[11p-W810-1~17] 31.1 Focused Session "AI Electornics"

[11p-W810-9] Construction of a machine learning system for conductive polymer wire memristive device

〇Masaru Okada¹, Yasumasa Sugito¹, Naruki Hagiwara¹, Tetsuya Asai², Yuji Kuwahara¹, Megumi Akai-Kasaya^1,3 (1.Osaka Univ., 2.Hokkaido Univ., 3.JST PRESTO)

Keywords:neuromorphic, machine learning, conductive polymer

We demonstrate a building of neural network architecture hardware consisting of conducting We demonstrate a building of neural network architecture hardware consisting of conducting polymer, PEDOT:PSS, wires. The wire grows and bridges between electrodes that are immersed in monomer solution. The conductance increases and can be kept as a memristive resistive change memory. In this study, we fabricate an electrode array for polymer wire growth. The array consists of 90 pairs of electrodes on 2×2 cm^2 glass substrate that is bounded to an electric circuit board. The polymer wires can be taught to perform data encoding function realizing feature extraction of three 3×3 binary letters through un autoencoder that is an un-supervised learning.