Optimizing Crossvalidation Setup of Movement EEG Classifier

Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F14%3A00219202" target="_blank" >RIV/68407700:21230/14:00219202 - isvavai.cz</a>
Result on the web
—
DOI - Digital Object Identifier
—

Result language
angličtina
Original language name
Optimizing Crossvalidation Setup of Movement EEG Classifier
Original language description
The contribution presents results of a study on optimal crossvalidation setup for Brain Computer Interface (BCI) classifier. A wide range of settings of N-times repeated stratified $K$-fold crossvalidation is examined in order to determine the optimal crossvalidation setup and to study the relation between result stability and computational demands. To perform the extensive experiments required by this study an parallel extension to the existing EEG toolbox was developed. The toolbox can thus utilize the performance of modern symmetric multiprocessing PCs. This allows us to do experiments on BCI setup more quickly than ever before, to use advanced optimizations and to perform more detailed experiments. Using the toolbox extension, a speed-up factor ofup to 4.4 is achieved. 10-times repeated stratified 5-fold crossvalidation is recommended for future setup.
Czech name
—
Czech description
—

Publication year
2014
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection
POSTER 2014 - 18th International Student Conference on Electrical Engineering
ISBN
978-80-01-05499-4
ISSN
—
e-ISSN
—
Number of pages
6
Pages from-to
—
Publisher name
Czech Technical University
Place of publication
Prague
Event location
Praha
Event date
May 15, 2014
Type of event by nationality
EUR - Evropská akce
UT code for WoS article
—

Similar results(10)