en-es
en
0.25
0.5
0.75
1.25
1.5
1.75
2
Analysing Non-stationarity in EEG-BCI
Published on Apr 03, 20142868 Views
EEG is a highly complex signal. One of the main challenges of EEG analysis is to robustify against artifacts, non-stationarities and task unrelated variability. This holds in particular for EEG experi
Related categories
Chapter list
BCI and Nonstationarity00:00
Noninvasive Brain - Computer Interface00:56
Towards imaginations: Modulation of Brain Rhythms01:49
BBCI paradigms02:44
Playing with BCI: training session (20 min)03:52
Machine learning approach to BCI: infer prototypical pattern04:51
BBCI Set-up05:27
Spelling with BBCI: a communication for the disabled05:54
Future Issues: Shifting distributions within experiment10:18
Mathematical flavors of non-stationarity14:06
Neurophysiological analysis21:12
Weighted Linear Regression for covariate shift compensation22:15
Projections - Nonstationary28:51
Source separation paradigms - 129:03
Source separation paradigms - 229:19
Source separation paradigms - 331:53
The Stationary Subspace Analysis model32:17
Splitting into stationary and nonstationary subspace: SSA32:25
SSA36:24
Inverting the SSA model - 136:47
Inverting the SSA model - 237:44
Inverting the SSA model - 338:13
Inverting the SSA model - 438:27
Inverting the SSA model - 538:33
Identifiability - 138:35
Identifiability - 238:36
Identifiability - 338:51
The SSA algorithm39:01
The algorithm: optimizing stationarity39:41
Simplifying the objective (symmetries!)41:04
Optimizing in the special orthogonal group44:09
SSA: how many epochs?44:59
Identifiability: theoretical results46:05
Simulations on synthetic data47:02
Application to Brain-Computer-Interfacing51:23
Application to EEG analysis54:44
What are the strongest changes in the data?55:11
Results on one subject - 156:21
Results on one subject - 257:27
PCA and ICA do not find nonstationarities57:57
Classification of SSA directions59:18
What happens during a trial? (on average) - 101:00:29
What happens during a trial? (on average) - 201:01:20
Summary: stationary subspace analysis01:01:59
Real Man Machine Interaction01:02:53
Multimodal - Nonstationary01:02:55
Towardsa subject independent BCI decoder01:03:10
Model formulation01:04:03
Linear Mixed EffectsModel: intuition01:04:04
Multimodal - Nonstationary01:04:06
Motivation: Shifting distributions within experiment01:04:08
Cartoon: learn from adverse nonstationary subspace across subjects01:07:48
Algorithm01:09:21
Changes are similar!01:11:48
Results - 101:13:21
Interpretation01:14:11
Feature distribution becomes stationary01:15:01
Summary 201:15:35
Multimodal - Nonstationary01:16:17
BCI Pipeline01:17:02
Divergence CSP Framework01:17:33
Robustness through Beta Divergence01:20:55
Robustness Property01:22:00
Beta divergence CSP01:23:16
Simulations01:24:12
Results - 201:24:18
Results - 301:26:00
Different Kinds of Regularization01:27:23
Results - 401:27:46
Reducing Shif tbetweenTraining and Test01:28:23
Regularization Towards other Subjects01:28:37
Invariance Through Regularization01:29:49
Summary 301:30:02
Illiterates - Nonstationarity01:30:46
Conclusion01:30:51