基于脑功能连接网络和CNN识别情绪的研究

Research on emotion recognition based on brain functional connectivity network and CNN

基于脑电信号的情绪识别是当前的研究热门,属于人机交互技术的一种。文章设计一种基于脑功能连接网络和CNN识别情绪的模型。研究中采用滑动窗口方法扩充样本量,筛选出与含有最多情绪特征信息的通道,再利用皮尔森相关系数构建脑功能连接网络。不同的网络密度控制下,将脑网络输入至CNN,并识别三类情绪。脑网络密度为30%时,识别精度达到82.3%±1.4%。实验结果表明,所提模型能有效识别积极、中性和消极三类情绪,为情绪识别提供了一种方法。

Emotion recognition based on EEG signals is currently hot research topic and is a type of human-computer interaction technology.A emotion recognition model based on brain functional connectivity networks and CNN is designed.The sliding window method is used to expand the sample size,and screen out the channels with the most emotional feature information,and then the Pearson correlation coefficient is used to construct the brain functional connectivity network.The brain network at different network densities is inputted into the CNN network to recognize three emotions.When the brain network density is 30%,the recognition accuracy can reach 82.3%±1.4%.The experimental results show that the proposed model can effectively identify three emotions:positive,neutral and negative,providing a method for emotion recognition.