Deep Learning Recommendation System for Course Learning Outcomes Assessment

Raed Abu Zitar, Ammar EL-Hassan and Oraib AL-Sahlee

This paper is improvising a system that utilizes the existing capabilities of a courses learning outcomes assessment application called ADAMS (Accreditation Data Analysis and Management System). ADAMS can provide us with the data regarding the weaknesses and strength in specific learning outcomes or group of learning outcomes of a study course or group of courses in a study plan. The assessment will be based on given rubrics for every course. The outcomes of ADAMS will be processed to extract best features and then later to be used in training a deep learning neural network that can give the advice about the best approach to mitigate the weaknesses explored by ADAMS in the learning process. Two category of deep learning networks will be tried; the classical MLP neural network and the Convolutional Neural Networks (CNN). The training details are presented and the results are analyzed and compared.The goal is to have an intelligent automated system that gives advice, directions, and recommendations on how to amend the teaching process to maximize the degree of an academic program learning outcomes achievement. The major contribution is to use deep learning technique in training a feed forward neural network to implement the remedial action. The idea here is to treat the outcomes of ADAMS as input training vectors that go under classification and recognition for deep data analysis. We use supervised learning in the MLP accompanied and Convolutional training in CNN. The capabilities of a connectionist approaches such as the MLP and CNN are used for huge data analysis. The system once trained can be used in recommendation purposes for remedial actions for data that was not used before in the training process. A promising approach for data analysis with application in education is purposed, comparisons were made and conclusions were reached.

Volume 11 | 10-Special Issue

Pages: 1478-1491

DOI: 10.5373/JARDCS/V11SP10/20192993