Final projects #

Your final evaluation in Math 2805 will take the form of a final project. My goal is to give you a chance to further explore an aspect of the course in a way that highlights the conversation between mathematics and machine learning. A few projects are outlined below; you should pick one, or discuss alternatives with me.

• Paper: Your work will be summarized in the form of a paper. It should detail the mathematical techniques used in your work as well as a through description of the data-centered experiments you have performed.

  • The mathematics should be well-developed and written with a mathematically-knowledgeable audience in mind, but one that has not necessarily been exposed to the intricacies of machine learning.
  • Computational experiments should be thoroughly described. A reader should be able to replicate your work based on your write-up.
  • Each project has a natural length, so I will not specify a page limit, but the historical average has been ten to twelve pages.
  • As you write your report, you should view your audience as a student who is well-versed in linear algebra and analysis and has only a very basic understanding of machine learning. One of the grading rubrics addresses how effectively you are able to summarize your work and communicate your ideas. The above is the perspective I will take when reading your work.

• Working in groups: You are encouraged to work in groups. Each group should contain no more than four members. You can share notes, computer code, data, and results, but the write-up of your project must be completed entirely on your own.

• Timeline: The projects descriptions and stubs are available on this website. You should:

  • form groups and meet with me with your project proposal by December 4;
  • check in with me as a group once before the start of finals week to update me on your progress; and
  • submit your project directly via email to me by 5PM on December 21.

• Grading rubric: Your project will be worth one hundred points toward your final grade. The points will be assigned based on the following criteria:

  • Clarity and completeness of the exposition of the mathematics involved (35 points). With the audience as described above in mind, the project should detail the mathematical tools used in the project.
  • Clarity and completeness of the exposition of the computational aspects of the project (35 points). A motivated reader should have enough detail to be able to replicate your work.
  • Appropriateness and support for your conclusions (20 points). Are your conclusions well-supported by your work?
  • Your two interim meetings with me (10 points). My expectation is that you and the members of your group are prepared for our meeting. You should be ready to discuss your progress on the project, the steps you have taken so far and the challenges that remain. You should be ready to share any computations that you have performed and be ready to answer questions about the mathematics behind your work.