Identification of Students' Confusion in Classes from EEG Signals Using Convolution Neural Network

Abstract

For a student, classes are vital factors for gaining knowledge. The lectures may be online or offline, but getting knowledge without confusion is a major issue. The confusion labels can be measured from the electroencephalography signals and the confusion can be solved after knowing that students are suffering from confusion. Different machine learning approaches were implemented on electroencephalography signals to identify the suffering of students from confusion. The performance of traditional machine learning approaches in predicting confusion status is found as poor. In this paper, the one-dimensional convolution neural network is implemented on the electroencephalography signals to detect confusion of the students at the time of watching video classes. Students' attention, mediation, electroencephalography signals, delta, theta, alpha1, alpha2, beta1, beta2, gamma1 and gamma2 are taken into consideration to train a one-dimensional convolution neural network classifier. The one-dimensional convolution neural network approach has achieved better accuracy in detecting the confusion of the students. Besides finding confusion labels of students, the experiment is performed when understandable classes are creating confusion and the difficult classes are understandable for the students. This second experiment is also performed on electroencephalography signals of students and after identification of confusion status, the improvement of students' deficiencies can be possible. For future work, more data and different aspects of the students can be taken into consideration for identifying confusion and different obstacles respectively which helps to improve in achieving perfect knowledge from the classes.