"Machine Learning for Adaptive Optics: from a shinny past to a bright future"

In this talk, after a quick introduction on the various flavors of Machine Learning (ML), i will cover several applications to adaptive optics (AO) systems, from model-based ML techniques used to fine tune a number of parameter for optimized operations, to modern data-driven Deep Learning techniques applied to AO real-time control and image reconstruction. I will show how MAVIS requirements and specifications are motivating a full fledged R&D program on these topics.