Statistical machine learning and its application to neonatal seizure detection

Statistics Seminar Series 2009
-
10



Title:

Statistical machine learning and its
application to neonatal seizure
detection


Speaker:

Dr. Andrey Temko
,

Neonatal Brain Research Group,


Department of Electrical & Electronic Engineering,
University College Cor
k
,


Venue and time:

We
stern Gateway (IT) Building,
G16

Mon
19
th

October

4
-
5pm


Abstract
:

Recent work on statistical machine learning has shown the advantages of discriminative
classifiers such as Support Vector Machines (SVM) in a range of applications, in
cluding
seizure detection. In this talk, basics of an SVM classifier will be given followed by
description of the SVM
-
based neonatal seizure detector developed at UCC. An
alternative developed at UCC which is based on Gaussian mixture model classifier will

be also described and the comparison of performance between systems based on
discriminative SVM and generative GMM classifiers will be proposed.