基于双分支卷积神经网络的运动想象分类
首发时间:2020-06-03
摘要:脑电信号是脑神经细胞电生理活动在大脑皮层或头皮表面的总体反映。脑电信号中包含了大量的生理与疾病信息,常被用于脑机接口研究中风患者的康复治疗等,而解决这些问题的关键是如何提高脑电信号的识别精度。为此,本文提出一种端到端的双分支卷积神经网络对运动想象脑电信号进行特征提取和分类. 首先,利用双分支卷积神经网络自动提取原始脑电信号的多种特征;然后,使用残差模块进行特征的融合;最后,利用全连接网络层进行脑电信号的分类.实验在BCI竞赛III 4a数据集中取得了很好的效果,证明该模型可以有效地分类脑电信号且无需手动设计特征,具有较高的应用价值.
For information in English, please click here
Motor imagery based on double branches convolutional neural network
Abstract:EEG signals are the overall reflection of brain nerve cell electrophysiological activities on the cerebral cortex or scalp surface. EEG signals contain a large amount of physiological and disease information, and are often used in brain-computer interface research and rehabilitation treatment of stroke patients. The key to solving these problems is how to improve the accuracy of EEG signal recognition. Hence, this paper proposes a end-to-end double branch convolutional neural network for feature extraction and classification of motor imaging EEG signals. First, the double branch convolutional neural network is used to automatically extract multiple features of the original EEG signals; Then, the residual module is used for feature fusion; finally, the fully connected network layer is used to classify the EEG signals. The experiment has achieved good results in the BCI competition III 4a dataset, the proposed model has higher application value for it classifies motor imagery EEG signals efficiently without manual feature extraction when spatial information is insufficient.
Keywords: motor imagery convolutional neural network EEG signal end-to-end
基金:
引用
No.****
同行评议
勘误表
基于双分支卷积神经网络的运动想象分类
评论
全部评论0/1000