Dominik Hintersdorf

Researcher@DFKI and final year PhD student@AI & ML Lab TU Darmstadt

prof_pic.png

My research focuses on the privacy and security of deep learning systems. As deep learning increasingly powers real-world applications, the data used to train these models becomes ever more critical. Many of these applications require models to be both reliable and secure, particularly when sensitive data, such as medical records or personal images, is involved in the training process. In my work, I investigate potential threats to the safety and security of deep learning models and develop strategies to mitigate these risks.

I received my Masters from TU Darmstadt and I am currenctly a PhD student at TU Darmstadt and the German Research Center for AI (DFKI) under the supervision of Prof. Kristian Kersting.

news

Apr 6, 2025 :mega: Our Workshop “The Impact of Memorization on Trustworthy Foundation Models” has been accepted at ICML 2025. You can find the CFP here.
Mar 26, 2025 :blue_book: I sucessfully submitted my PhD thesis titled “Understanding and Mitigating Privacy Risks in Vision and Multi-Modal Models”.
Oct 10, 2024 :tada: Our paper Class Attribute Inference Attacks: Inferring Sensitive Class Information by Diffusion-Based Attribute Manipulations was accepted at the AdvML Frontiers Workshop at NeurIPS 2024!

selected publications

  1. ECAI
    Defending Our Privacy With Backdoors
    Dominik Hintersdorf, Lukas Struppek, Daniel Neider, and Kristian Kersting
    In Proceedings of the 27th European Conference on Artificial Intelligence , 2024
  2. JAIR
    Does CLIP Know My Face?
    Dominik Hintersdorf, Lukas Struppek, Manuel Brack, Felix Friedrich, Patrick Schramowski, and 1 more author
    Journal of Artificial Intelligence Research, 2024
  3. NeurIPS
    Finding NeMo: Localizing Neurons Responsible For Memorization in Diffusion Models
    *Dominik Hintersdorf, *Lukas Struppek, Kristian Kersting, Adam Dziedzic, and Franziska Boenisch
    In Proceedings of the 38th Conference on Neural Information Processing Systems (NeurIPS) , 2024