TUK Students: BSc/MSc disseratations available

Please get in touch about a BSc or MSc disseration. Topics include

Table of Contents

Survival/time-to-event analysis

Survival/time-to-event analysis is an important field of Statistics concerned with understanding the distribution of events over time. Survival analysis presents a unique challenge as we are also interested in events that do not take place, which we refer to as ‘censoring’. Survival analysis methods are important in many real-world settings, such as healthcare (disease prognosis), finance and economics (risk of default), commercial ventures (customer churn), engineering (component lifetime), and many more. Recently there has been an increased interest in applying machine learning to survival analysis in order to make more powerful predictions. We encourage successful candidates to explore areas of machine learning within survival analysis that are of interest to them and to pursue novel methods. We also encourage open-source software development in R or Julia. In particular, the candidate will be expected to contribute to the survival analysis development in MLJ. As well as advances in model development, research will also be expected to explore practical aspects of model comparison including validation and benchmarking. This comprehensive programme will ensure any successful candidate is an expert in machine learning survival analysis. Successful candidates will work on cutting-edge research with access to state-of-the-art technology and will be at the forefront of survival analysis research. The project will further allow the candidate to develop the necessary skills to become an independent researcher by encouraging novel ideas and methods, as well as helping the candidate grow their own network of academics across the world.


  • Kvamme, H., Borgan, Ø., & Scheel, I. (2019). Time-to-event prediction with neural networks and Cox regression. Journal of Machine Learning Research, 20(129), 1–30.
  • Lee, C., Zame, W. R., Yoon, J., & van der Schaar, M. (2018). Deephit: A deep learning approach to survival analysis with competing risks. In Thirty-Second AAAI Conference on Artificial Intelligence.
  • Katzman, J. L., Shaham, U., Cloninger, A., Bates, J., Jiang, T., & Kluger, Y. (2018). DeepSurv: personalized treatment recommender system using a Cox proportional hazards deep neural network. BMC Medical Research Methodology, 18(1), 24.
  • Gensheimer, M. F., & Narasimhan, B. (2019). A scalable discrete-time survival model for neural networks. PeerJ, 7, e6257.

Fault and Efficiency Prediction in High Performance Computing


  • deep learning
  • event data modelling
  • time series
  • survival modelling

High use of resources are thought to be an indirect cause of failures in large cluster systems, but little work has systematically investigated the role of high resource usage on system failures, largely due to the lack of a comprehensive resource monitoring tool which resolves resource use by job and node. This project studies log data of the DFKI Kaiserslautern high performance cluster to consider the predictability of adverse events (node failure, GPU freeze), energy usage and identify the most relevant data within. The second supervisor for this work is Joachim Folz.

Data available:

Prometheus-kompatibles System is und wir haben


Speaker-identification from voice

The high fidelity identification of a person is crucial for a variety of security related issues. There has been a recent progress of doing so using techniques of image vision - this thesis aims at highly visible impact on a small training set with a twist.


  • Transfer learning
  • Youtube data scraping
  • High impact paper


Learning Robust Time-Frequency Transformation of Audio

Data Safe havens, federated learning and other privacy enhancing techniques

In order to answer questions like who is at higher risk of dying from Covid, how to improve treatment strategy or what are effective interventions for mental health. This requires to develop models and access data in way that minimises the risk of disclosing sensitive data. There is a trade-off between locking the data away (or not collecting it systematically in the first place) - keeping it safe at the cost of not extractingt the beneficial knowledge of the data.

This projects aims at reviewing state of the art and consider different reference implementations. Possibly contributing to the research project SEMLA: A SEcure Machine Learning Architecture.


Customer choice and identification to improve climate change

Green technolgy has many different shapes and forms. It is key requirement to understand the customer behaviour in order to optimise utility and replace more emissions. Keywords:

  • event data, predictive maintance
  • promotions, A/B testings

Feel free to reach out if the topic sounds interesting or if you have ideas related to this work. We can then brainstorm a specific research question together.

Other topics:

  • Privacy preservation - multiparty computations

  • Privacy preservation - Federated learning

  • Applications - Customer and Churn Modelling

  • Applications - A/B testing

  • User interfaces for Julia Programs

  • Deep Learning for choice modeling

  • Deep Learning And Differential Equations - Ground Truth and Recovery

Email me or get in touch. register your interest