CS-E407517 - Special Course in Machine Learning, Data Science and Artificial Intelligence: Seminar on NLP Research, Lectures, 4.9.2023-27.11.2023
This course space end date is set to 27.11.2023 Search Courses: CS-E407517
Topic outline
-
Format of the seminars
The seminars are held in person. Two papers are discussed in each session with a 10-minute break between the papers. The format of discussion of one paper:- Presentation (max 20 minutes).
- Opponent speech (max 5 minutes).
- Discussion moderated by the teachers, first in groups and then generally.
Grading
The grade will be based on your presentation (40%), your service as the opponent (20%), and your writing assignments (40%). The grading scale is 1-5, and the final grade will be the weighted average of the subcomponents.- The writing assignments will be graded on a scale of 0 (fail)/1 (ok)/2 (excellent)
- You must attend at least 10 of 12 sessions to pass the course. Well-justified failure to attend can be compensated by writing assignments graded as excellent. However, this must be agreed on separately with the course staff.
- You can earn up to (N_papers * 2) points for the writing assignments. However, (N_papers * 1) points are enough for a full score given sufficient attendance.
- You are also welcome to discuss the papers on Slack. Active participation on Slack can earn you up to 5 extra points = .5 of a grade.
The presentation and opponent talk will be graded according to the following criteria:- Structure and organization of the presentation
- Understanding of scientific content
- Clarity and accessibility of presentation
- Style and delivery
- Ability to answer questions (presentation only)
Presentation guidelines
Please read these instructions carefully because the quality of your presentation will affect your grade.- The duration of your presentation should be at most 20 minutes. We will stop the presentations after 20 minutes.
- Try to explain the main idea of the paper instead of
copy-pasting paragraphs from the paper page by page. You do not have to
use all 20 minutes.
- Spend minimum time on well-known concepts (such as transformer or cross-entropy loss) or concepts discussed earlier in the course. If you are not sure whether a specific topic should be covered in your presentation, you can consult the teachers in the course chat.
- Do not overload your slides with text. It is better to present figures/illustrations.
- Please be selective in presenting experimental results. You do not have to present all of them but the most important ones.
Guidelines for the opponent
Please try to make a short review of the discussed paper, for example, summarize the main strengths and possible shortcomings. Examples of reviews can be found on OpenReview. Please focus on your opinions and thoughts instead of just summarizing the main contents of the paper, as the audience has just heard the 20-minute presentation of the article. The duration of the opponent's talk should be 5 minutes.Writing assignments
To prepare for the meetings, you need to write down the following before the meeting for each paper:1) Main points and facts2) Questions3) Opinions, further comments, etc.As a general rule of thumb, you are expected to write down approximately three bullets in each category for each paper. Obviously, if you don't have any questions, you can give correspondingly more weight to the other categories or elaborate on some other bullets. The idea is not to spend overly much time on this, but just to briefly recap your thoughts during and after reading the chapter. During the meeting, we will take turns, and participants can be asked to present some points from their own lists from each category to others. Return your completed task for each week as a single PDF on the "Assignments" page. The writing task is due 15 min before the meeting starts.