基于多模态生理信号的熟练度识别方法研究

Research on Proficiency Recognition Method Based on Multi-modal Physiological Signals

基于生理信号识别用户情绪已成为当前研究热点,论文在为雷达训练模拟器中增加定位用户知识盲区的熟练度分析任务背景下,展开基于生理信号识别用户熟练度的方法研究。文中给出了基于脑电信号、眼动信号等多模态数据的两分类熟练度数据集自采集的实验范式,并在该数据集上进行了多种熟练度识别实验,实验结果表明,论文提出的熟练度数据集的数据丰富性足够支撑经典的深度学习模型,此外,综合多种生理信号识别用户熟练度模型平均识别精度均高于基于单模态模型的平均识别精度,准确率最高的基于多模态识别模型其平均精度可达84%。

Emotion recognition based on physiological signals has become a research hotspot.Under the background of increas-ing the proficiency analysis task of locating user knowledge blind spots in radar training simulator,this paper studies the method of identifying user proficiency based on physiological signals.The experimental paradigm of self-acquisition of two-class proficiency data sets based on multi-modal data such as EEG signals and eye movement signals is given,and a variety of proficiency recogni-tion experiments are carried out on this data set.It is concluded that the data richness of the proficiency data set proposed in this pa-per is sufficient to support the classical deep learning model.In addition,the average recognition accuracy of the proficiency model based on multiple physiological signals is higher than that based on the single-modal model.The average recognition accuracy of the multi-modal recognition model with the highest accuracy can reach 84%.