rement geometry and classifier performance. Numerical results\, including 
 a face recognition application\, validate this relationship in practice.\n
 \nThis represents joint work with Matthew Nokleby (Duke University\, USA) 
 and Robert Calderbank (Duke University\, USA)\n\n*BIO*: Miguel Rodrigues i
 s a Senior Lecturer with the Department of Electronic and Electrical Engin
 eering\, University College London\, U.K. He was previously with the Depar
 tment of Computer Science\, University of Porto\, Portugal\, rising throu
 gh the ranks from Assistant to Associate Professor\, where he also led the
  Information Theory and Communications Research Group at Instituto de Tel
 ecomunicações – Porto. He received the Licenciatura degree in Electr
 ical Engineering from the Faculty of Engineering of the University of Por
 to\, Portugal in 1998 and the Ph.D. degree in Electronic and Electrical E
 ngineering from University College London\, UK in 2002. He has carried ou
 t postdoctoral research work both at Cambridge University\, UK\, as well 
 as Princeton University\, USA\, in the period 2003 to 2007. He has also 
 held visiting appointments at Princeton University\, Duke University\, 
 Cambridge University\, and University College London in the period 2007 t
 o 2013.\n\nHis research interests are in the general areas of information 
 theory\, communications theory and statistical signal processing. He was 
 the recipient of the IEEE Communications and Information Theory Societies
  Joint Paper Award in 2011 for the work on Wireless Information-Theoretic
  Security (with M. Bloch\, J. Barros and S. W. McLaughlin).
LOCATION:BE4-38 (CBL Meeting Room)
END:VEVENT
END:VCALENDAR