The Uncertain Systems lab aims to identify the mechanisms by which complex networked systems evolve and behave. Our main objectives are to identify the low-dimensional regularities in these systems so that we can interpret their behaviour, to predict how they will act in their future environment, and to eventually exert control over them.
Of course, doing so is a daunting task. Not only are large-scale system dynamics difficult to identify in general, but our observations of these systems are often sparse, indirect and noisy. Dealing with both this epistemic and aleatoric uncertainty requires the development of new probabilistic models and techniques, which plays a central role in our group. We particularly make use of Bayesian nonparametric methods, which allows us to adjust model complexity on the fly.
We apply our work in several domains, including: network neuroscience, genetics, (human) learning behaviour, and (personalized) healthcare.