An EEG Study of the Student's Confusion Using Deep Learning

The possibility of detecting the confused state of the students would help the instructors validate their teaching methods during the lecture. For this purpose, a wireless electroencephalography (EEG) machine known as the Emotiv EPOC+ was used to record the brain electrical signals of 12 students while presenting visual and audio-visual stimuli to create visual and audio-visual EEG datasets. Following that, convolutional neural network (CNN) and bidirectional long short-term memory (BLSTM) models were trained on both datasets to classify the EEG signals as originating from a confused or non-confused state. The test accuracies achieved by the CNN and BLSTM models trained on the audio-visual dataset were 92.68% and 82%, respectively. In addition, the CNN and BLSTM models achieved test accuracies of 94.25% and 79%, respectively, when trained on the visual dataset.