Effect on signal magnitude thresholding on detecting student engagement through EEG in various screen size environment

Abstract

In this study, a new method was developed to detect student involvement in the online learning process. This method is based on convolutional neural network (CNN) as a classifier with an emphasis on the preprocessing process combined with a new feature in the form of signal magnitude area (SMA) thresholding. In this study, the data used as training data is a public dataset that emphasizes the decomposition of electroencephalography (EEG) signals into individual signal processing. Twenty subjects were taken to be used as test data, with each subject watching online learning lectures in the field of computer science on three different devices, either with a flat screen, a curved screen or a smartphone screen that is smaller than two standard computer monitors. Based on the study's results, it is known that the change in screen size is inversely proportional to the level of student attention, the smaller the screen, the lower the student's attention. For classification results, the model equipped with SMA thresholding outperformed the standard classifier by 8.33% with a test set of 20 people.