Rxivist logo

Mental Workload Classification of Oceanauts from EEG Data Using Multiclass Support Vector Machines

By Xiaoguan Liu, Lu Shi, Cong Ye, Yangyan Li, Jing Wang

Posted 09 Mar 2022
bioRxiv DOI: 10.1101/2022.03.08.483450

In the actual operation task, the workload of the oceanaut is mainly mental workload. For the oceanaut, too high or too low mental workload will significantly reduce work efficiency and even lead to major safety accidents. Classification of mental workload of the oceanaut in operational task research is one of the key problem in operational task research. However, the traditional mechanism modeling method is complex, computational complexity and low accuracy. In this paper, machine learning method got used to the model. Based-on Electroencephalograph (EEG) data collected from the simulation experiment of the operating manipulator, multiclass support vector machine(MSVM) was used to train the samples, and the optimal training sample size was selected. The parameters of the model were optimized by grid search mixed with particle-swarm optimization algorithm (GSPSO) to obtain the optimal classification of oceanauts mental workload. As the research results showed that the method could accurately classify the mental workload of oceanauts.Furthermore, GSPSO-MSVM in the classification of oceanauts mental workload had the advantage over K-nearest neighbors(KNN), BP neural network(BP), random forest(RF) and support vector machines(SVM).

Download data

  • Downloaded 212 times
  • Download rankings, all-time:
    • Site-wide: 178,743
    • In scientific communication and education: 894
  • Year to date:
    • Site-wide: 41,022
  • Since beginning of last month:
    • Site-wide: 159,232

Altmetric data

Downloads over time

Distribution of downloads per paper, site-wide