Overview

A key element of this course is the final project. You will work on your project in groups of 2 or 3. Groups are encouraged but not required to involve students from different departments. If your project involves human subjects research, you may need approval from the IRB.

Schedule

• Proposal due: before class Wednesday, March 20, 2024. Send an email to Shreyas at sgandlur@princeton.edu. Please send one email per group and include the names of all team members.
• Instructor feedback on proposal: before class Wednesday, March 27, 2024.
• Final project due: Tuesday, May 7, 2024 5pm ET (Dean’s Date).

Expectations

We will evaluate your project based on the creativity of your ideas, the thoroughness of your investigation, the quality of your writing, as well as connecting your project to related work. It’s basically like a research paper, except that you’re probably tackling a more modest problem, and we won’t penalize you if you accidentally missed prior work. See “Grading Rubric” below for more details.

Code submission

You should submit reproducible code along with your paper; we need to be able to run your code and verify your results. Learning to carry out reproducible research is an important learning goal of the class. Other than that, you won’t be graded based on your code, but we’ll consult it if we can’t understand something in your paper or if we want to get a deeper understanding of how your system works. This is what happens in peer review (at least in the ideal case; in practice, unfortunately, often authors don’t make code available and reviewers are overburdened and don’t have time to look at it).

Making your final project submission public

Recall that in addition to learning goals, this course also has the goal of enabling publishable student-led research. After the course is over, we (Arvind and/or Matt) may, at our discretion, offer to help you develop your project into a paper to submit for peer review. You are welcome to work with us to produce a publishable paper, or do so on your own, or choose to leave it as a class project. It is completely up to you.

Either way, we ask that you make your class project submission public in a way that can be cited by others. We are happy to do this for you by default. The reason for this requirement is that your work may contain new research insights; unless it is public, it will be difficult for others (including us) to build on those insights — i.e. we won’t be able to cite your paper for the insight nor, of course, claim credit for it as our own. Except for potentially citing your paper, we won’t publicize it (e.g. blog or tweet about it) unless you tell us that’s ok.

If you have a reason to be exempted from this policy, please talk to us.

Grading rubric

Problem selection (10 points)

• 10 pts – Interesting, important, and clearly defined problem
• 5 pts – Problem statement or importance is unclear

Creativity (20 points)

• 16-20 pts – Research design and techniques went beyond what was shown in class. Taught the reader something new and surprising.
• 11-15 pts – Used designs and techniques learned in class but showed some originality in their application and/or dataset choices.
• 5-10 pts – Used techniques learned in class in a straightforward way.

Correctness and thoroughness of investigation (20 points)

• 16-20 pts – Thorough investigation with no obvious holes.
• 11-15 pts – Covered a few bases but left some obvious holes.
• 5-10 pts – Left gaping holes in the investigation / incorrect data analysis etc.

Quality of writing (30 points)

This includes linguistic clarity, exposition of technical concepts, logical structure, justification of claims, explanation of background concepts, and discussion of results.

• 27-30 pts – The writing makes the work more interesting and accessible
• 21-26 pts – The writing does not add too nor detract from the research itself
• 10-20pts – The writing interferes with the reader’s ability to learn from work

Related work (20 points)

• 16-20 pts – Identified and discussed all categories of related work. Within each category, identified works that are truly state of the art. Made clear how the paper advances the state of the art compared to the papers cited.
• 11-15 pts – Identified and discussed most major categories of related work. Discussed works that are state of the art but missed a few major papers.
• 5-10 pts – Failed to identify major categories of related work. Works discussed are not state of the art. Unclear how the paper builds off of previous work.