Classification of multi-modal data in a self-paced binary BCI in freely moving animals

The goal of the present article is to compare different classifiers using multi-modal data analysis in a binary self-paced BCI. Individual classifiers were applied to multi-modal neuronal data which was projected to a low dimensional space of latent variables using the Iterative N-way Partial Least Squares algorithm. To create a multi-way feature array, electrocorticograms (ECoG) recorded from animal brains were mapped to the spatial-temporal-frequency space using continuous wavelet transformation. To compare the classifiers BCI experiments were simulated. For this purpose we used 9 recordings from behavioral experiments previously recorded in rats free to move in a nature like environment.