BRAIN-COMPUTER INTERFACE
Ilya Kuzovkin
7 June 2014
B RAIN -C OMPUTER I NTERFACE Ilya Kuzovkin 7 June 2014 Now I know - - PowerPoint PPT Presentation
B RAIN -C OMPUTER I NTERFACE Ilya Kuzovkin 7 June 2014 Now I know how your brain signal looks like when you think LEFT and RIGHT Now I know how your brain signal looks like when you think LEFT and RIGHT Try
Ilya Kuzovkin
7 June 2014
Now I know how your brain signal looks like when you think “LEFT” and “RIGHT”
Now I know how your brain signal looks like when you think “LEFT” and “RIGHT” Try me — think
Now I know how your brain signal looks like when you think “LEFT” and “RIGHT” Try me — think
It was
Now I know how your brain signal looks like when you think “LEFT” and “RIGHT” Try me — think
How would you use such technology? It was
Mental intention
Mental intention Neuroimaging
Mental intention Neuroimaging
Name some neuroimaging techniques
Mental intention Signal Neuroimaging
Mental intention Signal Data Neuroimaging
Mental intention Signal Data Algorithm Neuroimaging
Mental intention Signal Data Algorithm With 87% certainty I can say that you are thinking “LEFT” right now Neuroimaging
http://biomedicalengineering.yolasite.com
http://biomedicalengineering.yolasite.com
http://www.conncad.com/gallery/single_cells.html
http://en.wikipedia.org/wiki/Neural_oscillation
http://en.wikipedia.org/wiki/Neural_oscillation
http://en.wikipedia.org/wiki/Neural_oscillation
http://en.wikipedia.org/wiki/Neural_oscillation
What is the frequency in this example?
Delta
0-4 Hz
Theta
4-7 Hz
Alpha
7-14 Hz
Mu
8-13 Hz
Beta
15-30 Hz
Gamma
30-100 Hz
Delta
0-4 Hz
Theta
4-7 Hz
Alpha
7-14 Hz
Mu
8-13 Hz
Beta
15-30 Hz
Gamma
30-100 Hz
slow wave sleep, babies, lesions children, drowsiness, meditation, relaxed closed eyes, relaxed motor neuron in rest, mirror neurons motor activity, anxious thinking, concentration networking between populations of neurons
TIME CHANNELS
TIME CHANNELS
Alpha
7-14 Hz
Beta
15-30 Hz
Gamma
30-100 Hz
TIME CHANNELS
Alpha
7-14 Hz
Beta
15-30 Hz
Gamma
30-100 Hz
TIME CHANNELS
Jean Baptiste Joseph Fourier 1768 — 1830
*discrete
*discrete
*discrete
signal at time t frequency complex number
*discrete
signal at time t frequency complex number
Amplitude of the component with frequency k
*discrete
signal at time t frequency complex number
Amplitude of the component with frequency k
Why not like this?
Are we done?
Are we done? Hint:
300 MS
300 MS
300 MS
300 MS
300 MS
11 channels 50 frequencies on each 3 seconds of data 300 ms window
1 reading of 300 ms?
300 MS
11 channels 50 frequencies on each 3 seconds of data 300 ms window
1 reading of 300 ms?
all 3 seconds of data?
INSTANCES FEATURES CLASSES
. . .
set of sample objects (samples) is called training set Machine Learning algorithm learns from examples,
Each object
called features
Tail length : ...
f1
Furriness : ... Tail length : ...
f1
f2
Furriness : ... Tail length : ...
f1
f2
f = ( f1, f2)
Form a feature vector
Instance Feature 1 Feature 2 Class Cat 1 8 cm 546 h/cm M Cat 2 7.5 cm 363 h/cm M ... ... ... Cat N 11 cm 614 h/cm F Together feature vectors and corresponding classes form a dataset
Feature vectors live in a feature space
?
Feature vectors live in a feature space
Feature vectors live in a feature space
Feature vectors live in a feature space
handling ¡(GMDH) ¡
learning ¡(PAC) ¡
algorithms ¡
algorithms ¡