Models As Code Differentiable Programming With Zygote

Summary

Scientific computing is increasingly incorporating the advancements in machine learning and the ability to work with large ... According to Max Haughton, the calculation of gradients is a way to understand the universe. For the entire history of computing, ... For 70 years, to program a computer meant one thing: tell it exactly what to do, step by step. That entire era is quietly ending ... A presentation on back-propagation and automatic differentiation, and demonstration of how this method is used for calibration in ... This talk was presented as part of JuliaCon 2021. Abstract: Deep learning has grown steadily and there has been rising interest ... Derivatives are at the heart of scientific Presenter: Gordon Plotkin Presented at POPL'2020. Yet another example from my demonstrative project on Chris Rackauckas (MIT), "Generalized Physics-Informed Learning through Language-Wide Talk from HSF/IRIS-HEP Analysis Ecosystem 2 Workshop ( indico.cern.ch/event/1125222/).