Concept

Future Work (Using deep reinforcement learning for personalizing review sessions on e-learning platforms with spaced repetition)

In future work it would be interesting to explore how DR agent would perform in real systems and with the real student interaction data. Also, more complicated LSTMs or GRUs performances could be tested. It would be beneficial to gain a better insight why LSTMs had such a good performance when trained on Supermemo data. Also, testing with the dynamic amount of exercises instead of fixed one would be interesting.

0

1

Updated 2020-11-02

Tags

Data Science