Abstract

Student performance prediction is a primary goal for educational institutions to improve future performance of students. Early prediction of student’s future performance is critical and it may result interventions to students. Researchers focused in this field and attempt to give a better prediction at early stage. Previously existing algorithms namely fuzzy logic was implemented. However, these algorithms failed to explore this prediction problem. In this paper, a deep learning perspective is proposed for this prediction problem. Deep learning-based algorithm termed as Deep Belief Neural Network (DBNN) with Atom Search Optimization (ASO) optimization categorised students based on their historical performance. DBNN works through a cognitive divergence algorithm with multiple cascaded Restricted Boltzmann Machines (RBM). Initially the proposed model utilized an open-source educational dataset. It is pre-processed in first phase and classified in the second phase. In second phase, DBNN finds quite challenging to optimize the learning rate parameters. By optimization using ASO, the prediction of student level is done automatically. To evaluate the proposed work, performance metrics are measured and provided better result than previous algorithms. Performance of this proposed model outperforms to accuracy level of 90% and reduced error value below 20%. It is concluded that the proposed method of student performance prediction in the academic field is helpful to both students and academicians.

中文翻译：

---

使用基于原子搜索优化的深度信念神经网络的学生成绩预测

摘要

学生成绩预测是教育机构提高学生未来成绩的主要目标。对学生未来表现的早期预测是至关重要的，它可能会导致对学生的干预。研究人员专注于该领域，并试图在早期给出更好的预测。实现了先前存在的算法，即模糊逻辑。然而，这些算法未能探索这个预测问题。在本文中，针对这个预测问题提出了深度学习的观点。基于深度学习的算法被称为具有原子搜索优化 (ASO) 优化的深度信念神经网络 (DBNN)，根据学生的历史表现对学生进行分类。DBNN 通过具有多个级联受限玻尔兹曼机 (RBM) 的认知发散算法工作。最初提出的模型使用了一个开源教育数据集。它在第一阶段进行预处理，在第二阶段进行分类。在第二阶段，DBNN 发现优化学习率参数非常具有挑战性。通过使用 ASO 优化，学生水平的预测是自动完成的。为了评估所提出的工作，性能指标被测量并提供比以前的算法更好的结果。该模型的性能优于 90% 的准确度，并将误差值降低到 20% 以下。得出的结论是，所提出的学术领域学生表现预测方法对学生和院士都有帮助。DBNN 发现优化学习率参数非常具有挑战性。通过使用 ASO 优化，学生水平的预测是自动完成的。为了评估所提出的工作，性能指标被测量并提供比以前的算法更好的结果。该模型的性能优于 90% 的准确度，并将误差值降低到 20% 以下。得出的结论是，所提出的学术领域学生表现预测方法对学生和院士都有帮助。DBNN 发现优化学习率参数非常具有挑战性。通过使用 ASO 优化，学生水平的预测是自动完成的。为了评估所提出的工作，性能指标被测量并提供比以前的算法更好的结果。该模型的性能优于 90% 的准确度，并将误差值降低到 20% 以下。得出的结论是，所提出的学术领域学生表现预测方法对学生和院士都有帮助。