Deep Neural Networks based Personalized Learning Systems for Adaptive Education

In recent scenario, Artificial Intelligence (AI) is significantly transforming higher education by enhancing student assessment through enabling personalized learning experiences and streamlining administrative operations among institutions for improved outcomes. Traditional approaches for personalized learning system had faced challenges like limited adaptability and inadequate handling of diverse learner behaviours and preferences. Therefore, this research proposes Deep Neural Networks (DNN) for personalized learning systems in adaptive education. Initially, data is collected from Open University Learning Analytics Dataset (OULAD) dataset which consists of student demographics, assessments, interactions and performance records among different courses. Then, preprocessing is done by using one-hot encoding and min-max normalization which converts categorical data and scales features into a uniform numerical range. After that, features are extracted by using Principal Component Analysis (PCA) that reduces dimensionality, removes redundancy and retains most significant variance for efficient learning. Finally, DNN is used for modeling personalized learning paths by capturing complex patterns and predicting student performance with high accuracy. The proposed DNN achieved better results in terms of accuracy (0.93), precision (0.91), recall (0.90) when compared with existing Fully Connected Neural Network (FCNN) model.