A Hopfield network uses recurrent connections to store memories in an artificial neural network. Its relationship to statistical mechanics helped connect the physics of collective behavior with models of computation.
121 words · auto-generated from the episode video
1:12:53in 1982, formulation of the Hopfield network. >> Yes, this was John Hopfield. He introduced the recurrent fully connected artificial neural network that functions to store memories. He was at Caltech at the time and then um later on went to Princeton and continued his work. He established this profound bridge between statistical mechanics and all of the math that we've used to describe magnets and things like that with the Eising model and computer science and this idea of artificial neural networks. It really started a lot of physicists to go down this track of hey maybe these artificial neural networks are something that we can make sense of. He won the two 2024 Nobel Prize in physics. Very recent.