Learning Analytics for Adaptive and Self-Improving Learning Environments

EDIC candidacy exam
Exam president: Prof. Volkan Cevher
Thesis advisor: Prof. Pierre Dillenbourg
Co-examiner: Prof. Martin Jaggi

Abstract
Today, Massive Open Online Courses attract millions of students worldwide. In particular, EPFL MOOCs have reached more than a million participants. These large number reveal the hype of digital education and brings new opportunities for educational research by giving it the tools of the data sciences. For instance the implementation of large scale peer grading in MOOCs shows the convergence between pedagogical methods and the digital revolution. This project explores these new opportunities for developing adaptive and self-improving learning environments.

Background papers
Performance Factor Analysis - A new Alternative to Knowledge Tracing, by Pavlik P., et al.
Faster Teaching via POMDP Planning, by Rafferty A., et al.
Deep Knowledge Tracing, by PiechC., et al.