Authors:
N. S. Dias
1
;
M. Kamrunnahar
2
;
P. M. Mendes
1
;
S. J. Schiff
2
and
J. H. Correia
3
Affiliations:
1
University of Minho, Portugal
;
2
The Pennsylvania State University, United States
;
3
Industrial Electronics Department, University of Minho, Portugal
Keyword(s):
BCI, EEG, feature selection.
Related
Ontology
Subjects/Areas/Topics:
Applications and Services
;
Artificial Intelligence
;
Biomedical Engineering
;
Biomedical Signal Processing
;
Computer Vision, Visualization and Computer Graphics
;
Data Manipulation
;
Health Engineering and Technology Applications
;
Human-Computer Interaction
;
Informatics in Control, Automation and Robotics
;
Medical Image Detection, Acquisition, Analysis and Processing
;
Methodologies and Methods
;
Neurocomputing
;
Neurotechnology, Electronics and Informatics
;
Pattern Recognition
;
Physiological Computing Systems
;
Sensor Networks
;
Signal Processing, Sensors, Systems Modeling and Control
;
Soft Computing
;
Time and Frequency Response
;
Time-Frequency Analysis
Abstract:
A new formulation of principal component analysis (PCA) that considers group structure in the data is proposed as a Variable Subset Selection (VSS) method. Optimization of electrode channels is a key problem in brain-computer interfaces (BCI). BCI experiments generate large feature spaces compared to the sample size due to time limitations in EEG sessions. It is essential to understand the importance of the features in terms of physical electrode channels in order to design a high performance yet realistic BCI. The VSS produces a ranked list of original variables (electrode channels or features), according to their ability to discriminate between tasks. A linear discrimination analysis (LDA) classifier is applied to the selected variable subset. Evaluation of the VSS method using synthetic datasets selected more than 83% of relevant variables. Classification of imagery tasks using real BCI datasets resulted in less than 16% classification error.