Exploring the performance of different classifiers in valence and arousal detection for emotion recognition. Valence and arousal levels are used to determine the individual's location on the valence/arousal axis seen below.
In this work, the DEAP Dataset was used to obtain the EEG signals used in classification. Different features were extracted, and classifiers for valence and arousal were trained accordingly. This work explores the effectiveness of Extreme Learning Machines (ELM) in emotion classification from EEG signals. Additionally, shows that ELM is capable of achieving similar training and testing performances to the statistical methods typically used on this dataset, but it manages to do that in a significantly shorter amount of time. The work done and the results can be seen in this report and this presentation.