Advanced topics in causality and causal ML research

Hi all,

as discussed in class, we will have presentations for the course on July 16th from 9-12. 

The groups of 4 will have 48 minutes of presentation (~12 minutes each, but this can be a bit flexible as long as it is proportionate) + 12 minutes of Q&A (3 minutes each)
The group of 2 will have 24 minutes of presentation (~half each) + 6 minutes of Q&A (3 minutes each)

In general, the presentation format is quite flexible, but it should include information on:

- Background and Motivation
- Explanation of the method you are extending
- Explanation of the proposed extension 
- Experimental results (possibly including reproducing the paper, but not necessarily if you have more interesting results)

The presentation will be graded individually, also based on the Q&A and it will count for 50% of the grade.

Separately, I will also ask you to submit one Jupyter notebook per group (ideally in the form of a github link, but otherwise you can send it to me by mail) that shows your proposed extension and the experimental results that you presented. Please include also a short paragraph describing each team member contributions.
The deadline for this will be one week after the presentations, July 23rd, and it will be again graded individually and count for 50% of the grade.

Just a small reminder, next week we will not be meeting, so we will meet directly for the presentations.
Let me know if you have questions.

Cheers!
Sara

Hi all,

Good news! We got now 3 accounts (one per team) for the cluster access from the department IT, 

Here are the instructions:

"Here's some info on how to use the cluster: https://sic.link/condor and some more over in the FAQ: https://wiki.cs.uni-saarland.de/en/HPC/faq
We request that you recommend students to not use VS Code Remote SSH sessions or similar to build/run stuff on the submission nodes, but instead use Code Server inside interactive jobs: https://wiki.cs.uni-saarland.de/en/HPC/interactive"

These are the 3 accounts:
- causalml26_team000 - this is for the CRL group
- causalml26_team001 - this is for the causal discovery group
- causalml26_team002 - this is for the causality-inspired ML/downstream tasks group

I will give you the passwords in the meeting inperson today.

Cheers,

Sara

Hi all,

As discussed in class, nthe next meeting each group will prepare a presentation of 20 minutes on one of the papers they want to extend (+10 minutes questions from the others).

We currently have three groups: 

- causal discovery (full)

- downstream tasks (full) 

- causal representation learning (potentially still 2 spots)

If you are not in a group, but you still want to participate in the course, please let me know  asap.

 

Cheers!

Sara