Setting the Parameters for an Accurate EEG (Electroencephalography)-Based Emotion Recognition System

The development of a suitable EEG-based emotion recognition system has become a target in the last decades for BCI (Brain Computer Interface) applications. However, there are scarce algorithms and procedures for real time classification of emotions. In this work we introduce a new approach to select the appropriate parameters in order to build up a real-time emotion recognition system. We recorded the EEG-neural activity of 5 participants while they were looking and listening to an audiovisual database composed by positive and negative emotional video clips. We tested 11 different temporal window sizes, 6 ranges of frequency bands and 5 areas of interest located mainly on prefrontal and frontal brain regions. The most accurate time window segment was selected for each participant, giving us probable positive and negative emotional characteristic patterns, in terms of the most informative frequency-location pairs. Our preliminary results provide a reliable way to establish the more appropriate parameters to develop an accurate EEG-based emotion classifier in real-time.