Deep Frustration Severity Network for the Prediction of Declined Students' Cognitive Skills

Abstract

Prediction of declined Cognitive Skills (CS) is essential not only for students and tutors but also for policymakers to make appropriate policies for the effective educational systems, such as the evaluation of admission criteria, teaching method, class activities, and examination systems. Articles are saturated with the number of findings which statistically correlate students' weak CS with the influence of frustration severity. Prior approaches have predicted declined CS using biological factors and study-related attributes of a student. Nevertheless these studies are insufficient to predict students' CS using the adverse influence of frustration severity. In this work, we have proposed Deep Frustration Severity Network and the loopholes in the existing approaches are the primary source of inspiration for this network. The proposed network has four outer layers for frustration severity while 34 inner layers for the CS outcomes. During outer and inner iterations student's CS is iteratively estimated while considering the adverse influence of frustration severity layers. Eventually the deep network predicts declined students' CS with iterative calculation process. We have validated the network on a students' score dataset. It achieved significant results in terms of state-of-the-art evaluation measures.