Recursive feature elimination and least square support vector machine approaches to operator functional state feature selection and classification

Operator functional state is closely related to the human mental workload and mental fatigue in a complex human-machine system. This paper is based on the least square support vector machine and recursive feature elimination approaches to determine psychophysiological feature set with highest sensitivity to those human factors. Additionally, a nonlinear least square support vector machine was also trained to objectively recognize the high level of the mental workload and mental fatigue. Analysis demonstrated that the proposed method can effectively eliminate the redundant features without decreasing classification accuracy.