Multilayer Cross-Point Device Using IGZO as Synapses in Artificial Neural Networks

Abstract

We have developed a multilayer cross-point device using In-Ga-Zn-O semiconductor for synapse elements. There are 200 synapses on a glass substrates. We evaluate the change in the current value of the synapse. The current value gradually degrades by flowing current. The characteristic is available for modified Hebban learning.