[20a-F211-1~10] 31.1 Focused Session "AI Electronics"

[20a-F211-7] Real Neuromorphic Systems using Local Learning Rule and Thin-Film Devices - Working Confirmation of Letter Learning -

〇Mutsumi Kimura^1,2, Ryo Tanaka¹, Daiki Yamakawa¹, Yuki Shibayama¹, Hiroya Ikeda², Yuta Takishita², Yasuhiko Nakashima² (1.Ryukoku Univ., 2.NAIST)

Keywords:Real Neuromorphic Systems, Local Learning Rule, Thin-Film Device

Artificial intelligences are currently realized by software, but because of much waste to secure versatility, the size and power consumption are large. Therefore, we are focusing on real neuromorphic systems for brain-type integrated systems from the device level. The control circuit can be reduced by the local learning rule, and three-dimensional layered integration can be achieved by amorphous metal-oxide semiconductor thin-film devices. In this presentation, we succeeded in confirming the working operation of letter learning in Hopfield network with crosspoint synapse.