基于受限玻尔兹曼机的疲劳脑电特性分析

Analysis of Fatigue EEG Characteristics Based on Restricted Boltzmann Machines

中文关键词:  稳态视觉诱发电位  疲劳脑电  受限玻尔兹曼机算法  卷积神经网络

英文关键词:SSVEP  fatigue EEG  restricted Boltzmann machines  convolutional neural network

中文摘要:

      为了准确提取和分类视觉疲劳所引起的脑电特征，以此提醒过度用眼的工作人员及时休息，提出了多通道受限玻尔兹曼机算法和卷积神经网络(CNN)算法结合的深度学习混合模型，利用该模型对枕叶区10个通道的脑电信号进行自动提取内在特征和分类。在基于SSVEP的视觉疲劳脑电数据集上进行评估，深度学习混合模型的平均准确率达到88.63%,比传统的特征提取和分类方法高10%。实验结果证明了深度学习混合模型取得的分类效果较好，并且克服了传统手动提取特征方法不全面的不足，对疲劳脑电的研究具有现实的意义。

英文摘要:

      In order to accurately extract and classify the EEG features of visual fatigue,a deep learning hybrid model is proposed to remind the over-the-eye staff to rest in time,which combining the multi-channel restricted Boltzmann machines (MCRBMs) algorithm and the convolutional neural network (CNN) algorithm to automatically extract the intrinsic features and classification of 10 channels of EEG signals in the occipital zone of the cerebral cortex.The model is evaluated on the steady-state visual evoked potential(SSVEP) -based visual fatigue EEG dataset and compared with the traditional feature extraction and classification methods.The average accuracy of the deep learning hybrid model is 88.63% exceeding 10% of traditional models.The experimental results show that the mixed model of deep learning achieves a better classification effect,and overcomes the shortcomings of traditional manual feature extraction method that is incomplete,which has practical significance for the research of fatigue EEG.