Research On EEG Affective Computing And Brain Network Under Virtual Reality

Affective computing is one of the important research areas of artificial intelligence.Because EEG signals are easy to collect and can reflect brain thinking activities,so emotional computing based on EEG signals is one of the effective methods to analyze emotional changes,and it is also emotional The design of intelligent robots provides a certain technological basis.This study uses VR and EEG to design a set of visual evoked paradigms based on virtual reality scenes,and use this paradigm to conduct EEG emotion computing research and brain network research.At present,virtual reality technology is used as a new auxiliary treatment method in the medical field,but it is mostly an open-loop system application.To this end,this article combines virtual reality technology with EEG emotional computing,and analyzes the brain network:(1)Experimental paradigm design and data collection.This paper proposes an experimental paradigm using virtual reality video as an inducement method.It designs10 virtual reality stimulus segments,and combines the EEG data of 20 healthy subjects with the SAM emotional self-assessment scale.(2)Research on EEG Emotional Computing and Classification.This study use principal component analysis to conduct time-Frequency domain feature fusion.The results by SVM show that the differential entropy's accuracy is 80.18%.The full-band frequency domain feature(80.06%)has a better classification effect than the sub-band frequency.Combined feature reaches 82.83%.The result on the international sentiment DEAP database reached 77.31%,which shows that the emotional EEG database established by virtual reality technology in this paper is better.(3)Granger analysis of brain function network due to effect.In order to explore the brain connection status and information exchange efficiency in different virtual scenarios,this paper selects 6 channels of the prefrontal lobe,frontal lobe,and occipital lobe as regions of interest.Then uses Granger causality to establish causal networks and use statistical analysis to compare the results of different environments.The results show that there is no information exchanged between the prefrontal lobe in the virtual reality scenario;the overall information flow is mainly from the left hemisphere to the right hemisphere;the characteristics of each network also reflect that the left hemisphere is significantly more than the right hemisphere in the virtual reality scenario Brain is active.According to the one-way analysis of variance of the global efficiency,there is no difference between the virtual reality scenario.In this paper,the Granger factor-effect matrix of alpha and beta bands is calculated,and the results are statistically analyzed.It is indicating that there are obvious differences in the way of communication of brain waves in the frequency band.