Here are some things I’ve done recently:

  1. Built a CNN-based automatic essay scoring system and used anchors to investigate the susceptibility of the system to score manipulation by word-perturbation strategies. [poster]
  2. Used parallel feed-forward neural networks to imputate missing item responses in a socioemotional learning survey. [poster]
  3. Implemented a vanilla RNN version of Deep Knowledge Tracing using course interaction covariates to predict student performance in a statistical learning MOOC [poster]
  4. An R package (my first!) to implement the “manager’s regression” as described in this Stack Exchange thread. [github repo]