Computational Neuroengineering

We develop machine-learning methods for interpreting complex data for applications in science and engineering.

We want to understand the computations that neural networks use to process sensory information and to control intelligent behaviour. To this end, we develop statistical models and machine learning algorithms for large-scale data-analysis, and collaborate with experimental laboratories performing measurements of neural activity and behaviour. We use Bayesian inference as our core research methodology: It allows us to build structured probabilistic models which incorporate prior knowledge, and makes it possible to quantify uncertainty about model-parameters and predictions, and to identify which additional measurements would be most informative in improving them.