One line of my research focuses on efficient and reliable probabilistic machine learning ``in the wild'', that is when a model has to perform inference and learning in one environment that might not correspond to the one it has been trained on, and for which reasoning over calibrated uncertainties is of primary importance. I also focus on combining classical statistical machine learning with complex (logical) reasoning, which is fundamental to enable trustworthy AI. To do so, I investigate how we can certify that a probabilistic model either delivers exactly what we expect or could provide bounds over all the computations needed to perform complex reasoning.
Research keywords:
deep generative models, complex reasoning, neuro-symbolic AI, reliable and efficient probabilistic inferece, automated machine learning
Theme:
Machine Learning and AI (inc. multi-agent systems)
Verification and Security
Website:
Email (optional - published on profile page):
University: