AI Hardware and Real-world AI - JuliaCon 2024

Explore the intersection of AI hardware and real-world applications in this keynote address by Andrew Fitzgibbon, FRS, at JuliaCon 2024. Delve into the critical importance of efficient computational power usage in AI, spanning from theoretical understanding of machine learning models to transistor-level design. Examine case studies such as real-time computer vision on the Microsoft HoloLens HPU and training of massive AI models on Graphcore's IPU. Gain insights into empowering programmers to effectively communicate with hardware and designing frameworks and languages to bridge theory and practice. Cover diverse topics including mathematical optimization, automatic differentiation, programming languages, and silicon design in an accessible manner suitable for all computer enthusiasts.