Disjunctive normal unsupervised LDA for P300-based brain-computer interfaces

Official URL: http://risc01.sabanciuniv.edu/record=b1640583 (Table of Contents)

Can people use text-entry based brain-computer interface (BCI) systems and start a free spelling mode without any calibration session? Brain activities differ largely between people and across sessions for the same user. Thus, how can the text-entry system classify the target character among the other characters in the P300-based BCI speller matrix? In this thesis, we introduce a new unsupervised classifier for a P300-based BCI speller, which uses a disjunctive normal form representation to de ne an energy function involving a logistic sigmoid function for classification. Our proposed classifier updates the initialized random weights performing classification for the P300 signals from the recorded data exploiting the knowledge of the sequence of row/column highlights. To verify the effectiveness of the proposed method, we performed an experimental analysis on data from 7 healthy subjects, collected in our laboratory and used public BCI competition datasets. We compare the proposed unsupervised method to a baseline supervised linear discriminant analysis (LDA) classifier and Bayesian linear discriminant analysis (BLDA) and demonstrate its performance. Our analysis shows that the proposed approach facilitates unsupervised learning from unlabelled test data.