Dominik Fay
LightNews LightNews
dominikfay.me
Dominik Fay
@dominikfay.me
30 followers 86 following 4 posts
AI/ML Researcher @ Elekta, Stockholm
Posts Replies Media Videos
dominikfay.me
I wrote a little blog post about Wordle, differential privacy and chocolate 😋 dominikfay.me/blog/2025/ma...

Thanks to @desfontain.es for creating a fun contest!
May 7, 2025 at 8:36 PM
Reposted by Dominik Fay
rasmuspagh.net
Opportunity to do a PhD in a wonderful city, at a great university, with a fantastic supervisor! x.com/ioanabercea/...
Ioana O. Bercea on X: "I’m recruiting! Please help spread the word! 2PhDs available at KTH Royal Institute of Technology, in the area of Theoretical Computer Science. Deadline is April 3rd, 2025. More details here: https://t.co/rbDT01n5XI" / X
I’m recruiting! Please help spread the word! 2PhDs available at KTH Royal Institute of Technology, in the area of Theoretical Computer Science. Deadline is April 3rd, 2025. More details here: https://t.co/rbDT01n5XI
x.com
March 20, 2025 at 8:53 AM
Reposted by Dominik Fay
tmlr-pub.bsky.social
Personalized Privacy Amplification via Importance Sampling

Dominik Fay, Sebastian Mair, Jens Sjölund

Action editor: Antti Koskela

https://openreview.net/forum?id=IK2cR89z45

#privacy #private #personalized
January 15, 2025 at 3:07 PM
Reposted by Dominik Fay
dppapers.bsky.social
Locally Differentially Private Online Federated Learning With Correlated Noise
Jiaojiao Zhang, Linglingzhi Zhu, Dominik Fay, Mikael Johansson
http://arxiv.org/abs/2411.18752
Locally Differentially Private Online Federated Learning With Correlated Noise Jiaojiao Zhang, Linglingzhi Zhu, Dominik Fay, Mikael Johansson http://arxiv.org/abs/2411.18752 We introduce a locally differentially private (LDP) algorithm for online federated learning that employs temporally correlated noise to improve utility while preserving privacy. To address challenges posed by the correlated noise and local updates with streaming non-IID data, we develop a perturbed iterate analysis that controls the impact of the noise on the utility. Moreover, we demonstrate how the drift errors from local updates can be effectively managed for several classes of nonconvex loss functions. Subject to an $(\epsilon,\delta)$-LDP budget, we establish a dynamic regret bound that quantifies the impact of key parameters and the intensity of changes in the dynamic environment on the learning performance. Numerical experiments confirm the efficacy of the proposed algorithm.
December 2, 2024 at 5:27 AM
Reposted by Dominik Fay
nandofioretto.bsky.social
🆘Help needed!

Are you working on Privacy (from a Technical (e.g., Differential Privacy), Policy, or Law perspective)?

Please give your availability to review for PPAI (ppai-workshop.github.io) if you can!
We'd highly appreciate it! 🙏

forms.gle/dqjVsBsR2y81...
Sixth AAAI Workshop on Privacy-Preserving Artificial Intelligence
ppai-workshop.github.io
November 23, 2024 at 12:47 AM