USE OF ARTIFICIAL NEURAL NETWORKS IN THE SELECTION OF EDUCATIONAL CONTENT ON AN E-LEARNING PORTAL

An individualized learning path is a new solution that has been introduced in didactic materials made available to students on the educational portals of our universities. This approach stems from the fact that students participating in the educational process, on the one hand, have different needs and expectations, and on the other hand, different possibilities and limitations in acquiring knowledge. The conducted research allowed recognition of the differences and the selection of the most appropriate of the three learning paths offered. Indication of the learning path is made by means of a classification system, for the construction of which artificial neural networks were used. On the basis of the tests carried out, it was shown that when selecting the learning path, the best results are provided by a multi-layer network, with one hidden layer that contains 9 neurons. The network was taught in 50 epochs, the activation functions of the hidden and output layer neurons were hyperbolic tangent and linear function respectively. Over 98 percent correctness was achieved in the classification of new students starting the education process. The innovation of the proposed solution is to demonstrate in practice the possibilities of individualizing the student education process and thus its adaptation to the educational needs and competency gaps of participants in this process.