Artificial intelligence (AI) has shaped our society and influenced many scientific disciplines. Examples include applications in smartphones, autonomous cars, modelling the impact of political decisions on the spread of the COVID-19 virus or predicting climate disasters. Many of these require a probabilistic approach to AI that accounts for uncertainties. Unfortunately, probabilistic AI methods are often challenging and computationally heavy. Probabilistic circuits are a new technique that promises a remedy to these problems. This research project aims to develop tools for efficient computations in probabilistic models by utilizing core ideas of probabilistic circuits. In particular, the project will focus on statistical inference in stochastic processes and Bayesian neural networks. Although the project is on fundamental research in AI, the results will have a lasting impact in areas where decision making needs to be robust, reliable, and efficient.

2022

2025