Improved Classification of Mental Workload Using One Dimensional Convolutional Neural Network and SMOTE Technique

Mental Workload (MWL) can be measured as the amount of mental strength necessary to complete a work. Mental workload measurement can help in the reduction of mental stress, tension, anxiety, strain, and worry, in daily life or job-specific tasks. This might have a significant impact on tasks like studies/learning and driving etc. In this paper, the classification of mental workload levels through deep learning (one-dimensional convolutional neural networks (1DCNN))has been done. STEW: Simultaneous Task EEG Workload Dataset is used in this study. Electroencephalogram (EEG) signals are captured during the mental task. Based on the subject's rating after completing the mental task, the workload is categorized as low mental workload or high mental workload. To balance and increase thedataset size Synthetic Minority Oversampling Technique (SMOTE) has been used. Different classification measures are used to evaluate the performance of the deep learning model used. Finally, the mental workload classification accuracy of 97.77 was achieved.