基于同步性脑网络的支持张量机情绪分类研究  

Research on Support Tensor Machine Based on Synchronous Brain Network for Emotion Classification

作　　者：黄丽亚[1] 苏义博 马捃凯 丁威威 宋传承 HUANG Liya;SU Yibo;MA Junkai;DING Weiwei;SONG Chuancheng(School of Electronic and Optical Engineering&Microelectronics,Nanjing University of Posts and Telecommunications,Nanjing 210023,China;Bell Honors Shool,Nanjing University of Posts and Telecommunications,Nanjing 210023,China)

机构地区：[1]南京邮电大学电子与光学工程学院微电子学院,南京210023 [2]南京邮电大学贝尔英才学院,南京210023

出　　处：《电子与信息学报》2020年第10期2462-2470,共9页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(61977039)。

摘　　要：一直以来,情绪是心理学、教育学、信息科学等多个学科的研究热点,脑电信号(EEG)因其客观、不易伪装的特点,在情绪识别领域受到广泛关注。由于人类情绪是大脑多个脑区相互作用产生的,该文提出一种基于同步性脑网络的支持张量机情绪分类算法(SBN-STM),该算法采用相位锁定值(PLV)构建了同步性脑网络,分析多导联脑电信号之间的同步性和相关性,并生成2阶张量序列作为训练集,运用支持张量机(STM)模型实现正负情绪的二分类。该文基于DEAP脑电情绪数据库,详细分析了同步性脑网络张量序列的选取方法,最佳张量序列窗口的大小和位置,解决了传统情绪分类算法特征冗余的问题,提高了模型训练速度。仿真实验表明,基于支持张量机的同步性脑网络分类方法的情绪准确率优于支持向量机、C4.5决策树、人工神经网络、K近邻等以向量为特征的情绪分类模型。Emotion has always been a research hot spot in many disciplines such as psychology,education,and information science.Electro EncephaloGram(EEG)signal has received extensive attention in the field of emotion recognition because of its objective and not easy to disguise.Since human emotions are generated by the interaction of multiple brain regions in the brain,an algorithm of Support Tensor Machine based on Synchronous Brain Network(SBN-STM)for emotion classification is proposed.The algorithm uses Phase Locking Value(PLV)to construct a synchronous brain network,in order to analyze the synchronization and correlation between multi-channel EEG signals,and generate a second-order tensor sequence as a training set.The Support Tensor Machine(STM)model can distinguish a two-category of positive and negative emotions.Based on the DEAP EEG emotion database,this paper analyzes the selection method of synchronic brain network tensor sequence,the research on the size and position of the optimal tensor sequence window solves the problem of traditional emotion classification algorithm which always exists feature redundancy,and improves the model training speed.The results show that the accuracy of the emotional classification method based on SBN-STM is better than support vector machine,C4.5 decision tree,artificial neural network,and K-nearest neighbor which using vectors as input feature.

关 键 词：情绪分类 同步性脑网络 支持张量机 相位锁定值 

分 类 号：TN911.7[电子电信—通信与信息系统] TP391.4[电子电信—信息与通信工程]