Topics in Privacy-Preserving Machine Learning and Optimization

As explained at the beginning of the seminar, AI tools may be used only for general background questions. They may not be used to generate summaries, reviews, submitted text, or to polish written work. An AI-use statement is also required with every course deliverable.

Because the short paper review was the first submitted assignment, I am offering everyone an optional one-time opportunity to ensure that their submission complies fully with the course policy.

You may submit a revised version of your short paper review by Friday, 14 August at 12:00 CEST. The revised version should represent your own reading, understanding, and writing. It must also include an honest AI-use statement that either:

• confirms that no AI tools were used; or

• identifies every AI tool used and explains specifically how it was used.

Permitted use for general background or clarification should be described clearly. AI use for summarizing the assigned paper, drafting or rewriting the review, generating submitted text, or polishing the writing is not permitted.

There will be no penalty for making use of this one-time resubmission opportunity. If you resubmit, I will grade the revised version rather than the original submission. I will begin grading after the resubmission deadline.

Please remember that the presentation and question-and-answer portion will also assess your personal understanding of the assigned material. The same AI policy, including the required AI-use statement, applies to the presentation and final seminar paper.

Best,
Andrew

Below I have listed each student's assigned presentation/paper topic and review topic in the following format: Presentation Topic; Review Topic. 

For the Review, you may choose either one of the two papers under the assigned topic to review. 

Jilixin Tang: DP Bilevel Optimization; User-Level DP SCO

Bhakti Kate: DP SCO in Other Geometries; DP SCO

Prathamesh Shinde: Robustness and Privacy; Parameter-Free and Adaptive DP Optimization

Qian Liu: User-Level DP SCO; DP FL

Saugata Purkayashta: DP SCO; DP Non-Convex Optimization

Bushra Ashfaque: DP FL; DP SCO in Other Geometries 

Oliver Bintriem: Parameter-Free and Adaptive DP Optimization; DP Bilevel Optimization 

Krithiga Ravikumar: Machine Unlearning; Robustness and Privacy 

Neshmia Hafeez: DP Non-Convex Optimization; Machine Unlearning 

Dear students,

I would like to let you know that the kick-off meeting and introductory lecture block for the seminar has been moved one week earlier.

The kick-off meeting and introductory lectures will now take place on:

July 6–9, 2026, 13:00–16:00
Location: CISPA C0 building, 0.01 Presentation Room

The block seminar presentations remain scheduled for:

August 31–September 2, 2026, 10:00–17:00
Location: CISPA C0 building, 0.07 Presentation Room

Please keep the full presentation block available for now. The final daily schedule will be announced after topics are assigned; depending on the final number of presentations, not all reserved time may be used.

The short paper review is due on August 5, and the final seminar paper is due on September 11. The full syllabus is available on the CMS website. 

Best,
Andrew

Dear all,

welcome to the seminar Topics in Privacy-Preserving Machine Learning and Optimization. I’m looking forward to working with you.

Here are the key dates and structure of the seminar:

*Introductory lectures (kickoff phase):
July 13–16, 2026
13:00–16:00 each day

The first session (July 13) will include a brief overview of the seminar structure and expectations, followed by the first lecture. Topics will be assigned during this session.

Independent reading phase:
Mid July – late August

Short paper review due:
August 5, 2026

*Block seminar (presentations):
August 31 – September 2, 2026
(Full-day sessions; exact room details will be posted here)

Seminar paper deadline:
September 11, 2026

*Denotes in-person class meeting. 

Best regards,
Andrew