Classification of covariance matrices using a Riemannian-based kernel for BCI applications

摘要: The use of spatial covariance matrix as a feature is investigated for motor imagery EEG-based classification in brain-computer interface applications. A new kernel derived by establishing connection with the Riemannian geometry symmetric positive definite matrices. Different kernels are tested, combination support vector machines, on past BCI competition dataset. We demonstrate that this approach outperforms significantly state art results, effectively replacing traditional filtering approach.