Skip to main content
Advanced Topics in Sequential Decision Making AA229 / CS239: Winter Quarter 2027

Final Project

Project

The objective of this project is to explore topics and research in sequential decision making in greater depth than is permitted in class. The choice of topic is up to you, but it should be related to the general theme of the course. As part of the project you should:

  • describe an approach (existing or newly developed),
  • apply the approach to a problem of interest (which may or may not be related to aerospace), and
  • analyze the performance of the approach according to a set of metrics.

Your topic may be related to your graduate research or another class project (so long as permission is granted by the instructor of your other class); however, you must make the relationship to your other work clear in your proposal and final paper and describe how you have extended this research for this class. If your project is based of your previous project in AA228/CS238, you must explain in detail how you will be extending your project.

Your final project may be done in any programming language, and you may use any libraries or software available to you. If you have difficulty finding an appropriate topic, feel free to talk with the course staff during office hours or post to Ed.

Collaboration

You may collaborate with up to one other student. If you decide to work with a partner, you will be expected to tackle a more difficult problem and clearly explain the individual contributions of both you and your partner.

Proposal

There is a 20% penalty for each late day. Clearly describe the problem to be solved, the approach to be taken, and how you will measure success. Please reference any prior work that you plan to build upon. The proposal should be limited to two pages. Please submit as a PDF to Canvas.

Project Presentation

You will present your research during class on 3/5 and 3/7. You will be allotted 6 minutes to present and 2 minutes for questions. Time constraints will be enforced. Please submit your presentation to Canvas no less than one hour before class.

Paper

There is a 20% penalty for each late day. Write a 6-10 page paper describing your research. It should follow the typical conference style with an abstract, introduction, etc. State what the problem you are trying to solve, introduce your approach, and review the relevant literature. The experiments should be described in sufficient detail that someone with a reasonable background in the area could reproduce your results. Show your results and discuss the conclusions that can be drawn. You can use the AAAI, NeurIPS, or IEEE paper templates (the LaTeX version looks better, but Word will be accepted)—or you can use the template of another conference of your choice.

If you end up using LaTeX, I like using the biblatex package to manage references, pgfplots for plotting data, and tikz for drawing figures—but you are free to use whatever you want.

Please submit as a PDF to Canvas.

The final grade will be based on the following criteria:\n\n

  • Structure of paper
  • Writing quality
  • Appropriateness of the approach
  • Clarity of drawings, graphs, and tables
  • Appropriateness of abstract
  • Quality of discussion and conclusions
  • Adequacy of references and discussion of prior work

Students who are registered for 4 units are expected to spend 30 additional hours on the research component and will be graded according to this expectation. The paper that is produced should be ready for submission to a peer-reviewed conference.

Peer Review

There is a 20% penalty for each late day. Peer review is an important part of science and engineering. Understanding the review process will make you a better writer. As part of this class, you will be randomly assigned two student papers to peer review in the spirit of a real conference or journal article review. The task of the referee is outlined in this article by Alan Jay Smith. Although you will not be making recommendations for or against publication, you will need to answer the questions listed in Section 4 of Smith’s article.

Each of your reviews will be 1-2 pages each. You will be graded according to the quality and constructiveness of your review. Your reviews will be provided anonymously to the authors, so please do not include your name in your review. If you would like to nominate one or both of the papers for the “best paper award,” please note this in your review.

Please submit your two reviews to Canvas in PDF format. Please submit your completed reviews in the following TWO locations:

  1. Attach the PDF documents as comments on the submissions you were assigned to review. This will allow the authors to directly receive the feedback.
  2. Submit the documents to the “Final Project Peer Reviews” assignment on Canvas. The TAs will look here for your submission when grading. If you do not submit your peer reviews here then you will receive no points for the assignment.

Best Paper Award

An award will be given to the best paper as determined by the peer reviews and the reviews of the course staff. The award will be announced by email. The winner will be taken to lunch at the Stanford Faculty Club by the course staff.