PrivML Research Group
This page provides an overview on projects I am working on and students I am supervizing or have supervised in the past.
Current projects and research interest
- Individualizing privacy guarantees according to user preferences
- Measuring and auditing privacy leakage in ML
- Jointly optimizing for privacy and different aspects of trustworthy ML
Currently ongoing theses
Student | Program | Thesis Title |
---|---|---|
M. Nest | Master | Temporal Membership Inference Attacks |
I. Fendel | Bachelor | Group Membership Inference Attacks |
A. Meszaros | Bachelor | Taxonomy of Privacy Attacks in Machine Learning |
Past theses
Student | Program | Thesis Title | Link |
---|---|---|---|
M. Krüger | Bachelor | Application and Evaluation of Differential Privacy in Health Data Classification Tasks | |
O. Bouanani | Bachelor | Neural Network Architectural Choices for Privacy | |
C. Mühl | Master | Personalizing Private Aggregation of Teacher Ensembles | |
T. Känel | Bachelor | Practical Evaluation of Neural Network Watermarking Approaches | |
D. Wang | Bachelor | Evaluating and Adapting Existing Neural Network Watermarking Approaches to Online Learning Scenarios | |
D. Sosnovchyk | Bachelor | Evaluating Privacy of Synthetic Data Through Metrics | |
W. Gu | Bachelor | Differential Private Synthetic Data Generation | |
J. Ihrig | Master | Privacy Quantification Methods |
Open topics for collaboration
Currently, the following topics are available:
- Side Channel Attacks against Neural Networks
- Property Inference Attacks against Neural Networks
- Bias in AI
If you are interested in collaborating in one of the topics mentioned above, please feel free to reach out.