论文部分内容阅读
通过使用单次提取脑电信号的分类技术进行情绪词的脑电(EEG)识别研究.以中文情绪双字词为实验材料,通过其诱发的EEG信号,对正性词与中性词、负性词与中性词分别进行分类.使用时域正则化的共空间模式对单次提取脑电信号进行特征提取,并利用线性判别分析方法进行特征分类,分类准确率集中于55%~65%.置换检验验证了实验分类准确率的统计学显著性,表明了情绪词和中性词的成功识别,也有效地证实了基于脑电信号的语言情绪信息的可识别性.此外,在15名被试中,10名被试的负性词与中性词识别率显著,而仅有4名被试的正性词与中性词识别率显著,说明负性情绪更易被识别.
EEG recognition of emotional words was conducted by using the single classification of EEG signals.With Chinese double word of emotion as the experimental material and positive and negative words Sexual words and neutral words were classified respectively.Using space-time regularized co-space model to extract single EEG signals, the features were classified by linear discriminant analysis, and the accuracy of classification was 55% -65% The permutation test verifies the statistical significance of the accuracy of experimental classification and indicates the successful recognition of emotional words and neutral words and also effectively confirms the recognizability of linguistic emotional information based on EEG.Furthermore, In the subjects, the positive rate of negative words and neutral words of 10 subjects were significant, while the positive rate of positive words and neutral words of only 4 subjects was significant, which showed that negative emotion was more easily identified.