Application of Mental Fatigue Classification in Cross Task Paradigm

Zhengxing Xing,Enzeng Dong,Jigang Tong,Zhe Sun,Feng Duan

DOI: https://doi.org/10.1109/ICMA54519.2022.9855978

2022-01-01

Abstract:Mental fatigue refers to the state that individuals feel uncomfortable after high-intensity or long-term continuous mental activities. Mental fatigue will affect the efficiency of learning and work, and even lead to accidents and loss of personnel and property. Although previous studies have achieved effective single task classification, the efficiency of collecting corresponding training data for different work tasks is often low. Here, we use the Cross task paradigm model to test the mental fatigue data, so as to improve the applicability of the mental fatigue detection model. Eight healthy subjects performed n-back task and simple mental arithmetic (MA) task respectively. The features of the collected EEG signals were extracted by fuzzy entropy algorithm as the feature vector of the experiment, and then input into support vector machines (SVM) to evaluate its classification performance in a Cross task paradigm manner. Our framework achieves an average classification accuracy of 84.50%. At the same time, a variety of feature extraction methods and classifiers are used to verify our conclusion that the brain fatigue classification across task paradigm is feasible.