Precision Simulations using Machine Learning (Tilman Plehn)

Like many fields of numerical research, particle physics is being transformed by modern machine learning. In LHC physics, data science and machine learning provide a common language for theory and experiment, allowing us to make optimal use of the coming LHC runs. I will introduce some key ML-concepts for LHC applications, including uncertainty-aware networks and generative networks, and show how they can be used to cope with the simulation and analysis requirements at the LHC. An important aspect of modern ML is explainability, and I will give examples how we can meet the particle physics requirements for example defined by uncertainties and approximate analytic expressions.