Abstract: I will describe a research program aimed at advancing the potential for discovery and interdisciplinary collaboration by approaching fundamental physics challenges through the lens of modern machine learning (ML). This research program has two complementary components. Ab initio simulations are a powerful tool of fundamental physics and the first component described in the talk will be the optimal combination of simulations with ML. The second component will focus on simulation-free problems where ML can be used to identify patterns in high-dimensional feature spaces that would be unfindable with traditional methods. I will give illustrative examples of physics analysis we can do now that would be unimaginable before deep learning!
PHY Colloquium