Emotion Recognition Based on Brain Connectivity Reservoir and Valence Lateralization for Cyber-Physical-Social Systems

As an important application of pattern recognition, emotion recognition can make Cyber-Physical-Social Systems (CPSS) provide more efficient services for humans. In order to improve the recognition accuracy, this paper proposes an electroencephalogram (EEG) emotion recognition method based on brain connec-tivity reservoir (BCR) and valence lateralization (VL) for CPSS. First, for the purpose of comprehensively considering the temporality, nonlinearity, and correlation of EEG signals, an emotion recognition model based on BCR is established. Specifically, according to the connectivity index, the correlation between EEG channels is calculated to determine the brain connectivity structure of BCR, and the features of EEG sig-nals are represented through BCR, then the classification result is obtained by the fully connected neural network according to the feature representation. Second, for the purpose of enhancing the feature rep-resentation capability of BCR, a training algorithm of BCR based on VL is proposed. Specifically, BCR is divided into two parts, i.e., the left hemi-BCR and the right hemi-BCR. These two parts are trained sep-arately, so that the lateralization characteristic of the brain is better reflected. Finally, the experimental results on DEAP demonstrate that the proposed method achieves a recognition accuracy of 85.55% which is higher than the state-of-the-art methods.(c) 2022 Elsevier B.V. All rights reserved.

Automated summary

This paper proposes an EEG emotion recognition method based on brain connectivity reservoir (BCR) and valence lateralization (VL) to improve the efficiency of Cyber-Physical-Social Systems (CPSS) in providing services for humans. The method establishes an emotion recognition model based on BCR to consider the temporality, nonlinearity, and correlation of EEG signals. Furthermore, a training algorithm based on VL is introduced to enhance the feature representation capability of BCR. Experimental results demonstrate that the proposed method achieves a recognition accuracy of 85.55%, outperforming state-of-the-art methods.