The goal of this challenge was to discriminate the Magneto-Encephalography (MEG) responses to happy vs non-happy faces recorded from four subjects with 204 planar gradiometers. Six possible facial expressions could be presented to the subject. The subject’s task was to press a button whenever a happy face was presented. Therefore, the MEG activity of interest is likely to be dominated by 1) the event-related field (ERF) specific to happy faces and 2) the neural activity induced by the following button press.

I entered this challenge with Jean-Remi King, and the winning solution was a ensemble of 3 Riemannian geometry based classification pipeline.