AIAtomic-spin memory reportedly improves recall over the classical Hopfield model
A paper coauthor says a study published in Science uses atomic spins as neurons and light exchanged between them as synapses—the connections between neurons.
TLDR
A coauthor announced the Science paper “High-capacity associative memory in a quantum-optical spin glass,” describing memory recall in a Hopfield system built from atomic spins and light. The coauthor credits two effects with enhancing recall compared with the classical Hopfield model: different dynamics that enlarge memory basins, and changes in connectivity linked to atomic positions and spin orientations. They describe the changing connectivity as a direct analogue of short-term Hebbian synaptic plasticity and a precursor to learning in a quantum-optical system.
