One time during the semester it will be your responsibility to read an assigned research paper and prepare a 10 minute presentation for the entire class. The goal of this assignment is that we, as a class, will be exposed to more of the reinforcement learning research field without the need to greatly add to the amount each individual must read. Of course, the selected readings are influential and engaging papers and so you may well want to read more than just your assigned paper.

You may sign-up for a reading and presentation slot here.

Note that in each class period we will have up to two speakers. To ensure that presentations go smoothly, please submit a draft of your presentation slides two days in advance so that I can provide feedback. I also ask that you prepare your presentation with Google Slides so that it is easy to start your presentation without the need to switch laptops in class.

Questions to answer:
  1. What are the main contributions of the paper? What is possible now that wasn't before? What was learned from this research?
  2. How does the paper support its stated contributions? Experiments (and if so which ones)? Theoretical analysis?
  3. What were the primary challenges?
  4. What are the limitations?
  5. What next steps do the author's propose? What next steps would YOU propose?
  6. Why does this paper matter?
  7. Additional critiques or praise for the paper?
Rubric: Tips for a good presentation:
  1. Remember that not everyone has read the paper. Even if they have, they likely have not thought about it as deeply as you.
  2. Assigned readings pertain to the week's reading and lecture topic but likely include additional material that some in the class may be unfamiliar with. Be prepared to provide additional background in your presentation.
  3. It is not important that the class understands everything in the paper. In 10 minutes, your goal should be to convey the most important points, contributions, and methods of the paper.
  4. A useful guide for timing is to plan to spend 1 minute / slide. If you have 20 slides, it is likely you will have to rush to finish.
  5. When explaining experimental results, make sure to explain what each axis represents on plots.