Cellular Neural Network using IGZO Thin Film as Synapses and LSI as Neurons

Artificial neural networks are infomation processing models of human brains. Especially, hardware neural networks have functions such as robustness, parallel distributed processing, and low power consumption. We think hardware neural networks are superior to software neural networks. However, hardware neural networks need high integration. Therefore, we propose a neural network using a LSI and IGZO thin films. We used current value as the synaptic bond strength. It is found that current value of synapses using IGZO thin films gradually decreased as time gose by. The neural networks succeeded in letter recognition.