Nils Lukas

Guten Tag , I am an incoming Assistant Professor at MBZUAI in Abu Dhabi, beginning August 2024. Previously, I did my PhD at the University of Waterloo in Canada, where I was supervised by Florian Kerschbaum. During my PhD I interned at Microsoft Research and Borealis AI and I was generously supported by the David R. Cheriton scholarship. In 2024, I received the Mathematics Doctoral Prize at the University of Waterloo and was nominated for the Governor General’s Gold Medal.

[Curriculum Vitae], [Research Statement]

Please reach out to me via e-mail if you are interested in joining MBZUAI as a student or a visting researcher/visiting postdoc with me.

My research focuses on designing safe and reliable Machine Learning systems in the presence of untrustworthy

Providers: Confidential computing via Homomorphic Encryption & Secret Sharing.
Data: Mitigate data poisoning during training & prompt injection during inference.
Models: Protect training data privacy through PII scrubbing & differential privacy.
Users: Control misuse by detecting generated (mis)information with watermarking

My work includes studying privacy attacks against large language models fine-tuned on private datasets, developing defenses against data poisoning, and creating multiple methods for controlling model misuse. In collaboration with our group, I have also contributed to developing Secure Multi-Party Computation protocols for Private Information Retrieval and the Secure Inference of Deep Neural Networks. I received a Master of Science from the RWTH-Aachen (Germany) in 2019.

If you have any questions or would like to learn more about my work, please feel free to contact me. You can also find my work on GitHub.

News

Apr 25, 2024	I am excited to have received the Mathematics Doctoral Prize and being nominated for the University of Waterloo Governor General’s Gold Medal.
Jan 16, 2024	Two papers accepted at ICLR’24 in Vienna. These include our paper on Adaptive Attacks against Image Watermarks and our work on Universal Backdoor Attacks.
Dec 11, 2023	I gave a research talk at Meta about the privacy of personal information in fine-tuned LLMs. [Meta slides]
Oct 9, 2023	I gave an research talk at the University of California, Berkeley and I won the Best Poster Award at the 2023 Cheriton Symposium at the University of Waterloo. [Announcement] [Cheriton poster] [Berkeley slides]
Jun 26, 2023	I gave a research talk at Google and MongoDB. Our IEEE S&P’23 paper was featured on the Microsoft Research Focus blog. [Google slides] [MongoDB slides] [Focus blog]
May 24, 2023	One paper accepted at USENIX’23 Security in Anaheim, California. We worked on Learnable Watermarks for Pre-trained Image Generators.
Apr 3, 2023	One paper accepted at IEEE S&P’23 in San Francisco, California. We worked on studying the Leakage of Personal Information in Language Models. 🏆 Our IEEE S&P’23 paper won the Distinguished Contribution Award at Microsoft’s internal Machine Learning, AI & Data Science Conference (MLADS) in Fall 2023.
May 1, 2022	I am doing a research internship at Microsoft Research supervised by Shruti Tople and Lukas Wutschitz.
Feb 1, 2022	🏆 I was awarded the David R. Cheriton Scholarship for outstanding academic achievements.
Jan 1, 2022	One paper accepted at IEEE S&P’22 in San Francisco, California. We handcraft Adaptive Attacks against Eleven Model Watermarks.

Selected Publications

Leveraging Optimization for Adaptive Attacks on Image Watermarks

Nils Lukas, Abdulrahman Diaa, Lucas Fenaux, and 1 more author

The Twelfth International Conference on Learning Representations (ICLR 2024), 2024

arXiv Bib

@article{lukas2023leveraging,
  title = {Leveraging Optimization for Adaptive Attacks on Image Watermarks},
  author = {Lukas, Nils and Diaa, Abdulrahman and Fenaux, Lucas and Kerschbaum, Florian},
  journal = {The Twelfth International Conference on Learning Representations (ICLR 2024)},
  year = {2024}
}

PTW: Pivotal Tuning Watermarking for Pre-Trained Image Generators

Nils Lukas, and Florian Kerschbaum

32nd USENIX Security Symposium, 2023

arXiv Bib Code

@article{lukas2023ptw,
  title = {PTW: Pivotal Tuning Watermarking for Pre-Trained Image Generators},
  author = {Lukas, Nils and Kerschbaum, Florian},
  journal = {32nd USENIX Security Symposium},
  year = {2023},
}

Analyzing Leakage of Personally Identifiable Information in Language Models

Nils Lukas, Ahmed Salem, Robert Sim, and 3 more authors

44th IEEE Symposium on Security and Privacy (SP), 2023

arXiv Bib Code

@article{lukas2023analyzing,
  title = {Analyzing Leakage of Personally Identifiable Information in Language Models},
  author = {Lukas, Nils and Salem, Ahmed and Sim, Robert and Tople, Shruti and Wutschitz, Lukas and Zanella-B{\'e}guelin, Santiago},
  journal = {44th IEEE Symposium on Security and Privacy (SP)},
  year = {2023},
}

Sok: How Robust is Image Classification Deep Neural Network Watermarking?

Nils Lukas, Edward Jiang, Xinda Li, and 1 more author

In 43rd IEEE Symposium on Security and Privacy (SP), 2022

arXiv Bib Code

@inproceedings{lukas2022sok,
  title = {Sok: How Robust is Image Classification Deep Neural Network Watermarking?},
  author = {Lukas, Nils and Jiang, Edward and Li, Xinda and Kerschbaum, Florian},
  booktitle = {43rd IEEE Symposium on Security and Privacy (SP)},
  pages = {787--804},
  year = {2022},
  organization = {IEEE},
}

Deep Neural Network Fingerprinting by Conferrable Adversarial Examples

Nils Lukas, Yuxuan Zhang, and Florian Kerschbaum

Splotlight Presentation at The Ninth International Conference on Learning Representations (ICLR 2021), 2021

arXiv Bib

@article{lukas2021deep,
  title = {Deep Neural Network Fingerprinting by Conferrable Adversarial Examples},
  author = {Lukas, Nils and Zhang, Yuxuan and Kerschbaum, Florian},
  journal = {Splotlight Presentation at The Ninth International Conference on Learning Representations (ICLR 2021)},
  year = {2021},
}