0.25
0.5
0.75
1.25
1.5
1.75
2
EEG Feature Representations and Associated Spatial Filters for Brain-Computer Interfaces
Published on Dec 03, 20126738 Views
When designing EEG-based Brain-Computer Interfaces (BCI), a crucial signal processing component is the feature extraction step. It consists in representing EEG signals by a number of values that descr
Related categories
Chapter list
EEG Features and Spatial Filters towards “practical” BCI design00:00
Designing a practical BCI?00:28
Take home message02:25
Architecture of a BCI03:59
Today’s talk04:28
Outline 1.05:09
Feature extraction in context: the pattern recognition approach06:49
Oscillatory activity based BCI07:42
Example: Features for MI-based BCI08:52
Band power features09:33
Band power features (2)10:26
Basic features for left and right hand motor imagery11:11
That’s all? It’s that simple?12:09
Using more channels12:52
Spatial Filtering14:36
Some basic spatial filters15:04
Inverse Solutions15:49
Some remarks17:38
2 main types of inverse solutions19:44
Inverse solutions for BCI20:55
Example 23:12
FuRIA24:34
Evaluation26:38
Feature interpretability27:33
Results28:54
BUT!29:51
Why using Inverse solutions for BCI?32:31
Common Spatial Patterns (CSP) informally...35:55
CSP formally36:51
CSP in action38:56
CSP in action (2)40:31
CSP pros and cons41:06
Towards a more robust CSP?43:35
Regularized CSP (RCSP)45:15
What prior knowledge to use? Spatial knowledge to deal with noise47:06
Spatial filters obtained50:27
Regularization terms to deal with non-stationarities52:30
Combining multiple regularization terms54:41
Regularization terms to deal with limited data55:38
Evaluations: BCI competition IV, data set 2a57:37
Sparse CSP for channel selection59:10
Using a-priori knowledge for CSP01:00:18
Using Subject-specific frequency bands01:01:39
How to find the right frequency band with CSP?01:02:54
FBCSP Results01:04:20
Other Spatio-Spectral Filters01:05:07
Summary of features for ERD/ERS-based BCI01:05:51
Evoked potentials-based BCI01:06:46
Spatial Filters for Evoked Potentials: do we need them?01:07:13
Should we use spatial filters for Evoked Potentials?01:08:37
Spatial Filtering for EP: Why not using CSP?01:10:42
Fisher Spatial Filter (1)01:11:43
Fisher Spatial Filter (2)01:12:45
xDAWN spatial filter01:13:48
xDAWN spatial filter (2)01:16:36
xDAWN vs other spatial filters01:17:42
Impact of the training data size01:19:17
Alternative features01:21:02
Using multiple features01:23:13
Spatial filters for alternative features? 01:24:24
Example: Time Domain Parameters (TDP)01:25:23
Spatial filters for TDP01:26:34
Evaluation01:27:18
Alternative SF are valuable!01:28:32
Do we really need spatial filters?01:30:15
Classifying Covariance matrices01:31:38
Open research challenges01:32:50
Conclusion01:34:05
Thank you for your attention!01:35:14