Considering this, out of LANL, impressed by their work, another of our enterprise connections. Will we able to print networks with for specific need?
Printing Circuits on Nanomagnets Yields New Breed of AI
IEEE Spectrum, Dexter Johnson, April 5, 2022
Los Alamos National Laboratory (LANL) researchers have developed an artificial spin glass made of nanomagnets that is configured to mimic a neural network, paving the way for artificial intelligence algorithms to be printed as physical hardware. The artificial spin glass is comprised of thin layers of iron-nickel alloy, and the positions and orientations of the microscopic bar magnets align with the interaction structure of an artificial neural network. LANL's Michael Saccone said their artificial spin glass, a proof-of-principle Hopfield neural network, is akin to a slide rule in which "the rules of the geometry encode simple arithmetic." The architecture of a Hopfield network and nanomagnetic system involves information flowing constantly between all nanomagnets in all directions. Saccone explained, "This takes a while for a sequential algorithm to simulate, but in a physical system there is no cost to the inherent parallelization. The universe just does its thing."
No comments:
Post a Comment