Paper Title
Students Engagement Level in Online Learning

Abstract
Online learning has replaced in-person instruction as a result of the COVID-19 epidemic. Although this change has benefits such removing time and location restrictions from the learning process and allowing education to take place at anytime, it can be difficult to gauge student interest in online learning because of the lack of interaction. One important element impacting the entire learning process is student engagement, which is the active participation of students in the educational process. In order to overcome this difficulty, this study suggests a methodology that applies bagging (bootstrap aggregating) ensemble learning to 1-dimensional residual networks (1D ResNet), 1-dimensional convolutional neural networks (1D CNN), and hybrid ensemble deep learning models. Our results using the DAiSEE dataset show that the bagging ensemble of the 1D CNN model outperforms the individual model by 3.25%, achieving 93.25% accuracy. The deep learning ensemble bagging achieves 93.75%, which is 3.5% better than the unique 1D ResNet model. Furthermore, the hybrid ensemble bagging attains the greatest accuracy of 94.25%, which is 0.5% better than the 1D ResNet model and 1% better than the 1D CNN model. Keywords - Online learnig, Students Engagement, COVID- 19Pandemic, In person instruction, Active partcipation, Bagging, Bootstrap aggregating, Ensemble learning, 1D ResNet, 1D CNN, Hybrid Ensemble, DAiSEE dataset, Convolution Neural Net- works, Engagement detection, Educational technology, Remote educational.