Classification methods for Brain-Computer Interfaces

This study considers a real dataset provided by Brain Computer Interface competition-III (http://www.bbci.de/competition/iii/#goals). 41680 EEG samples were collected corresponding to one of the three classes (Imaginative left hand movement, Imaginative right hand movement and Word Association). The data collected are Raw EEG potentials and thus are noisy and have outliers. The dataset contains a set of 96 dimensional vector Precomputed features which are preprocessed raw EEG potentials. We can implement several classification techniques covered in class CSC 591 and can select the most efficient model depending upon the accuracy obtained in our experiments.

1. Khelan Patel (kjpatel4)
2. Saloni Desai (sndesai)