Clustering strategies for optimal trial selection in multisensor environments. An eigenvector based approach.

Highlights: [•] In this paper we have considered real, clinical scenarios in magnetoencephalography. [•] We improve the performance of methods such as LCMV by using clustering techniques. [•] The basis for these clustering techniques is the covariance matrix diagonalization. [•] The eigenspace/eigenvector corresponding to the highest eigenvalue must be used. [•] The projection of this eigenspace across trials provides the dissimilarity matrix. [ABSTRACT FROM AUTHOR]