hide
Free keywords:
-
Abstract:
Constructing a reliable and stable emotion recognition system is a critical but challenging issue for realizing an intelligent human-machine interaction. In this study, we contribute a novel channel-frequency convolutional neural network (CFCNN), combined with recurrence quantification analysis (RQA), for the robust recognition of electroencephalogram (EEG) signals collected from different emotion states. We employ movie clips as the stimuli to induce happiness, sadness, and fear emotions and simultaneously measure the corresponding EEG signals. Then the entropy measures, obtained from the RQA operation on EEG signals of different frequency bands, are fed into the novel CFCNN. The results indicate that our system can provide a high emotion recognition accuracy of 92.24% and a relatively excellent stability as well as a satisfactory Kappa value of 0.884, rendering our system particularly useful for the emotion recognition task. Meanwhile, we compare the performance of the entropy measures, extracted from each frequency band, in distinguishing the three emotion states. We mainly find that emotional features extracted from the gamma band present a considerably higher classification accuracy of 90.51% and a Kappa value of 0.858, proving the high relation between emotional process and gamma frequency band.
Incorporating emotion interaction into the human-machine interaction (HMI) process is of vital importance for building a much more intelligent HMI system. To realize intelligent emotion interaction, one of the most primary requisites is to design a reliable emotion recognition system with high accuracy and robustness. Electroencephalogram (EEG)-based emotion classification systems have been widely used on account of their high accuracy and objective evaluation. We here propose a novel emotion recognition system which combines recurrence quantification analysis (RQA) with channel-frequency convolutional neural network (CFCNN), based on the successful applications of RQA in analyzing nonlinear EEG signals and the remarkable identification ability of convolutional neural network (CNN). The results suggest that our system has a reliable and stable emotion identification capability. In addition, we feed the RQA-based features extracted from each of the five frequency bands into CFCNN to discuss their respective effectiveness and find the critical frequency band which is closely linked to emotional process. The result indicates that emotional arousal is in particular relevant to gamma frequency band activities
