Theses
If you are interested in writing your thesis at our chair, please review our available topics. If you are interested in a specific topic, please contact the corresponding assistant.
If you are interested in multiple research areas at the chair, please let us know. We will gladly help you choose a suitable topic.
| Topic | Type | Assistant | Status |
| Understanding and analysing safety alignment in LLMs | B.Sc. / M.Sc. Thesis | Hamid Dashtbani | Reserved |
| Analysing representations of harmful concepts in LLMs | B.Sc. / M.Sc. Thesis | Hamid Dashtbani | Available |
| Protecting privacy in LLMs with homomorphic encryption | B.Sc. / M.Sc. Thesis | Hamid Dashtbani | Available |
| Facilitating proof of work in cryptocurrencies with LLMs | B.Sc. / M.Sc. Thesis | Hamid Dashtbani | Available |
| Are safety neurons identifiable? Non-uniqueness of safety mechanisms in LLMs | B.Sc. / M.Sc. Thesis | Ansgar Schäfftlein | Available |
| Comparing neuron-, weight-, and subspace-level vulnerabilities in safety aligned LLMs | B.Sc. / M.Sc. Thesis | Ansgar Schäfftlein | Available |
| Improving safety alignment training through safety neuron splitting | B.Sc. / M.Sc. Thesis | Ansgar Schäfftlein | Available |
| Defending against low-rank malicious fine-tuning with high-rank safety alignment | B.Sc. / M.Sc. Thesis | Ansgar Schäfftlein | Available |
| Dynamic analysis for automated vulnerability repair | B.Sc. / M.Sc. Thesis | Varun Gadey | Reserved |
| White-box mitigation of vulnerabilities in AI-generated code | B.Sc. / M.Sc. Thesis | Varun Gadey | Reserved |
| Secure code generation for mixture-of-experts LLMs | B.Sc. / M.Sc. Thesis | Varun Gadey | Available |
| Mitigating hashjacking attacks on AI browsers using ML | B.Sc. / M.Sc. Thesis | Dr. Venkata Marella | Available |
| Memory-Constrained Neural Architecture Search for On-Device Deepfake Audio Detection on ESP32 | B.Sc. / M.Sc. Thesis | Behrad Bayat | Reserved |
| Secure and robust model watermarking in federated learning | B.Sc. / M.Sc. Thesis | Behrad Bayat | Reserved |
| Cross-client memory backdoor attacks in federated learning | B.Sc. / M.Sc. Thesis | Behrad Bayat | Reserved |