Review of candidate devices for neuromorphic applications

Abstract

Artificial intelligence technology has attracted much attention in recent years, and technological progress of this technology is anticipated with the development of semiconductor technology. This talk focuses on synaptic mimic devices to realize artificial intelligence with semiconductor memory technology. These synaptic devices affect cognitive accuracy along with conductance quantization and architecture. Therefore, we will first discuss from the architectural point of view and examine the characteristics of candidates for various synapse devices being reported. In particular, we concentrate on synaptic imitation devices that creatively use the functions of several flash memory devices. Finally, we discuss device variation and IR drop along metal wires as common challenges for synaptic devices, and how neuromorphic technology will evolve.