Eeg And Eye Movement Features And Classifier Fusion Method For Spatial Cognitive Assessment

Spatial cognitive ability can measure human body’s perception of other places and understanding of spatial motion.With the growth of age,people’s spatial cognitive ability will gradually decline,so it is of great significance to improve people’s spatial cognitive ability.As important indicators to measure spatial cognitive ability,eye movement characteristics and electroencephalogram characteristics need to be constructed to evaluate the changes of electroencephalogram and electroencephalogram characteristics after spatial cognitive training.The specific research content is as follows.Firstly,a multi-scale convolution co-attention noise reduction fusion model is proposed.The multi-scale convolution common attention noise reduction fusion model(MSCAMFDFM)is based on the principle that the initial electroencephalogram feature and the initial eye movement feature are output new multi-mode features through the common attention noise reduction module respectively,and then the new two mode features are passed through the multi-mode fusion module to realize high-order feature fusion.As the input of the network,the high-order features are used for the binary classification before and after training.The classification accuracy of each electroencephalogram frequency band can determine whether the features after the fusion of the frequency band change significantly before and after training.Secondly,a multi-modal feature fusion and multi-classifier fusion model is proposed.That is,the output of MSCAMFDFM is fused with the output of other classifiers as input to a second-level classifier,which outputs the final classification result.The SRMSCA-Hist model was adopted in this study,whose first-level classifiers were SVM,Radom Forest and MSCAMFDFM,and the second-level classifiers were Hist Gradient Boosting.Thirdly,the spatial cognitive training scene of this study was the scene game in which the virtual UAV looked for the target according to the fixed route,and the spatial cognitive test scene was the virtual city roaming.The subjects needed to go through 35 days of training and testing,which were divided into 5 rounds,and the game routes were different each time.By collecting electroencephalogram and eye movement data of subjects before and after training,the changes of electroencephalogram and eye movement characteristics can be compared to realize spatial cognitive evaluation.Finally,MSCAMFDFM and SRMSCA-Hist proposed in this paper analyzed the classification results before and after training electroencephalogram features and eye movement features from the perspective of 7 electroencephalogram frequency bands,and the classification accuracy of the two algorithms in each frequency band was more than93%.The results showed that the electroencephalogram characteristics of each frequency band changed significantly before and after the training,and the subjects showed significant improvement after the spatial cognitive training.