EurekAlert  February 1, 2021 The quest to implement machine learning algorithms in hardware has focused on combining various materials to create device functionality. This approach limits functionality, efficiency, complicates scaling and on-chip learning. Researchers in the Netherlands created an atomic spin system that emulates a Boltzmann machine directly in the orbital dynamics of one well-defined material system. They fabricated the prerequisite tunable multi-well energy landscape by gating patterned atomic ensembles using scanning tunneling microscopy. The anisotropic behaviour of black phosphorus, provided plasticity with multi-valued and interlinking synapses that led to tunable probability distributions. They observed an autonomous reorganization of the […]