Paper Presentation

Due: Varies based on presentation slot.

Through your paper presentation, you will form a group of 3 to present a 25-minute deep dive into one of the papers from a set of suggested papers.

We will send you a Sign-Up Google Sheet, including the list of papers and presentation dates. Sign ups for the paper presentation will be first come, first serve.

When preparing your presentation, keep your intended audience in mind: you may assume that your fellow students have given the paper a light read, but do not assume that they have the same intimate familiarity with the work that you do. Additionally, it would be fruitful to give a brief overview and background of the broad class of machine learning techniques to which the paper belongs, if you anticipate your classmates may not be familiar with the concept (a good rule of thumb for this is, “in preparing my presentation, did I need to research additional background on this to feel comfortable with the paper?”).

For a 25 minute presentation you should aim to have no more than 20 slides.

We ask you to explicitly include the following template slides in your presentation:

  • Problem Setup: What problem is the paper trying to solve?
  • Motivation: What motivated the work? Who should care about it, and what applications do you foresee for this work?
  • Related Work: What is a short summary of existing work? How do they solve the problem?
  • Technical Detail: Describe the paper's core technical or practical contributions - explain why the paper is new to the literature and what it allows practitioners to do, which could not be done before.
  • Limitation and Future Work: What are the limitations of the work? What would you do to improve the work if you had infinite time?
Rubric
CriterionWeight
Clearly motivates and defines the problem.15%
If applicable, presents requisite background on the techniques implemented in the paper.15%
Clearly explains at least one the main ideas from the paper.20%
Creates a compelling visual representation of the main idea.5%
Explains the experiments and/or results. Discussion of what the machine learning community stands to gain from this work.15%
Provide a conclusion which is clear-eyed about both the strengths and limitations of the paper.10%
Finish under time (25 minutes).10%
Includes the template slides.10%