Research On Subliminal Emotional Face Recognition And Classification Based On EEG Signals

Emotion is a state that integrates people's feelings,thoughts,and behaviors,which plays an important role in person-to-person communication.Emotional faces are of great significance for humans to express emotions.Many psychologists emphasize the role of consciousness in affective priming and believe that emotion can be conscious and perceived by humans.However,exploring the issue of emotions from an unconscious perspective is also an important direction of psychological research.Studies show that subliminal emotional face stimulation can trigger emotions,confirming the subliminal affective priming,which is also called unconscious affective priming.The paper combines subliminal unconscious emotion and feature extraction and classification methods of EEG signals together,to classify the subliminal emotional faces based on EEG signals,and investigate whether subliminal unconscious emotion is related to the positive or negative emotional faces of the subliminal stimulate.The main innovative points and contributions of the paper are shown as follows:?1?The paper proposes EEG feature extraction methods based on multi-scale sample entropy?MSpEn?and wavelet packet decomposition.The feature extraction method based on multi-scale sample entropy combines the multi-scale analysis with the sample entropy algorithm together.The paper calculates the MSpEn of the EEG signals and perform KS test on it.The result shows that the MSpEn can effectively distinguish the different subliminal emotional faces.The EEG feature extraction method based on wavelet packet decomposition extracts the wavelet packet entropy?WpEn?of the EEG signals and the wavelet packet energy?E_i?corresponding to the?4,0?^4,3nodes after the wavelet packet decomposition.Experiments prove that the features extracted by the above methods can effectively distinguish different subliminal emotional faces.?2?The paper combined the classification algorithms with subliminal emotional face classification research together,proposed a subliminal emotional face classification method based on decision tree algorithm.Firstly,the decision tree algorithm principle is introduced.Then,the the classification results are obtained by inputting feature vector.The classification results of different features are compared.Next,the paper proposed an improved random forest algorithm.In the selection of the splitting attribute,the information gain rate is compensated by introducing the Pearson coefficient,thereby reducing the bias of the feature with more values.At the same time,a weighted voting method is proposed to improve the accuracy of random forests.The subliminal emotional face classification method based on decision tree and improved random forest algorithm achieves good classification results of subliminal emotional faces.