[Mandatory] Written Paper Critiques

[Due] Before each class.
(e.g., the deadline for Tuesday's class papers is 9:59:59 AM on Tuesday.)

The required readings are posted on the course schedule, and you need to write a critique for one out of the 2-3 papers per class. Keep in mind that this is not light pleasure reading. You may need to read a few more relevant papers to understand the reading materials. After you are used to this reading, it may take 2-3 hours to read each paper.

Your critique for each paper MUST have the following fields:

• Summary of the paper (a paragraph)
• Contributions (typically 2-3 bullets for each paper)
• Strengths and weaknesses (2-3 bullets for each)
• Your opinions (on the strengths and weaknesses)

Please make your best effort to write concise and precise statements. For example:

• In the summary, it is better to write statements that are more summarizing than descriptive. A common mistake is to write "this paper proposed a methodology..." This isn't valuable for you later on; you can write "the key idea (or intuition) of this paper is XYZ."
• Contributions / strengths / weaknesses. You need to extract the essence—what the authors really want us to remember about the paper. For example, the fact that the research is well-evaluated is not a contribution; it's necessary for the authors to convey their ideas. Similarly, bad writing is not a weakness for those selected papers. You can mention them in the "opinions" field.

In your critiques, I will look for evidence that you thought about the problem discussed in the paper; you incorporated personal experience, you looked for how the security issues of machine learning discussed in the paper are addressed today, and so on. Think critically about the paper and do not accept the authors' opinions as-is. It is the authors' job to convince you!

Here, I also share a sample critique of mine for reference [example].

Grading Policy

• 0 point: No submission before the deadline / a simple paraphrase of the abstract and introduction.
• 1 point: You made an effort to understand the papers / you also carefully think about the pros and cons of the papers.
• 2 points: You make an insightful comment.

Submission Instructions

Upload your critique to Canvas.

[Mandatory] In-Class Paper Presentation and Discussion

[Note on Paper Selection]

• Choose one paper that is relevant to the course topic covered on the day you are presenting (see schedule).
• The paper must be from one of the following venues:
  • Security Conferences: USENIX Security, ACM CCS, IEEE S&P Symposium, and NDSS
  • ML Conferences: NeurIPS, ICML and ICLR
• The paper must NOT be one of the required readings already listed in the course schedule.
• You are expected to independently find a paper from the venues above.

You only need to sign up once during the term. During your assigned class, you are expected to present the paper as if you are the author and lead the in-class discussion.

Please sign-up using this Google Sheet [link] (you must use your ONID to access this sheet). A maximum of 3 students can sign-up for each presentation slot. You MUST sign-up at least one week before your chosen presentation date.

You MUST meet Sanghyun ONCE—3-4 days before your presentation. This meeting is for reviewing and revising your presentation materials. Make sure to have a draft presentation and 1-2 discussion topics.

Your presentation and discussion will be tentatively 35-45 minutes.

• 25 min: Paper presentation.
• 10-25 min: In-class paper discussion.

Grading Policy

• 0%: No sign-up or no presentation.
• 5%: Sign-up; but the lack of quality in your presentation.
• 8%: Sign-up; you present a paper and lead the discussion.
• 10%: You make an outstanding presentation and discussion.
• Miss the meeting with Sanghyun: If you miss the meeting, your final score will decrease by 2%.