Doctoral position in Sample-Efficient Probabilistic Machine Learning (4 years, fully funded)

The Machine and Human Intelligence research group led by Assistant Professor Luigi Acerbi is looking for a PhD candidate eager to work on new machine learning methods for smart, robust, sample-efficient probabilistic inference, with applications to computational and cognitive neuroscience. The candidate will join a newly established research group at the Department of Computer Science of the University of Helsinki (Finland) with strong links to the Finnish Center for Artificial Intelligence (FCAI).

Our group is developing novel machine learning approaches for approximate Bayesian inference that use only a small number of likelihood evaluations, which can be a game-changer for complex computational models or when resources are limited. A state-of-the-art framework being developed in our group is Variational Bayesian Monte Carlo (VBMC), which combines Gaussian process surrogates, active learning, variational inference and Bayesian quadrature (Acerbi, NeurIPS; 2018, 2020). Promising thesis projects include extending the representational power of VBMC (e.g., discrete variables, more complex posteriors, higher dimension); exploiting recent advances in Gaussian process inference for superior scalability; combining VBMC with Bayesian deep learning; strengthening the connections with simulator-based inference; and exploring the theoretical properties of the framework.
 
The position is full-time, funded for four years and will be filled as soon as possible, with a negotiable starting date in early 2021. --