Machine learning and inference are both concerned with finding patterns in data. In statistical inference this is done via formal mathematical and information theoretic models. Machine learning, in contrast,  uses numerical algorithms in which parameters are updated in response to sequentially presented data. We focus mainly on transparent machine learning, based on algorithms with clear (Bayesian) statistical interpretations, and interpretable parameters.  

Quantum machine learning is the interesting interface between machine learning and quantum physics. Quantum annealing has been successfully applied to optimisation problems. We are interested in how the quantum formalism can be used to advance machine learning. First, to exploit quantum properties such as entanglement in classical data analysis. Second, to accelerate machine learning by implementation on quantum hardware.

In medical statistics one tries to infer statistical patterns from patient data with which to predict the evolution of their disease or their response to treatments (`personalized precision medicine'). In collaboration with the UMC, we focus specifically on the statistical and computational problems associated with latent heterogeneity of patients and diseases, and on developing new methods for handling very high dimensional data without overfitting.