A comparison of different dimensionality reduction and feature selection methods for single trial ERP detection
Tian Lan1, Deniz Erdogmus, Lois Black
1Department of Science and Engineering, Oregon Health & Science University, Beaverton, Oregan, USA.
Summary
Principal Component Analysis (PCA) effectively reduces dimensions in electroencephalography (EEG) data for brain-computer interfaces. This method enhances event-related potential detection in both online and offline systems.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Dimensionality reduction and feature selection are crucial for electroencephalography (EEG) based event-related potential (ERP) detection.
- Brain-computer interfaces (BCIs) rely on accurate ERP detection for effective operation.
Purpose of the Study:
- To compare various dimensionality reduction and feature selection methods for ERP detection.
- To identify the optimal method for enhancing ERP detection in BCI systems.
Main Methods:
- Implemented a Rapid Serial Visual Presentation (RSVP) paradigm to collect EEG data.
- Utilized Linear Discriminant Analysis (LDA) as the ERP detector.
- Applied and compared different dimensionality reduction techniques within a greedy wrapper framework.
Main Results:
- Principal Component Analysis (PCA) with the first 10 principal components per channel demonstrated superior performance.
- The chosen PCA method proved effective for both online and offline ERP detection systems.
Conclusions:
- PCA is a highly effective technique for dimensionality reduction in EEG data for BCI applications.
- The optimized PCA approach enhances the accuracy and efficiency of ERP detection systems.
