| The rapid development of social economy has significantly improved people’s material living standards,but also brought a faster pace of life.The accelerated pace of life compresses the time that marriageable individuals can spend looking for their favorite partner,which is a very time-consuming and challenging task.Therefore,the fast-paced lifestyle virtually increases the difficulty for marriageable individuals to find their favorite partner,reduces the willingness of marriageable individuals to enter marriage,and then reduces the marriage rate of marriageable youth groups.The advent of the Internet era has brought new opportunities to solve this problem.People see that more and more marriageable individuals begin to find their favorite partner with the help of convenient Internet matchmaking platform and mobile dating software.On the Internet platform,the information available to marriageable individuals is very limited.Therefore,marriageable individuals usually make decisions on whether they want to further interact and develop romantic relationships with a potential romantic partner according to their first impression or romantic attraction.In this context,the role of romantic attraction in human mate selection becomes more and more important.However,so far,little is known about the neurophysiological activities of romantic attraction,and there is also no research on identifying individuals’ preference for potential romantic partners based on EEG signals.Studying the neurophysiological response of romantic attraction based on EEG signals can not only provide an effective window for understanding the subconscious process behind mate selection preference,but also provide an an effective reference for obtaining objective and effective neurophysiological characteristics to identify romantic attraction.The main purpose of this paper is to explore the neurophysiological activities closely related to the generation of romantic attraction based on EEG signals,and extract the neurophysiological indicators that can characterize the romantic attraction of marriageable individuals based on the laws found in the brain mechanism of romantic attraction,which can be used to identify the preference state of marriageable individuals for potential romantic partners,so as to help improve the effectiveness of the marriage algorithm used by Internet matchmaking platform and mobile dating software,help marriageable individuals find their favorite partner.In our research,we first simulated and built a mate selection platform similar to mobile dating software,and asked the subjects to use the platform to select their favorite partner.When participants assessed their romantic attraction to potential romantic partners on the simulation platform,their EEG signals and behavioral data were recorded in real time.We then analyze the neurophysiological activities related to the generation of romantic attraction from multiple dimensions such as time domain,source domain,and time-frequency domain.Then,the neurophysiological indexes associated with the generation of romantic attraction are extracted from EEG signals to recognize the romantic attraction of marriageable individuals,and investigate the possibility of predicting individual mate selection intention based on neurophysiological indexes.Finally,for portable human-computer interaction system,explore how to find the most effective EEG features and electrodes under the constraint of classification accuracy,so as to reduce the dimension of feature space and lead space,and optimize the convenience and practicability of EEG based romantic attraction recognition system.The main research contents and innovations of this paper are as follows:1.Romantic attraction data sets(RAD)are constructed.How to induce romantic attraction and obtain enough EEG data is an important and difficult problem.Aiming at the problems that EEG research requires a high amount of data,but the induction rate of romantic attraction is very low,and there is a lack of experimental paradigm to effectively increase the induction times of romantic attraction,this paper puts forward the solutions of "increasing the average induction rate of romantic attraction by increasing the proportion of high attractive stimulus materials" and "flexibly increasing the number of stimuli by designing a new simulated mate selection experimental paradigm",Successfully induced romantic attraction,collected a sufficient number of EEG data,and finally constructed the romantic attraction data set RAD required for the study.2.The brain mechanism of romantic attraction was studied based on EEG signals.Connecting the emotional state of interest(i.e.,romantic attraction)with the required neurophysiological signals(i.e.,EEG)is a key aspect of emotional computing research,but the pattern of EEG activity for romantic attraction is unclear.To solve this problem,based on the RAD data set constructed by the simulated mate selection experiment,this paper makes a systematic analysis from the three dimensions of time domain,source domain and time-frequency domain,and finds the EEG activity law closely related to the generation of romantic attraction.The study found for the first time that compared with the situation without romantic attraction,the generation of romantic attraction will lead to:(1)the amplitude of P300 and late positive potential(LPP)components of eventrelated potential(ERP)is significantly increased.The amplitude enhancement of P300 component is caused by the enhancement of neurophysiological activities in insula and cingulate cortex,the amplitude enhancement of LPP component is caused by the enhancement of neurophysiological activities in orbital frontal cortex,dorsolateral prefrontal cortex,cingulate cortex,frontal eye movement area,visual cortex and insula;(2)The phenomenon of event related de synchronization(ERD)in alpha band(8-13 Hz)and beta band(14-20 Hz)is enhanced,and this effect mainly occurs in frontal parietal region and lateral occipital complex region.3.A recognition model of romantic attraction is constructed.In this paper,we propose a specific solution to the problem that it is not clear whether participants’ preference states for potential romantic partners can be identified based on information extracted from EEG signals.Firstly,the preprocessed EEG signal is decomposed into power related features of different frequency bands by wavelet transform;Then,the proposed nested ten-fold cross-validation architecture(T-NCVA)model is used for feature selection and model training;Finally,the classification performance of the model is evaluated on the RAD data set constructed in the simulated mate selection experiment.The results of ten-fold cross-validation show that the average recognition accuracy can reach 85.28% when using the feature vector mainly composed of hemispheric asymmetry score(AS)features in alpha,beta and theta(21-30 Hz)bands.This result proves the feasibility of recognizing romantic attraction based on EEG signal.4.The optimization of romantic attraction recognition model is studied.Reducing the dimension of feature space and lead space can improve the convenience and practicability of mate preference recognition system based on EEG.Therefore,this paper proposes a frequency based feature subset reconstruction algorithm(F-FSRA)to reduce the dimensions of feature space and lead space,and applies the feature subset reconstructed by the algorithm to romantic attraction recognition,so as to optimize the romantic attraction recognition model.Firstly,the data set is divided according to the subjects,and the recursive feature elimination(RFE)algorithm is used to generate the middle feature subset under each data segmentation scheme;Secondly,the middle feature subset is vectorized and the middle feature matrix is constructed;Then,the feature weight is calculated based on the middle feature matrix,and a series of new feature subsets are generated according to the feature weight;Finally,the classification accuracy is calculated based on the newly generated feature subset,and the dimension of feature space and the number of electrodes required are reduced under the constraint of classification accuracy.Experimental results show that the proposed F-FSRA algorithm can improve the classification accuracy while reducing the dimensions of feature space and lead space.When using the best feature subset mainly from the frontal and parietal lobes,the average classification accuracy is up to 88.09%.The effectiveness of the proposed F-FSRA algorithm in optimizing the romantic attraction recognition model is proved.To sum up,the paper first induced the interested emotional state of romantic attraction by increasing the proportion of highly attractive stimulus materials and proposing a new experimental paradigm,and established the RAD data set.Then,taking the brain mechanism of romantic attraction as the starting point,this paper systematically analyzes the EEG activity law of romantic attraction,extracts effective features based on this,and verifies the feasibility of identifying romantic attraction based on EEG signals.Finally,starting with reducing the dimension of feature space and the number of electrodes needed,the romantic attraction recognition model is optimized by improving the feature selection algorithm.The research results of this paper not only help people understand the subconscious process behind mate selection preference,but also provide evidence support for evaluating and predicting the mate selection preference of marriageable individuals based on EEG signals.It is believed that in the near future,the mate selection preference recognition system based on EEG signal will be applied to online or offline mate selection scenarios to help marriageable individuals find their favorite partner. |
