Skip to main content

Project and dissertation

In the third term, students do their final project and write a dissertation. We run three different types of projects in MISCADA:

  • Research-led projects in the respective specialisation areas
    (such as “apply your HPC skills to a particular astrophysics simulation” or “use machine learning for a particular problem from financial mathematics”)
  • Research-led projects in the methodological core disciplines
    (such as “extend a particular statistical technique” or “prototype a challenging algorithm with a new supercomputing API”)
  • Projects in collaboration with industry partners (typically in R&D),  with joint academic and industry supervision.

Examples for successful projects and other activities can be found under the rubric success stories.

All information below is for your information only. Please do not worry or start activities re the project before you join the course. You will need to the few months of taught content to understand what projects can, in theory, cover, and you will have enough time to decide what route you want to go down and what you want to do actually for your dissertation after the first term has completed.


The workflow how projects are managed changes slightly from year to year, but in principle it follows the following steps:

  • We launch a call for projects early in term 2. That is, academics put their project proposals into a database.
  • If an industry partner proposes a project, we analyse this project (does it meet our standards?), pair the industry partner up with one academic, and then put it into our database. Industry projects thus always have one designated academic supervisor.
  • Students can propose academic projects themselves as well. In this case, we again show them around to the other academics. If the project meets our standards, then an academic who is interested in the student’s proposal puts the project into the database and serves as supervisor for this project.
  • Students sign up for projects from the database with their preferences (if they have proposed something themselves, then the preference should be clear).
  • We pair up academic supervisors and students – trying to meet the students’ preferences.

Can I bring in “my” company?

Every year, a couple of students wants to do projects in collaboration with a company for which they have worked before as interns. In principle, we try to support this. For us, this is like any other company that approaches us to host a dissertation: we analyse the project to ensure that it meets our academic criteria (it should go beyond sole code monkeying or plain data crunching) and then pair the company brought in by the student up with a colleague who supervises the project from MISCADA’s point of view.

So if you have a project in mind, you have to find someone at the company who would be willing to supervise you, and you have to write us a brief description what this project is about. We try to find a good academic peer, and then this academic peer will sort out with your industry supervisor whether this is a well-suited MISCADA project.

Example dissertation titles from previous years

Below are some examples for projects that MISCADA students worked on in previous years. This list is not complete. Often, several students work on projects with the same title, but develop their project into different directions. All projects are, in principle, open to all specialisation streams, though students are given a list of recommended previous knowledge and hence typically find some canonical fits to their background. Some of the projects involve industry, others ask the student to transfer methodological core competences into totally new application domains. Others again combine different areas or involve industry partners (even though they are not enlisted as industry-driven here).


Performance investigation of a novel 6D torus network on a High Performance Computing system

GPU programming with C++ 23 (or maybe 27)

Performance profiles considered harmful


Mixed-precision algorithms for computational linear algebra

Implementation and Analysis of Graph Algorithms on GPUs

Earth & Environment

Using Machine Learning to fill in the gaps in emission line diagnostics

Calculating crop yield from UAV-based imagery

Statistical Machine Learning to Feed the World

Fast and efficient Bayesian inference for stochastic epidemic models

Using image classification and machine learning to identfy mineral deposits in Greenland

Mine water geothermal heat … what are the carbon savings?

Climate change: where is changing and why?

A Statistical Analysis of the Effect of Ambient Temperature on the Accuracy of Weighing Systems

Earthquake detection classification and forecasting using machine learning

Machine learning for detection classification and forecasting of earthquakes

Automated processing of global seismic data
Financial mathematics

Regression analysis for big data

BSDEs and their application in option pricing problems

Cellular Automata Modeling of Financial Markets

Nonlinear Time Series Modelling

Modeling Financial Time Series via Machine Learning

Stochastic differential equations with Markovian switching
Machine learning methodology

Aid nonlinear dynamical system modelling with Machine Learning

Uncertainty and Robustness in Neural Density Estimation

History Matching of Complex Computer Models

Statistical and Machine Learning methods for classification and clustering
Industry-driven projects

Modelling the Demand for Gas by Industry

Analysis of acoustic data for industrial applications

Statistical and Machine Learning for Medicine

Manufacturing Innovation

Crash bang wallop!: Collecting and interpreting acoustic signatures of collisions

Machine Learning structure formation in alternative theories of gravity

Machine Learning structure formation in alternative theories of gravity

Predicting biological activities using QSAR (Quantitative Structure Activity Relationship) models

Understanding and Reducing Prediction Uncertainty When Extrapolating from Regression Models

Developing algorithms to automated strong gravitational lensing mass modelling from space-based imaging of galaxy clusters
Particle Physics

Machine Learning for Particle Physics signatures

Learning Scattering Amplitudes with Neural Networks
Education & training

Plagiarism detection for Jupyter notebooks
Health Data Science

Using chest X-ray images and deep learning for automated detection of pathologies

Anomaly Detection in Medical CT Volumes with Geometric and/or Radiodensity Priors

Unpaired Image-to-Image Translation in Medical Imaging

Modelling COVID-19 – Did we do something wrong?

Applied computer science

Real-Time Hand Gesture Recognition in Cluttered Environments using Deep Learning

Deep Generative Modeling for Pixel Art


Human object interaction recognition
Game-based Learning
Theoretical computer science

Scheduling Problems: Planes Trains and Boarding Algorithms

Learning-Augmented Algorithms for Minimum Spanning Trees under Uncertainty

Machine learning for board games – Reversi/Othello or similar

Limit theorems in game-theoretic probability