Brain-Computer Interface (BCI) can provide users with an alternative/augmentative interaction path, based on the interpretation of their brain activity. Steady State Visual Evoked Potential (SSVEP) is a good candidate for BCI-enabled communication/control applications. In this paper, we compare different reference signal processing methods, including two we developed ad hoc, assessing how they perform with respect to different indicators (not necessarily convergent, such as accuracy, computational effort and responsiveness). All the tests are performed on the subject population as a whole, in an effort to produce subject-independent methods. We also discuss a strategy for improving the classification accuracy by introducing an indicator related to the prediction confidence. Finally, a method for adaptively changing the length of the observed EEG window is presented.
Subject-independent, SSVEP-based BCI: Trading off among accuracy, responsiveness and complexity / Mora, Niccolo'; DE MUNARI, Ilaria; Ciampolini, Paolo. - ELETTRONICO. - 2015-July:(2015), pp. 146-149. (Intervento presentato al convegno 7th International IEEE/EMBS Conference on Neural Engineering, NER 2015 tenutosi a Montpellier; France nel 22 April 2015 - 24 April 2015) [10.1109/NER.2015.7146581].
Subject-independent, SSVEP-based BCI: Trading off among accuracy, responsiveness and complexity
MORA, Niccolo';DE MUNARI, Ilaria;CIAMPOLINI, Paolo
2015-01-01
Abstract
Brain-Computer Interface (BCI) can provide users with an alternative/augmentative interaction path, based on the interpretation of their brain activity. Steady State Visual Evoked Potential (SSVEP) is a good candidate for BCI-enabled communication/control applications. In this paper, we compare different reference signal processing methods, including two we developed ad hoc, assessing how they perform with respect to different indicators (not necessarily convergent, such as accuracy, computational effort and responsiveness). All the tests are performed on the subject population as a whole, in an effort to produce subject-independent methods. We also discuss a strategy for improving the classification accuracy by introducing an indicator related to the prediction confidence. Finally, a method for adaptively changing the length of the observed EEG window is presented.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
