The paper “Predicting Exact Valence and Arousal Values from EEG”, authored by Filipe Galvão, Soraia M. Alarcão and Manuel J. Fonseca, LASIGE researchers, has been published at MDPI Sensors Journal, a top-ranked journal (h5-index: 104).
In the paper, the authors propose a model for predicting the exact values of valence and arousal from EEG signals, in a subject-independent scenario. To create the regression model, the authors studied the best features, brain waves, and machine learning models that are currently in use for emotion classification. The systematic analysis revealed that the best prediction model uses a KNN regressor (K = 1) with Manhattan distance, features from the alpha, beta and gamma bands, and the differential asymmetry from the alpha band. Results, using the DEAP, AMIGOS and DREAMER datasets, showed that the model can predict valence and arousal values with a low error (MAE < 0.06, RMSE < 0.16) and a strong correlation between predicted and expected values (PCC > 0.80). A comparison with previous works (in a classification scenario), using the DEAP dataset, showed that the identified model presents the highest accuracies for two and four classes, achieving 89.8% and 84.4% respectively.
The paper is available here.