Presentation Information
[18a-PA2-7]Recognition of All 26 Letters of the Alphabet in Optical Pattern Classification Using a Physical
Reservoir
〇(B)Koki Masuda1, Tsuyoshi Hasegawa1 (1.Waseda Univ.)
Keywords:
reservoir computing,physical reservoir
By employing a solid electrolyte with photoconductivity and ion conductivity as the reservoir layer, it is possible to classify light patterns directly irradiated onto the reservoir layer. This arises from diverse output changes caused by the enhancement of local photoconductivity and ion redistribution, both dependent on the irradiation pattern. Previously, high-precision classification has been achieved for limited character patterns. Therefore, this study focused on the number of recognizable patterns and conducted classification experiments for optically inputted 26 alphabet characters to evaluate their recognition performance.