Developing advanced learning algorithms to reliably classify the brain signals, for example, event-related potentials (ERP) and motor imagery signals. Following are relevant techniques for this topic:
-
Committee of classifiers, ensemble learning and boosting, SVM and Neural Networks;
-
Effective preprocessing methods, such as principal component analysis (PCA), independent component analysis ( ICA), sparse component analysis (SCA) and so on;
-
Feature extraction and feature selection;
-
Hidden Markov Models for tracking temporal variability;
-
Signal verification methods for asynchronous BCIs;
-
Semi-supervised learning for online adaptation;
-
Multiple sources of signals (eg. EEG, NIR, etc) and their possible combination to enhance the performance;
-
Adaptive and collaborative learning strategy and mechanism for more robust and more accurate BCIs.
