Research On Emotion Recognition Method Based On Improved Brain Connectivity Reseroir

Emotion is the attitude experience and corresponding behavior reaction of human towards external objective things.Emotion recognition is an indispensable part of establishing a harmonious,friendly and intelligent human-computer interaction environment.Electroencephalogram(EEG)contacts directly with human central nervous system,reflecting emotional changes more realistically and intuitively.Consequently,the research of EEG emotion recognition has attracted more and more attention.Traditional EEG emotion recognition methods represent the EEG features in the temporalityfrequency domain,and then input these features into the classification model for emotion recognition.However,it is difficult for such methods to incorporate the correlation,temporality and nonlinearity of EEG in the feature representation of EEG.At the same time,traditional EEG emotion recognition methods do not take into account the spatiality and lateralization of EEG.In order to solve the problems mentioned above,this thesis studies an emotion recognition method based on improved brain connectivity reservoir.The specific work is as follows:(1)In view of the correlation,temporality and nonlinearity of EEG,an emotion recognition method based on brain connectivity reservoir is proposed.First,considering the correlation of EEG,a construction algorithm of brain connectivity reservoir is proposed.Second,considering the temporality of EEG,a bidirectional feature representation algorithm for EEG based on brain connectivity reservoir is proposed.Third,considering the nonlinearity of EEG,an emotion classification algorithm based on fully connected neural network is proposed.Forth,the specific process of the emotion recognition method based on brain connectivity reservoir is summarized.Finally,the experimental results show that the proposed method incorporate the correlation,temporality and nonlinearity of EEG in the feature representation of EEG,and improves the accuracy of emotion recognition,compared with traditional emotion recognition methods.(2)In view of the spatiality and lateralization of EEG,the brain connectivity reservoir is improved,and then an emotion recognition method based on graph convolutional neural network and valence lateralization is proposed.First,considering the topological relationship between EEG channels and the temporality of EEG,a feature enhancement algorithm of brain connectivity reservoir based on graph convolutional neural network is proposed.Second,considering the valence lateralization mechanism of the human brain,a brain connectivity reservoir training algorithm based on valence lateralization is proposed.Third,the specific process of the emotion recognition method based on graph convolutional neural network and valence lateralization is summarized.Finally,the experimental results show that the proposed method obtains more accurate feature representation and further improves the accuracy of emotion recognition,compared with the emotion recognition method based on the brain connectivity reservoir.