Student projects

I continually supervise student projects and special courses at all levels. If you are interested in working with me within one of the project themes listed here, you are very welcome to contact me.

Project proposals

You can find a list of specific project proposals on the Cogsys projects page.

Recommended courses

As a prerequisite to project work I recommend the following courses
 

Open project themes

  1. Bayesian deep learning

    Approximate Bayesian inference in deep neural networks, using techniques such as stochastic gradient Markov chain Monte Carlo sampling, Kronecker-factored approximate curvature, ensemble methods, stochastic weight averaging, variational inference, etc.

  2. Uncertainty quantification

    Methods for quantifying uncertainty including Bayesian methods, conformal prediction, and uncertainty distillation.

  3. Graph neural networks

    Applications of graph neural networks to models of molecules and materials.

  4. Generative models for molecules and materials

    Development of models for generating novel molecular and material structures e.g. based on diffusion og stochastic interpolants, enabling efficient exploration of chemical and structural design spaces.