Cross-temporal Mental Workload Identification Based on Transfer Learning Boosting
The effective detection of mental workload in space missions can ensure the efficiency of mission execution and work safety,and predict the performance of astronauts.To detect the cross-temporal mental workload,an experimental paradigm that includes tasks of varying difficulties was designed based on the MATB task and EEG data was collected from 14 subjects at three different time points.Based on the transfer learning Boosting method,a cross-temporal mental workload rec-ognition algorithm was designed based on TrAdaboost by introducing auxiliary data and cross-time classification and recognition were carried out without the participation of target data.In addition,the influence of the optimal segment length and the proportion of auxiliary samples on the recognition effect were explored and decision-making research was conducted based on multiple data samples.The optimal recognition accuracy for mental workload across time reached 74.73%.The results demonstrate that the cross-temporal mental workload classification framework proposed in this study can effectively identify the mental workload.
万方数据
维普
