Machine Learning Technique for Facial Datasets to Detect Examination Fraudulent Activities in the Online Examination: A Systematic Review Approach

Abstract

Online Examination (OE) is the most challenging part of E-Learning (EL) since there is no proper mechanism provided to reduce the OE's fraudulent activities by the students. All the previous research provided different methods to avoid this issue, but those techniques could not be applied due to a few drawbacks of those methods. Recent studies suggested using facial recognition with Machine learning (ML) applications to reduce OE malpractices. This systematic review confirmed that Convolutional Neural Networks (CNN) can be applied to identify students' facial recognition with the help of the CK+ dataset compared to other ML techniques and datasets. Furthermore, future research can be conducted to develop an automated OE proctoring system in real-time. It is noted that this study could not be included a few more recent study results due to no funding. Also, there are no studies found related to this study for comparison of ML techniques and datasets.