This page is a preview og should only be accessed through the link in the Kursusbasen administration page. We can see you are not accessing it from there, so we fear there has been an error.In almost all cases the link you need would be the link to the official public version.Not this version that has not been approved yet.Please contact the person that gave you the link (or the editor that published the link on a webpage) so it can be corrected.

02971 Latest developments in machine learning at NeurIPS


Course information
Latest developments in machine learning at NeurIPS
Ph.D., Professionally focussed course
The course runs full time in the days from Dec. 7, 2020 to December 11, 2020.
The course runs in small groups that are expected to meet either virtually or physically on campus.
Group work where students read and discuss papers. Discussions should be contrasted with publically available peer reviews for the read papers.
[The course is not following DTUs normal Schedule]
Evaluation of exercises/reports
The student is evaluated based on an individual report.
pass / not pass , internal examiner
Søren Hauberg , Ph. (+45) 4525 3899 , sohau@dtu.dk
Aasa Feragen-Hauberg , afhar@dtu.dk
Ole Winther , Lyngby Campus, Building 321, Ph. (+45) 4525 3895 , olwi@dtu.dk
Lars Kai Hansen , Lyngby Campus, Building 321, Ph. (+45) 4525 3889 , lkai@dtu.dk
Morten Mørup , Ph. (+45) 4525 3900 , mmor@dtu.dk
01 Department of Applied Mathematics and Computer Science
At the department
December 7, 2020
General course objectives
This course revolves around the latest developments in machine learning research presented at the virtual NeurIPS conference. The course takes place on the same days as the NeurIPS conference, and is structured as a collection of multiple concentrated reading groups running in parallel. Students will be split into smaller groups (approx. 3-6 students per group) according to topics of interest. Each group will then read and discuss a collection of papers within the topic of interest.

A side-goal goal of this course is to increase the scientific interaction for PhD students due to not being able to physically attend the NeurIPS conference. The course will be announced as part of the ELLIS Copenhagen Unit to engage with students from both KU and ITU, adding to the interactive and networking potential of the course.
Learning objectives
A student who has met the objectives of the course will be able to:
  • Read and discuss with peers a NeurIPS 2020 machine learning paper
  • Explain the scientific content of a machine learning content to other machine learning researchers and PhD students
  • Write an easily accessible explanation of a NeurIPS 2020 paper, e.g. in the form of a blog.
  • Contrast paper and discussions with publically available peer reviews.
  • Discuss strengths, weaknesses to NeurIPS papers.
  • Explain key points of NeurIPS 2020 tutorials to peers.
  • Reflect upon possible impacts of NeurIPS 2020 papers
  • Give a short oral presentation of a paper to peers.
Each group will collectively read and discuss at least 5 papers from the main conference and associated workshops. This should also include the paper reviews, which are publicly available for the main conference papers. An expected outcome of the discussions is that the groups can engage in qualified discussions with the authors during the conference. The groups will prepare a 5-minute pitch of their favorite paper to be presented to the remaining groups on Friday December 11.

Each student will individually prepare a “blog post style” report of one particularly interesting paper from the conference or a workshop. The report should both explain the scientific content of the paper and discuss its strengths, weaknesses and potential community response.

Course schedule:

Monday Dec. 7, 2020
Full day of tutorials.
NeurIPS provides a selection of virtual tutorials (lectures), which will be followed and discussed within each group.

Tuesday Dec. 8, 2020
Read and discuss papers within each group.

Wednesday Dec. 9, 2020
Read and discuss papers within each group.

Thursday Dec. 10, 2020
Read and discuss papers within each group.

Friday Dec. 11, 2020
From 10:00 to 12:00 all groups gather (virtually). Each group will present the key ideas, tricks, and insights of their favorite paper so far to the other groups, followed by discussions in plenum.
The remainder of the day is devoted to reading and discussing workshop papers within the group.
Last updated
10. december, 2020