BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Logistic regression with a Laplacian prior on the singular values:
   convex duality and application to EEG classification. - Ryota Tomioka (
 冨岡亮太)\, University of Tokyo / Fraunhofer FIRST IDA
DTSTART:20070323T140000Z
DTEND:20070323T150000Z
UID:TALK6890@talks.cam.ac.uk
CONTACT:Christian Steinruecken
DESCRIPTION:We consider a matrix coefficient probabilistic classification 
 problem. We put a Laplacian prior on the singular values of the coefficien
 t matrix. The Laplacian prior not only keeps the singular value spectrum o
 f the regression coefficient sparse\, thus offering good interpretation of
  the solution\, but also is a key to good generalization. In addition\, we
  propose an efficient optimization algorithm based on interior-point metho
 d. The convex duality plays the key role in this implementation. We apply 
 the mehtod to motor-imagery EEG classification problem in the context of B
 rain-Computer Interface (BCI). Classification results on 162 BCI datasets 
 show significant improvement in the classification accuracy against $\\ell
 _2$-regularized logistic regression\, rank=2 approximated logistic regress
 ion as well as Common Spatial Pattern (CSP) based classifier\, which is a 
 popular technique\nin BCI. Connections to LASSO\, GP classification with a
  second order polynomial kernel\, and SVM are discussed.\n\nI would also l
 ike to talk about "Multiple-ouput Gaussian Process for Nonlinear System Id
 entification"\, which is still a vague idea but I hope I get some stimulat
 ing feedbacks. \n
LOCATION:Mott Seminar Room\, Cavendish Laboratory\, Department of Physics
END:VEVENT
END:VCALENDAR