Class discovery from semi-structured EEG data for affective computing and personalisation

Many approaches to recognising emotions from metrical data such as EEG signals rely on identifying a very small number of classes and to train a classifier. The interpretation of these classes varies from a single emotion such as stress [24] to features of emotional model such as valence-arousal [4]. There are two major issues here. First classification approach limits the analysis of the data within the selected classes and is also highly dependent on training data/cycles, all of which limits generalisation. Second issue is that it does not explore the inter-relationships between the data collected missing out on any correlations that could tell us interesting facts beyond emotional recognition. This second issue would be of particular interest to psychologists and medical professions. In this paper, we investigate the use of Self-Organizing Maps (SOM) in identifying clusters from EEG signals that could then be translated into classes. We start by training varying sizes of SOM with the EEG data provided in a public dataset (DEAP). The produced graphs showing Neighbour Distance, Sample Hits, Weight Position are analysed holistically to identify patterns in the structure. Following that, we have considered the ground- truth label provided in DEAP, in order to identify correlations between the label and the clustering produced by the SOM. The results show the potential of SOM for class discovery in this particular context. We conclude with a discussion on the implications of this work and the difficulties in evaluating the outcome.