摘要
情绪是由某种特定的对象或情景所引发的一系列反应,会影响人的生理状态,因此可通过生理信号进行识别。提出一种融合脉搏波、皮肤电反应、呼吸、皮肤温度等多种信号的特征,结合SVM-RFE-CBR(基于支持向量机可减少相关性偏差的递归特征消除)特征排序算法进行特征选择,利用支持向量机进行分类的情绪识别模型,并通过DEAP数据集验证该模型在愉悦度、唤醒度、优势度上的二分类效果,分别获得了73.5%、81.3%、76.1%的准确率。结果表明,利用多种生理信号融合可有效地进行情绪识别。
Emotion is a series of reactions triggered by a specific object or situation that affects a person's physiological state and can,therefore,be identified by physiological signals.This paper proposes an emotion recognition model.Extracted the features of physiological signals such as photoplethysmography,galvanic skin response,respiration amplitude,and skin temperature.The SVM-RFE-CBR(Recursive Feature Elimination-Correlation Bias Reduction-Support Vector Machine)algorithm was performed to select features and support vector machines for classification.Finally,the model was implemented on the DEAP dataset for an emotion recognition experiment.In the rating scale of valence,arousal,and dominance,the accuracy rates of 73.5%,81.3%,and 76.1%were obtained respectively.The result shows that emotional recognition can be effectively performed by combining a variety of physiological signals.
作者
陈沙利
张柳依
江锋
陈婉琳
缪家骏
陈杭
CHEN Shali;ZHANG Liuyi;JIANG Feng;CHEN Wanlin;MIAO Jiajun;CHEN Hang(College of Biomedical Engineering and Instrument Science,Zhejiang University,Hangzhou,310027;Key Laboratory of Biomedical Engineering of Ministry of Education,Zhejiang University,Hangzhou,310027;Zhejiang Provincial Key Laboratory of Cardio-Cerebral Vascular Detection Technology and Medicinal Effectiveness Appraisal,Zhejiang University,Hangzhou,310027;Department of Psychology and Behavioral Science,Zhejiang University,Hangzhou,310027)
出处
《中国医疗器械杂志》
2020年第4期283-287,共5页
Chinese Journal of Medical Instrumentation
关键词
情绪识别
多生理信号
支持向量机
emotion recognition
multiple physiological signals
support vector machine
