An Algorithm to Detect P300 Potentials Based on F-Score Channel Selection and Support Vector Machines

To improve the classification accuracy of P300 potentials and the training speed of optimal support vector machines (SVM) classifier, a novel P300 detection algorithm based on F-score channel selection and SVM is proposed in this paper. Using F-score channel selection method, we reduce the task-irrelevant EEG channels to enhance the detection accuracy of P300 potentials. Meanwhile, by a new training set selection method given in this paper, we divide the primal training set into a training set and a validation set. With this validation set, the test error of the SVM classifiers can be predicted more accurately and quickly. Our algorithm was tested with a P300 dataset from the BCI competition 2003. And the results showed that the algorithm achieved an accuracy of 100% in P300 detection within four repetitions.