Deep learning-based mental task classification using a muse 2 EEG headset

One of the most significant qualities of the human brain is cognition, which incorporates both attention and recall. Attention is the initial process used when dealing with sensory inputs, prior to cognition, which acts to identify the nature of the stimuli as it impacts the individual's sensory system to select which details will be retained, processed, and recognised. Based on the increased use of e-learning, it has thus become necessary to determine the extent of student attention under the learning circumstances by looking at the state of the brain during this process. This can be achieved by reading brain signals using modern and advanced technologies with sensors that detect brain signals such as electroencephalography (EEG). The attention spans of students and their situational interest during learning have been studied for many years with respect to classroom learning; however, e-learning is becoming a highly popular method of learning, and the goal of this study is thus to use brain signals to measure learner attention levels in e-learning settings as compared to those seen in traditional classroom learning. The standard approach records EEG signals and their frequency bands across several subjects as they listen to a lecture in a classroom setting, characterising students' mental states as attentive or no attentive by means of machine learning classification models and algorithms. In this work, the best result were achieved with training using the Random Forest Classifier on the frequency band data set, which gave an accuracy of 91.4%, and by training using the Gated Recurrent Units neural network on the raw signal data set, which gave an accuracy of 96.4%.