Electroencephalographic(EEG)-based emotion recognition has received increasing attention in the field of human-computer interaction(HCI)recently,there however remains a number of challenges in building a generalized emotion recognition model,one of which includes the difficulty of an EEG-based emotion classifier trained on a specific task to handle other tasks.Lit-tle attention has been paid to this issue.The current study is to determine the feasibility of coping with this challenge using feature selection.12 healthy volunteers were emotionally elicited when conducting picture induced and videoinduced tasks.Firstly,support vector machine(SVM)classifier was examined under within-task conditions(trained and tested on the same task)and cross-task conditions(trained on one task and tested on another task)for pictureinduced and videoinduced tasks.The within-task classification performed fairly well(classification accuracy:51.6%for picture task and 94.4%for video task).Cross-task classification,however,deteriorated to low levels(around 44%).Trained and tested with the most robust feature subset selected by SVM-recursive feature elimination(RFE),the performance of cross-task classifier was significantly improved to above 68%.These results suggest that cross-task emotion recognition is feasible with proper methods and bring EEG-based emotion recognition models closer to being able to discriminate emotion states for any tasks.
LIU ShuangMENG JiayuanZHAO XinYANG JiajiaHE FengQI HongzhiZHOU PengHU YongMING Dong
A brain-computer interface(BCI)-based electric wheelchair control system was developed, which enables the users to move the wheelchair forward or backward, and turn left or right without any pre-learning. This control system makes use of the amplitude enhancement of alpha-wave blocking in electroencephalogram(EEG) when eyes close for more than 1 s to constitute a BCI for the switch control of wheelchair movements. The system was formed by BCI control panel, data acquisition, signal processing unit and interface control circuit. Eight volunteers participated in the wheelchair control experiments according to the preset routes. The experimental results show that the mean success control rate of all the subjects was 81.3%, with the highest reaching 93.7%. When one subject's triggering time was 2.8 s, i.e., the flashing time of each cycle light was 2.8 s, the average information transfer rate was 8.10 bit/min, with the highest reaching 12.54 bit/min.
In order to investigate the characteristics of sensorimotor cortex during motor execution(ME), voluntary, stimulated and imaginary finger flexions were performed by ten volunteer subjects. Electroencephalogram(EEG) data were recorded according to the modified 10-20 International EEG System. The patterns were compared by the analysis of the motion-evoked EEG signals focusing on the contralateral(C3) and ipsilateral(C4) channels for hemispheric differences. The EEG energy distributions at alpha(8—13 Hz), beta(14—30 Hz) and gamma(30—50 Hz) bands were computed by wavelet transform(WT) and compared by the analysis of variance(ANOVA). The timefrequency(TF) analysis indicated that there existed a contralateral dominance of alpha post-movement event-related synchronization(ERS) pattern during the voluntary task, and that the energy of alpha band increased in the ipsilateral area during the stimulated(median nerve of wrist) task. Besides, the contralateral alpha and beta event-related desynchronization(ERD) patterns were observed in both stimulated and imaginary tasks. Another significant difference was found in the mean power values of gamma band(p<0.01)between the imaginary and other tasks. The results show that significant hemispheric differences such as alpha and beta band EEG energy distributions and TF changing phenomena(ERS/ERD) were found between C3 and C4 areas during all of the three patterns. The largest energy distribution was always at the alpha band for each task.
