In which I share my opinions about what makes for a good paper presentation for a graduate seminar course, ECE 750-T5.
We’ll aim for talks of about 30 minutes. During the synchronous meeting time we will start with the presenter’s evaluation and talk about the strengths and weaknesses of each paper and how it can inspire your future work (whether academic or industrial).
Please upload your talks 24 hours before the scheduled class time (whenever that ends up being) and share the URL with the class.
I am going to give you e-mail feedback on content and on presentation style (i.e. how well you spoke and how good your slides are) soon after you record it.
The talk itself
I’m looking for three main parts to these presentations:
What is the paper trying to establish? (your summary)
What are the results? What is the evidence?
What is your evaluation of the paper?
This part should account for most of your presentation (about 25 minutes). People generally lift figures from the paper if they help you make your point, but if they are unsuitable you may have to re-draft them. If the paper has a talk posted you can refer to that for inspiration. However, you should be able to go into more detail than in the paper’s official presentation, because you have more time than the authors did, and because your audience is somewhat heterogeneous and isn’t assumed to know about particular topics in programming languages. e.g. for the week 2 Rust paper I would give examples of what context-sensitivity means. (I see the actual conference presentation more as an ad for actually reading the paper, perhaps with some intuition as to the key idea. This isn’t that.) Try to avoid an info-dump, but do explain the technique (primarily using examples).
Include some discussion of the importance of the paper.
Results and Evidence
In this part, you present the results the paper is claiming, along with the evidence that the authors provide to support their results. You can include some critical evaluation of the evidence here.
Finally, you evaluate the paper. Is it clearly-written? Is the problem important? Do you believe the evidence? Does it provide actionable information or results that future researchers or practitioners can build on?
I recorded this talk this summer. It’s an essay rather than a paper, and it’s not a class presentation, so you can’t exactly copy the form. But I recommend this sort of presentation style. Onward! 2020