EEG Emotion Recognition of Different Brain Regions Based on 2DCNN-DGRU

In order to explore the differences in the activation of brain regions under emotional stimuli, and the best effect of emotion recognition under the combination of different brain regions, this paper divides the electrode channels into four different regions: Frontal lobe (Fron), Temporal lobe (Temp), Parietal lobe (Pari) and Occipital lobe (Occi), and Combine them into 9 schemes for further exploration. The differential entropy feature (DE) extracted on the emotion datasets SEED and DEAP is used as the input of the model, and the emotion categories are defined as positive, neutral, and negative. Secondly, in order to evaluate the effect of emotion recognition in brain regions, we propose a hybrid deep model— 2DCNN-DGRU. In the end, an average accuracy rate of 97.74% and 92.18% was obtained for the Whole brain region, respectively. In terms of single brain region, the Temp was the most critical, with an average accuracy rate of 94.99% and 80.16%. In the combination of the three brain regions Fron-Temp-Occi, the highest accuracy rate of 98.40% was achieved, surpassing that of the Whole brain region.