PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
The rich physics present in spintronic materials and devices provide a wide range of opportunities for neuromorphic computing systems. This presentation will overview three distinct proposals to efficiently leverage spintronic phenomena to mimic the behavior of the human brain. The presentation will begin with a purely-spintronic neuromorphic system in which the analog states of magnetic domain wall neurons and synapses emulate neurobiological behavior and enable CMOS-free inference and unsupervised online learning. This will be followed by an approach for neuromorphic inference that marks the first experimental demonstration of a neuromorphic network directly implemented with MTJ synapses, as well as a solution for exploiting stochastic MTJ switching for unsupervised learning in concert with CMOS neuron circuits. This presentation will conclude with a reservoir computing system based on the dynamics of irregular arrays of frustrated nanomagnets that increases overall hardware efficiency by a factor of 10,000,000.
Joseph S. Friedman
"Spintronic neuromorphic computing: Domain walls, stochasticity, and frustration", Proc. SPIE PC12656, Spintronics XVI, PC1265613 (5 October 2023); https://doi.org/10.1117/12.2682719
ACCESS THE FULL ARTICLE
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
The alert did not successfully save. Please try again later.
Joseph S. Friedman, "Spintronic neuromorphic computing: Domain walls, stochasticity, and frustration," Proc. SPIE PC12656, Spintronics XVI, PC1265613 (5 October 2023); https://doi.org/10.1117/12.2682719