Trustworthy Machine Learning

Dear Students,

The video from today's Q&A session can be found here: https://youtu.be/k2U2MRKINWE

Next week, there will be two tutorials: (1) on data provenance for generative models, and an additional one (1) on Assignment 3: Robustness. 

We will cover the lectures on Collaborative (Federated Learning) on June 17th and you can already post your questions on the Forum. 

With kind regards,

The TML Team

Dear students,

The assignment 3 on the Robustness topic is out. You can find the task description as well as tutorial in the materials section on CMS. The task description also contains a link to the video that explains the task. The deadline for this assignment is on 16.06 23:59. 

Best of luck,

TML team

Dear everyone,

Thanks again for your patience. Eventually, the university has been able to organize us a room that fits everyone! We are extremely grateful that we can use 

 

HS I in Maths, E 2.5.

 

So, meet all of you there. The exam will start on time at 4PM, so please be sure to arrive early enough to find your seat.

Please be reminded that we only printed exams for those students who registered on CMS. If you have not registered, you will not be able to write the midterm.

Kind regards
Franziska and Adam

 

Dear everyone,

Thanks again for your patience. Eventually, the university has been able to organize us a room that fits everyone! We are extremely grateful that we can use 

 

HS I in Maths, E 2.5.

 

So, meet all of you there. The exam will start on time at 4PM, so please be sure to arrive early enough to find your seat.

Please be reminded that we only printed exams for those students who registered on CMS. If you have not registered, you will not be able to write the midterm.

Kind regards
Franziska and Adam

 

Dear All,

First, a reminder: the midterm is in E1 3 - HS 001 this Wednesday (June 3rd) from 16:00 to 17:00. --> See room update news: Midterm takes place in HS I, Maths in E 2.5. 

After the midterm, we will cover the topic of "Data Provenance for Generative Artificial Intelligence". The new lecture is uploaded here: https://youtu.be/rxJ7S7OCZc0 (we started the premiere and the full video will be available in a few minutes). Please, post your questions on the forum. The corresponding lecture slides can be found in the materials: https://cms.cispa.saarland/tml2026/dl/21/06-DataProvenanceForGenerativeArtificialIntelligence.pdf

With kind regards,

The TML Team

Dear Students,

Assignment 1 (Membership Inference Attack) marks and feedback are now available. You can view them on your CMS page Personal Status tab.

Regarding the email you received yesterday: the marks mentioned in that email are correct, however the grade of 5 was assigned automatically by the system and should be ignored. The final course grade will be determined at the end of the semester based on all components.

If you have any questions about your marks, feel free to reach out to us via the forum.

Thank you. 

Dear Students,

We are forwarding to you the message from the IT:

Hi everyone,

We are aware that several students are experiencing SSH connection timeouts to the HPC cluster (both conduit and conduit2). This appears to be a cluster-side issue and unfortunately something we cannot resolve directly. For such issues, please reach out to the HPC support team directly as they are best placed to help.

As a workaround, plain terminal SSH seems to be working for most students, while the VS Code Remote SSH connection is more affected. You can connect directly with:
ssh <username>@conduit2.hpc.uni-saarland.de or ssh <username>@conduit.hpc.uni-saarland.de

In the meantime, Google Colab is also a good temporary alternative if you need GPU access to continue working on the assignment.

Apologies for the inconvenience, and good luck with the submission!

TML Team

Dear students,

You can find the question sheet for the 4th tutorial on the first part of model stealing in the materials section. Please review them before tomorrow's lecture (20.05). The tutorial will be held after the question answering session. Also, we will have the next tutorial on Model Stealing part II (SSL) and Adversarial & Robustness concepts next week on 27.05. 

Best regards,

TML team

Dear Students,

Please, watch the lecture on Robustness for the next week (May 20th) and post the questions on forum beforehand. 

Next week, we also would like the 3 best teams from the scoreboard on assignment 1 to give short (up to 5 min) talks on their solutions. You can gain additional bonus points (up to 2 points) per assignment for this presentation part. 

With kind regards,

The TML Team

Dear students,

Assignment 2 is released. You can find the task description under the materials section on CMS. The deadline for this assignment is on 26.05.2026 23:59. Please read the task description and the following comments carefully:

• The assignment deadline is due on 26.05.2026 23:59. Both CMS submissions for the code and report as well as the API submission to the leaderboard will be closed at that time. Please consider the deadline and do not postpone the submission to the last moments. 
• No submissions via email will be accepted under any circumstances. You can make a submission to CMS and later, update that submission to avoid any issues.
• Please note that for assignment 2 and the rest of the assignments you should submit the report as well as your code on CMS. You can refer to the task description for more information. 
• API keys and cluster credentials are changed for some students due to change in teams. Please refer to the Personal Status page on CMS to find your new credentials and use those keys for the remaining assignments.

Best of luck in your assignment.

TML team

Dear Students,

Please note the following room changes:

• On 01.07.2026 and 29.07.2026, the class will take place in HS002, Building E1.3 instead of room 0.05, Building C0.
• We will also need to use a different room on 03.06.2026. We will inform you of the new room as soon as it has been booked.

With kind regards,

TML team

Dear students,

For the teams that have two members we have placed the API keys (for submission to the evaluation system) and cluster credentials (for computation jobs) on their personal status page.

If you do not already have a teammate we suggest forming a team before the deadline 29.04. After forming the teams you will get the aforementioned credentials to start working on the assignment 1 (due: 06.05). In order to have enough time to work on your solutions and enter your results on the leaderboard we recommend forming the teams as soon as possible. If you form a team after this news, please let the tutors know via email such that you can get access as soon as possible.

We will have our first tutorial on Wednesday 22.04 where we discuss how to set up the cluster and submit your solutions to the leaderboard. 

Best of luck,

TML team

Dear Students,

The first assignment was released here: https://cms.cispa.saarland/tml2026/dl/9/Assignment1-MembershipInference.pdf 

Note: the assignments must be completed in pairs. API keys will only be generated for teams of two (if you do not have a partner, you will not receive an API key and will not be able to submit solutions for the assignments).

Have a good week!

 

 

Dear Everyone,

The signup for the team pairing has opened. You have until the submission of the first assignment to sign up there with a partner. As noted during the intro lecture: single-participant submissions cannot be considered for grading. I made a thread in the "Off-Topic" section of a forum where you can look for assignment partners if you have not found one, yet.

See you next week,

Franziska and Adam