Data set I
‹motor imagery in ECoG recordings, session-to-session transfer›
Data set provided by University of
Dept. of Computer Engineering (Prof. Rosenstiel) and Institute of
Medical Psychology and Behavioral Neurobiology (Niels Birbaumer), and
Max-Planck-Institute for Biological Cybernetics, Tübingen, Germany
(Bernhard Schölkopf), and
Universität Bonn, Germany, Dept. of Epileptology (Prof. Elger)
Correspondence to Michael Schröder
The design of a classifier for a BCI system is
very challenging when a
classifier that was trained on the first day shall classify data
recorded during following days (if possible, without retraining):
the patient might be in a different state concerning motivation,
so that his brain will show different electrical activity. In addition,
the recording system might have undergone slight changes
concerning electrode positions and impedances.
Our data set reflects this sitation: training data and test data were
recorded from the same subject and with the same task, but on two
different days with about 1 week in between. Learn on the
data of the first session and do your best on the data of the second
During the BCI experiment, a subject had to
perform imagined movements of
either the left small finger or the tongue.
The time series of the electrical brain activity was picked up
during these trials using a 8x8 ECoG platinum electrode grid
which was placed on the contralateral (right) motor cortex.
The grid was assumed to cover the right motor cortex completely, but
its size (approx. 8x8cm) it partly covered also surrounding cortex
All recordings were performed with a sampling rate of 1000Hz.
After amplification the recorded potentials were stored as microvolt
Every trial consisted of either an imagined tongue or an imagined
finger movement and was recorded for 3 seconds duration. To avoid
visually evoked potentials being reflected by the data, the recording
intervals started 0.5 seconds after the visual cue had ended.
Format of the Data
The labeled training data (from the first
session) is stored as a zipped
matlab file called Competition_train.mat.gz. Use this data
to train your classification algorithms.
It consists of two parts:
The unlabeled test data is also stored as a
zipped matlab file called Competition_test.mat.gz.
It contains 100 trials of brain activity in matrix X (3D
format is the same as described above) but it contains no labels Y.
Part 1: the brain activity during 278
This part is stored in a 3D matrix named X using the
following format: [trials x electrode channels x samples of time
Part 2: the labels of the 278 trials.
This part is stored as a vector of -1 / 1 values named Y.
Please try to correctly classify the test data
set (100 trials).
Provide a list of the labels (-1/1) of these 100 trials in either ascii
or matlab format.
Send your labels by email to
with the subject "SOLUTION BCI COMPETITION".
How many of the labels you provided match the
true labels of the test set?
You can use our data set for your own
publications, if you please
- Thomas Lal, Thilo Hinterberger,
Guido Widman, Michael Schröder, Jeremy Hill, Wolfgang Rosenstiel,
Christian Elger, Bernhard Schölkopf, Niels Birbaumer.
Methods Towards Invasive Human Brain Computer Interfaces.
Advances in Neural Information Processing Systems (NIPS),
2004 (to appear).
Competition III ]