Machine Learning and Probabilistic Methods for Power Grids/Networks



Machine learning tools allow us to study complex systems in novel ways. A defining feature of many complex systems are the underlying interaction networks. A crucial challenge in the application of machine learning tools to complex systems science is the development of ML methods that are network aware. A first significant and active step in this direction is the ongoing rapid development of Graph Neural Networks. We offer BA/MA thesis topics that investigate GNNs in the context of power grids and complex system science, but also contribute novel architectures that arise out of network science.

Probabilistic Methods

Probabilistic methods help us systematically understand and design systems that defy analytic or exact understanding. In recent years we have developed a number of novel methods that can be used to study but also design complex dynamical networks to fulfil certain functions. Applying these methods in new contexts, working out their implications, or further refining and developing them are all challenging and interesting Thesis topics.