Prerequisites

Grad standing or permission of the instructor. It would be helpful to take a class on introduction of machine learning.

Communication

We will use Brightspace for all announcements, general quesionts about the course, and clarifications about assignments. We also have Piazza, which is mainly for student questions to each other. We strongly encourage all students to participate in discussion, ask, and answer questions in class as well as through Brightspace/Piazza!

Grading

Grades will be based on class participation and the final project.

  • Reading assignments: 20%
  • Paper presentation: 30%
  • Course project: 40%
  • Class participation (Discussion): 10%

Reading Assignments

(Adapted from David Held’s 16-881 Spring 2021 and Zhihao Jia’s 15-849 Spring 2022.)

Starting from the fourth lecture, we will be reading and discussing two to three papers during each class (the paper list will be posted with the class schedule). Every paper review deadline will be the time when our class for the corresponding paper discussion begins. Your paper reviews should consist of at least the following three paragraphs:

  • 1 short paragraph summarizing the first paper, in your own words (do not copy sentences from the paper)
  • 1 short paragraph summarizing the second paper, in your own words (do not copy sentences from the paper)
  • 1 short paragraph on any connections you see between the papers, such as:
    • Compare and contrast
    • How one could apply ideas from one paper to solve the problem in the other paper
    • A new idea that would incorporate results from both papers etc

For each paper review, you will receive 0, 1, or 2 points for each paragraph (full score: 6 points), based on the following rubric:

  • 0: paragraph is missing or any sentence is entirely copied from the paper
  • 1: paragraph misses the point of the paper
  • 2: paragraph demonstrates a clear understanding of the paper(s)

Paper Presentation

You will present two papers in the paper discussion class once a semester (15 mins presentation + 5 mins QA). You should submit your presentation slides (in PDF) to the “Paper Presentation” assignment in Brightspace before our class for your paper presentation begins. Late submission will deduct your presentation score. Here are the detailed rubrics for paper presentation.

  • Summary slides for high-level ideas: 4 points
  • Presentation of problems: 4 points
  • Presentation of method: 4 points
  • Discussion of this paper : 4 points
    • e.g., your own thought such as your interpretation of experiment results, potential improvement, main factors of performance improvement
  • Discussion questions: 2 points
  • Timed correctly: 2 points

Course Project

The course project will be completed by groups of 1-3 students (sign up before week 8). If necessary, we will provide some potential candidate project ideas in the area of machine learning systems (link). Still, you are also more than welcome to bring your own ideas that are related to your research. The Course project will have the following three components:

  • One-page proposal (5%)
  • Final course project presentation (10%)
  • Final course project report (25%)

One-page proposal (Deadline: Oct 22)

For one-page proposal, The team is expected to write the following components:

  • Title / Team members
  • Introduction: What is the key high-level idea behind your project?
  • Problem: What research problems does your project try to address? Clearly define the problem you are solving and include a comprehensive description about the problem.
  • Status quo: What are the existing approaches to addressing the problem? What are the limitations with existing solutions? How do they relate to your approach? How can you improve on what has already been done?
  • Implementation plans: How do you plan to implement and evaluate your ideas? What ML frameworks do you plan to build your solution upon? What datasets do you plan to use?

Each team will submit a one-page proposal in PDF. We understand that any component of the initial proposal can change dramatically over the course of your project. This proposal is to make sure everyone has a concrete idea of what to work on.

Intermediate check-in (Deadline: Oct 29 - Nov 26)

As an intermediate check-in of course project, each team will come to an office hour to discuss your course project or alternatively send us a short update note. The intermediate updates won’t be graded. The goal is to help you resolve any issues you will face in your project and make sure the progress is on track.

Final course project presentation (Deadline: Dec 3 - Dec 5)

Every team will present the final course project in class. Please submit your slides to Brightspace until the deadline. For the format and rubric, please refer to the following.

  • Introduction and overview of your project: 4 points
  • Problem: 3 points
  • Related work: 3 points
  • Method: 4 points
  • Evaluation: 4 points
  • Timed correctly: 2 points (17 mins for presentation, 3 mins for questions)

Note that the deadline depends on the day (Dec 3, or 5) your team is assigned to. For example, if your team is assigned to Dec 3, then the deadline is Dec 3, 4:30pm EST.

Final course project report (Deadline: Dec 12)

For your final project report, the team is expected to write a final report (up to 8 pages without references) that generally follows the format of publication in the MLSys conference with following components:

  • Title / Team members
  • Introduction: Highlight your motivation and contribution; how important is your problem? Why is your work important in this field? What are your main contributions?
  • Problem: What research questions are your project trying to answer? Clearly define the problem you are trying to solve. Include a precise description/background about the problem.
  • Related work: Organize previous work related to your own; what problem do they solve and how do they relate to yours? How did you improve on what has already been done?
  • Overview: Present a system (or architecture) overview of your system using a single figure. Your system can be just one or a few system components rather than an end-to-end stack. Still, you should clearly show what you have done for this course project.
  • Method (or system): Provide a detailed description of your own method or system. It is also fine not to have your own ideas as long as you provide meaningful insights to this field (e.g., exploring the communication/computation tradeoff by evaluating various federated learning algorithms). In addition, describe how you implemented your work. Specifically, which language and library did you use? Clearly distinguish between what you implemented and what you adopted.
  • Evaluation: How did you evaluate your system? What are experimental settings (e.g., hardware)? If any, what datasets did you use? What are evaluation metrics and hyperparameters?
  • Conclusion: Summarize your work in a single paragraph

You are required to submit your codes and final report in the MLSys 2024 format (PDF) to the Brightspace. Again, please make sure that all names of the team members are listed in the final report.

Regrade Policy

If you feel that we have made a mistake in grading your report, please request a regrade on Brightspace and we will consider your request. Please note that regrading may cause your grade to go either up or down.

Academic Integrity and More

Read here.

Listeners outside Purdue

Please feel free to reuse any of these course materials that you find of use in your own courses.

We ask that you retain any copyright notices, and include a written notice indicating the source of any materials you use.