Thesis topics

The following are examples of potential topics for a bachelor or master thesis. Other topics are also possible, to be agreed on by the student and the supervisor.

• Optimizing the memory footprint of distributed secure neural network inference
• Secure and explainable distributed inference with neural networks
• Secure inference with dynamic neural networks
• Video analysis with secure neural network inference
• Analysis of security threats in retrieval-augmented generation
• Simulation of cyber attack and defense strategies
• Applying optimization algorithms to the prioritization of IT security risks
• Applying machine learning to the prioritization of IT security risks
• Formal verification of security properties in neural networks
• Implementation and analysis of a privacy-preserving robotaxi service
• Usage of large language models in privacy engineering

Hall of Fame

Some bachelor or master theses are so good that they can serve as the basis for a scientific publication. Some examples:

• The paper "Secure Neural Network Inference for Edge Intelligence: Implications of Bandwidth and Energy Constraints" was based on the bachelor thesis of Jorit Prins, supervised by Prof. Mann when working at the University of Amsterdam
• The paper "Secure Neural Network Inference as a Service with Resource-Constrained Clients" was based on the bachelor thesis of Rik de Vries, supervised by Prof. Mann when working at the University of Amsterdam
• The paper "Simulation-based analysis of threats to location privacy in fog computing" was based on the bachelor thesis of Theresa Wettig, supervised by Prof. Mann when working at the University of Duisburg-Essen
• The paper "Classification of optimization problems in fog computing" was based on the master thesis of Julian Bellendorf, supervised by Prof. Mann when working at the University of Duisburg-Essen. The paper has been cited over 150 times, making it an important reference in its field of study