EEG Signal Based Multi Class Emotion Recognition using Hybrid 1D-CNN and GRU

Abstract

In this study, a hybrid architecture combining a Convolutional Neural Network (1D-CNN) and Gated Recurrent Unit (GRU) is proposed for multi-class emotion recognition using EEG signals.Emotion recognition using EEG signals is a challenging task due to the ever-changing nature of EEG signals and the high dimensionality of the feature space. The proposed approach aims to address these challenges by utilizing a hybrid architecture that combines the strengths of both 1D-CNN and GRU. The 1D-CNN is used to retrieve relevant spatial features from the EEG signals, while the GRU is employed to capture the temporal dependencies in the signals. The models were used to classify multi-class emotions: four and sixteen emotions based on the valence-arousal and valence-arousal-liking-dominance planes, respectively, using the benchmark DEAP dataset. The experiment results showed that the proposed models achieved high accuracy in classifying emotions for both four and sixteen emotions as compared to state of art methods. The results of this research have significant implications for the development of affective computing systems in various fields, including healthcare, human-computer interaction, and education. In conclusion, this study demonstrates the potential of deep learning models in affective computing and provides a foundation for future research in this field. The use of reliable physiological signals and the combination of different architectures have shown to be effective in accurately classifying emotions.