BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Machine Learning @ CUED
SUMMARY:Logistic Regression with a Laplacian prior on the
Eigenvalues: Convex duality and application to EEG
classification - Ryota Tomioka (University of Tok
yo / Fraunhofer FIRST)
DTSTART;TZID=Europe/London:20070315T130000
DTEND;TZID=Europe/London:20070315T140000
UID:TALK6851AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/6851
DESCRIPTION:We propose a matrix coefficient logistic regressio
n for the classification of\nsingle-trial ElectroE
ncephaloGraphy (EEG) signals. The method works in
the\n feature space of all the variances and covar
iances between electrodes.\nThe problem is formula
ted in a single convex optimization problem with t
he\nspectral $\\ell_1$-regularization of the coeff
icient matrix. In addition\, we propose\nan effici
ent optimization algorithm based on a simple inter
ior-point method.\nThe convex duality plays the ke
y role in this implementation.\nClassification res
ults on 162 Brain-Computer Interface (BCI) dataset
s\n show significant improvement in the classifica
tion accuracy against $\\ell_2$-regularized\nlogis
tic regression\, rank=2 approximated logistic regr
ession as well as\nCommon Spatial Pattern (CSP) ba
sed 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
CONTACT:Zoubin Ghahramani
END:VEVENT
END:VCALENDAR