Emotion Understanding And Expression Based On Individual Factors

In order to make the computer have higher and more comprehensive intelligence,we must give the computer the ability of cognition,understanding,expression and adaptation to human emotion to build a harmonious man-machine environment.This is the ultimate goal of affective computing research and the only way to move from weak AI to strong AI.There are two kinds of problems in affective computing,one is affective understanding,the other is affective expression.According to the current research,a key common problem in emotion understanding and expression is the important role of individuation in it.Generally speaking,the emotion understanding and expression are different with the individual's personality.Adaptively analyzing and expressing emotion according to specific personality is an inevitable trend of sentiment analysis research.In order to solve this problem,in terms of emotional understanding,this paper first considers the particularity of emotion,and different individuals may have different subjective responses to the same external stimulus.Therefore,the emotion recognition method based on EEG signal should be personalized.Personalized EEG emotion recognition is modeled from macro and micro levels in this paper.At the macro level,we use personality characteristics to cluster individual personalities from the perspective of "Birds of a feather flock together and people flock together".At the micro level,the deep learning model is used to extract the spatial and temporal characteristic information of EEG signals.In addition,in order to better characterize the specificity of individual stress response,we introduced channel weight layer to highlight the influence of EEG channels on personalized emotion recognition.In order to evaluate the effectiveness of this method,we conducted an experiment of EEG emotion recognition on the acknowledged data set.The experimental results show that the recognition accuracy of the proposed method in Valence and Arousal is 72.4% and75.9%,respectively,which is 10.2% and 9.1% higher than that of excision without considering individualization.Secondly,in terms of emotional expression,in order to improve the ability of machine or agent to understand and express emotions,this paper first inputs the above EEG emotion recognition results into the artificial emotion model,which can produce emotions similar to human emotions recognized,and adds the influence of the Big Five personality to the model.Secondly,we study the influence of Big Five personality on discrete emotional space,and propose a method of mapping between discrete emotional space and dimensional emotional space,which can easily transfer users' emotional state from one emotional space to another.Based on this mutual mapping,the Big Five personality model is integrated into the hidden Markov model for emotion regulation,so as to build a machine emotion model that can represent individuation.The simulation results show that the machine personalized emotion model proposed in this paper can better describe different personality characteristics and produce reasonable emotional responses.