Emotion Recognition Based On Multichannel Physiological Signals

Multi-channel physiological data sets are usually nonlinear separable in emotion recognition.Some researches have applied linear or partial nonlinear processing in feature reduction and classification,but didn’t work well.A nonlinear method is proposed to solve this problem.First,as traditional feature reduction may cause the loss of significant amounts of feature information,Kernel Principal Component Analysis(KPCA)based on radial basis function(RBF)was introduced to map the data into a high dimension space,extract the nonlinear information of the feature,and then reduce the dimension.This method can provide a large number of features carrying information about the structure in the physiological dataset.Next,considering its advantages of predictive power and feature selection from large number of features,Gradient Boosting Decision Tree(GBDT)was utilized as a nonlinear ensemble classifier to improve recognition accuracy.In the design of experiments,this paper collected four physiological signals(electrocardiogram,galvanic skin response,electromyography and photoplethysmography)of 29 subjects under four emotions(happiness,fear,sadness and anger),and constructed the multi-channel physiological dataset with 623 samples and 136 features by preprocessing and feature extraction.The nonlinear processing method had a good performance on the physiological dataset,classification accuracy of four emotions achieved93.42% through 10-fold cross-validation,which was higher than traditional emotion recognition model.Besides,we also compared the emotion discrimination ability of di?erent physiological signals,and the classification accuracy on four emotions.The results shown that among the four signals,the emotion recognition accuracy from ECG signal was the highest;among the four emotions,the classification accuracy on fear emotion was the highest.It can be inferred that ECG signal contains more emotionrelevant information;the physiological response of subjects to fear is stronger,or fear is more easily induced.