Presentation Information
[19p-P06-7]Evaluation on recording error of Magnetic domains in Magneto-Optical Diffractive Deep Neural Networks using Magneto-Optical Effect
〇(M1C)Juri Ikeda1, Hotaka Sakaguchi1, Hirofumi Nonaka2, Hiroyuki Awano3, Fatima Zahra Chafi1, Takayuki Ishibashi1 (1.Nagaoka Univ. Tech, 2.Aichi Inst. Tech., 3.Toyota Tech. Inst.)
Keywords:
Magneto-Optical,neural network
Optical diffraction deep neural networks are a type of stacked optical neural network that is expected to perform calculations with low power consumption and high speed. However, the development of devices that can operate in visible light and are rewritable has been a challenge. We have proposed a magneto-optical diffractive deep neural network (MO-D2NN) that utilizes the magneto-optical effect of magnetic materials to solve these problems. In this study, we investigated the effect of recording error on the correct classification rate of handwritten digits in MO-D2NN.