This study addresses the problem of classifying emotional words based on recorded electroencephalogram （EEG） signals by the single-trial EEG classification technique. Emotional two-character Chinese words are used a...This study addresses the problem of classifying emotional words based on recorded electroencephalogram （EEG） signals by the single-trial EEG classification technique. Emotional two-character Chinese words are used as experimental materials. Positive words versus neutral words and negative words versus neutral words are classified, respectively, using the induced EEG signals. The method of temporally regularized common spatial patterns （TRCSP） is chosen to extract features from the EEG trials, and then single-trial EEG classification is achieved by linear discriminant analysis. Classification accuracies are between 55% and 65%. The statistical significance of the classification accuracies is confirmed by permutation tests, which shows the successful identification of emotional words and neutral ones, and also the ability to identify emotional words. In addition, 10 out of 15 subjects obtain significant classification accuracy for negative words versus neutral words while only 4 are significant for positive words versus neutral words, which demonstrate that negative emotions are more easily identified.展开更多
A local and global context representation learning model for Chinese characters is designed and a Chinese word segmentation method based on character representations is proposed in this paper. First, the proposed Chin...A local and global context representation learning model for Chinese characters is designed and a Chinese word segmentation method based on character representations is proposed in this paper. First, the proposed Chinese character learning model uses the semanties of loeal context and global context to learn the representation of Chinese characters. Then, Chinese word segmentation model is built by a neural network, while the segmentation model is trained with the eharaeter representations as its input features. Finally, experimental results show that Chinese charaeter representations can effectively learn the semantic information. Characters with similar semantics cluster together in the visualize space. Moreover, the proposed Chinese word segmentation model also achieves a pretty good improvement on precision, recall and f-measure.展开更多
基金The National Natural Science Foundation of China(No.61375118)the Program for New Century Excellent Talents in University of China(No.NCET-12-0115)
文摘This study addresses the problem of classifying emotional words based on recorded electroencephalogram （EEG） signals by the single-trial EEG classification technique. Emotional two-character Chinese words are used as experimental materials. Positive words versus neutral words and negative words versus neutral words are classified, respectively, using the induced EEG signals. The method of temporally regularized common spatial patterns （TRCSP） is chosen to extract features from the EEG trials, and then single-trial EEG classification is achieved by linear discriminant analysis. Classification accuracies are between 55% and 65%. The statistical significance of the classification accuracies is confirmed by permutation tests, which shows the successful identification of emotional words and neutral ones, and also the ability to identify emotional words. In addition, 10 out of 15 subjects obtain significant classification accuracy for negative words versus neutral words while only 4 are significant for positive words versus neutral words, which demonstrate that negative emotions are more easily identified.
基金Supported by the National Natural Science Foundation of China(No.61303179,U1135005,61175020)
文摘A local and global context representation learning model for Chinese characters is designed and a Chinese word segmentation method based on character representations is proposed in this paper. First, the proposed Chinese character learning model uses the semanties of loeal context and global context to learn the representation of Chinese characters. Then, Chinese word segmentation model is built by a neural network, while the segmentation model is trained with the eharaeter representations as its input features. Finally, experimental results show that Chinese charaeter representations can effectively learn the semantic information. Characters with similar semantics cluster together in the visualize space. Moreover, the proposed Chinese word segmentation model also achieves a pretty good improvement on precision, recall and f-measure.
