Ensembles of temporal filters enhance classification performance for ERD-based BCI systems
Siamac Fazli, Cristian Grozea, Marton Donaczy, Benjamin Blankertz, Klaus-Robert Müller and Florin Popescu
Proceedings of the 4th International Brain-Computer Interface Workshop and Training Course 2008 2008.

Abstract

Finding suitable subject-dependent temporal and spatial filters is of paramount importance for achieving high information transfer rates in BCI systems, depending on Event- Related (De-)Synchronization (ERD/ERS). The temporal filter can be chosen manually, or automatically by use of a heuristic. We employ a multi-classifier system (MCS), based on a predefined filter-bank of temporal filters and apply it to 91 datasets, comprising 2-class experiments from 45 subjects. Our results indicate that this approach is a viable alternative to existing methods and can completely automate the temporal filter choice while promising higher performance than either a broad, subject-independent frequency band choice or the current heuristic.

PDF - Requires Adobe Acrobat Reader or other PDF viewer.

[Edit]