Stanford Seminar - Computing with Physical Systems

Explore the cutting-edge field of analog computing and Physical Neural Networks in this Stanford seminar presented by Peter McMahon from Cornell University. Delve into the concept of training complex physical systems to perform as neural networks for machine learning tasks, with experimental demonstrations across mechanical, electronic, and photonic systems. Discover the potential applications of this technology, including large-scale photonic accelerators for server-side machine learning, smart sensors for pre-processing signals, and new types of quantum neural networks. Learn about the limitations of conventional digital computing and the renaissance of analog computing across various physical substrates. Gain insights into future research directions and the possibilities of endowing analog physical systems with unexpected functionality.