Title: A dynamic P300-based BCI speller using a language model

Authors: Anca Ralescu; Mengxia Wang; Shikha Chaganti; Qiang Han

Addresses: Machine Learning and Computational Intelligence Lab, Department of Electrical Engineering and Computing Systems, University of Cincinnati, ML 0030, Cincinnati, OH 45221-0030, USA ' Machine Learning and Computational Intelligence Lab, Department of Electrical Engineering and Computing Systems, University of Cincinnati, ML 0030, Cincinnati, OH 45221-0030, USA ' Machine Learning and Computational Intelligence Lab, Department of Electrical Engineering and Computing Systems, University of Cincinnati, ML 0030, Cincinnati, OH 45221-0030, USA ' Machine Learning and Computational Intelligence Lab, Department of Electrical Engineering and Computing Systems, University of Cincinnati, ML 0030, Cincinnati, OH 45221-0030, USA

Abstract: The dynamic P300-based speller adjusts the number of flashes per character according to the character's probability of occurrence, as predicted by a language model. The speller consists of two modules: the modified P300 speller using a row-column paradigm, and the prediction by partial matching (PPM) language module. Two cases are considered, prediction hit and prediction miss, according to whether the character predicted by the model coincides with the character intended by the subject. Preliminary experimental results point to the possible advantages of the modified P300 speller which reduces total flash time, while preserving performance.

Keywords: brain computer interface; BCI; P300 speller; oddball paradigm; prediction by partial match; PPM; language model; modelling; flashes per character; character occurrence probability; total flash time.

DOI: 10.1504/IJAIP.2014.059587

International Journal of Advanced Intelligence Paradigms, 2014 Vol.6 No.1, pp.1 - 13

Published online: 28 Jun 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article