Bayesian Networks were originally invented for inference purposes, but have since become the leading approach for machine learning with uncertainties. I will introduce the structure and the training of such Bayesian networks and show how they can be applied to regression and classification tasks in jet physics. This will include a treatment of statistical uncertainties as well as systematics, through data augmentation. I will then introduce Bayesian generative networks, show how they can be applied to LHC simulations, and how they are crucial for
a comprehensive uncertainty treatment of network-based simulations.
P.-A. Delsart