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 Eigenvalues: Con
 vex duality and application to EEG classification - Ryota Tomioka (Univers
 ity of Tokyo / Fraunhofer FIRST)
DTSTART:20070315T130000Z
DTEND:20070315T140000Z
UID:TALK6851@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:We propose a matrix coefficient logistic regression for the cl
 assification of\nsingle-trial ElectroEncephaloGraphy (EEG) signals. The me
 thod works in the\n feature space of all the variances and covariances bet
 ween electrodes.\nThe problem is formulated in a single convex optimizatio
 n problem with the\nspectral $\\ell_1$-regularization of the coefficient m
 atrix. In addition\, we propose\nan efficient optimization algorithm based
  on a simple interior-point method.\nThe convex duality plays the key role
  in this implementation.\nClassification results on 162 Brain-Computer Int
 erface (BCI) datasets\n show significant improvement in the classification
  accuracy against $\\ell_2$-regularized\nlogistic regression\, rank=2 appr
 oximated logistic regression as well as\nCommon Spatial Pattern (CSP) base
 d classifier\, which is a popular technique\nin BCI. Connections to LASSO\
 , GP classification with a second order\npolynomial kernel\, and SVM are d
 iscussed.
LOCATION:LT1 (Inglis Building) Engineering\, Department of
END:VEVENT
END:VCALENDAR