Week 1 Post (5/20 - 5/26)

This was the first week of my DREU Research Program. I spent the first day getting acquainted with the Professors I would be working with. We were also informed about our responsiblity in contributing to the project - we will be working on developing and testing autograders for different assignments for Duke University’s courses on Coursera. Moreover, we will be learning how to read research papers.

I also spent time with my research team on developing the autograder for the first assignment we were assigned. We started with first understanding how Pytest works, and also how the autograder code works. We started with the development of our code and worked on debugging to resolve the errors we faced.

During this week, I attended UR2PhD’s online courses which focused on what research is, and how we can collaborate with others. I also worked on completing their assignments. I believe this course aligns well with my current timeline of the tasks assigned as part of the research I am working on.

Finally, I read 2 research papers this week that focused on open-ended programming and machine learning autograders:

• “A Fast and Accurate Machine Learning Autograder for the Breakout Assignment” by Evan Zheran Liu, David Yuan, Ahmed Ahmed, Elyse Cornwall, Juliette Woodrow, Kaylee Burns, Allen Nie, Emma Brunskill, Chris Piech, Chelsea Finn

  The paper talks about the development and deployment of of an AI based autograding tool for breakout assignments called Dreamgrader in a CS1 course. The tool helps to significantly reduce the grading labor required for such interactive and time-consuming assignments. Dreamgrader employs a reinforcement learning agent to play and evaluate each student’s game, providing instructors with videos of detected bugs and suggested grades.

• “Novices’ Learning Barriers When Using Code Examples in Open-Ended Programming” by Wengran Wang, Archit Kwatra, James Skripchuk, Neeloy Gomes, Alexandra Milliken, Chris Martens, Tiffany Barnes, Thomas Price.

  Discusses how open-ended assignments can be challenging for students as they might not have the required knowledge for the additional features, or might not know how to integrate knowledge from their classes into projects. They use a tool called “Exam Helper” to provide examples to students to learn how it helps students working on these projects.

Overall, the first week involved getting comfortable with my responsibilities as part of a team, getting started with developing autograders and reading research papers.