Get our free extension to see links to code for papers anywhere online!

Chrome logo  Add to Chrome

Firefox logo Add to Firefox


Deep Learning to Predict Student Outcomes

Apr 27, 2019
Byung-Hak Kim


Share this with someone who'll enjoy it:


The increasingly fast development cycle for online course contents, along with the diverse student demographics in each online classroom, make real-time student outcomes prediction an interesting topic for both industrial research and practical needs. In this paper, we tackle the problem of real-time student performance prediction in an on-going course using a domain adaptation framework. This framework is a system trained on labeled student outcome data from previous coursework but is meant to be deployed on another course. In particular, we introduce a GritNet architecture, and develop an unsupervised domain adaptation method to transfer a GritNet trained on a past course to a new course without any student outcome label. Our results for real Udacity student graduation predictions show that the GritNet not only generalizes well from one course to another across different Nanodegree programs, but also enhances real-time predictions explicitly in the first few weeks when accurate predictions are most challenging.

* Accepted as oral presentation to ICLR 2019, AI for Social Good Workshop. arXiv admin note: substantial text overlap with arXiv:1809.06686, arXiv:1804.07405 


   Access Paper Source



Share this with someone who'll enjoy it: