About
Vincent is part of the SPG group, he works at the intersection of Bayesian statistics and artificial intelligence. His research focuses on developing inference methods that push the boundaries of how uncertainty can be quantified and exploited in real world applications. During his PhD, he advanced Bayesian deep learning by introducing novel approaches using Sequential Monte Carlo, ensemble Kalman filtering, variational inference and Bayesian autoencoders, with applications ranging from generative modelling to state-space systems. In his postdoctoral work, he extended these ideas to interpretable models such as Bayesian decision trees, decision making under epistemic regret, and to reinforcement learning in complex, real-world environments. Vincent’s contributions bridge theory and practice, aiming to deliver algorithms that are both mathematically rigorous and impactful across domains where decision-making under uncertainty is critical.