Ensemble of Riemannian Classifiers for Multimodal Data: FUCONE Approach for M/EEG Data
Résumé
Interfacing computers directly with brain signals is a fascinating but difficult task, due to subject variability. Methods
based on Riemannian geometry have improved the state of the art for Brain-Computer Interface (BCI), capturing some invariant signature in electroencephalography (EEG) signals. So far, simple features have been used and alternative
ones, reflecting interconnected nature of brain activity have been overlooked. Here, we evaluated a framework, called FUCONE, that consists in combining functional connectivity estimators, Riemannian geometry, and ensemble learning, and we assessed its relevance for fusion from magnetoencephalography (MEG) and EEG signals.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|
