A COMPARISON OF THE UTILITY OF VARIOUS NEURAL NETWORK MODELS IN IMPROVING EDUCATION AND DESIGNING LEARNING PATHS

Neural networks are self-improving computational systems used for prediction. Artificial Neural Networks (ANNs) computationally process information in a way that is similar to the human brain. There are myriad existing prediction models that can be used for various purposes, and this report aims to identify the predictive model most useful in the realm of education. It is taken into account that different students require different types of media to learn most effectively. In this project, different predictive models are compared to one another in their effectiveness specifically in predicting learning performance in certain subjects. Additionally, various activations (i.e., tanh, sigmoid, identity) and filtering methods (i.e., content-based, collaborative, and hybrid) are compared. These findings are then used to describe a possible recommendation algorithm to improve education by creating learning paths.