Classifying cognitive workload using eye activity and eeg features in arithmetic tasks

Abstract

The paper presents the results of classification of mental states in a study of the cognitive workload based on arithmetic tasks. Different classification methods were applied using features extracted from eye activity and EEG signal. The paper discusses results of two datasets. The first one covers binary classification discriminating between the cognitive workload condition and the no-task control condition. The second one discriminates between three mental states: the high cognitive workload condition, the low cognitive workload condition and the no-task control condition. The results obtained for the first dataset reached the accuracy of 90% with 6 eye-tracking features as input and an SVMs classifier. The second dataset was classified with the maximum accuracy of 73% due to its complexity.