Presentation Information

[25a-P01-3]Development of hidden layers for MO-D2NN

〇Takuma Homma1, Hotaka Sakaguchi1, Jian Zhang1, Satoshi Sumi2, Hiroyuki Awao2, Hiroshi Nonaka3, Chafi Fatima Zahra1, Takayuki Ishibashi1 (1.Nagaoka Univ. of Tech., 2.Toyota Tech. Inst., 3.Aichi Inst. of Tech.)

Keywords:

magneto-optics,Magnetic Thin Film

Neural networks using physical phenomena are attracting attention in order to significantly reduce power consumption and speed up operations, which are problems in neural networks. We have proposed a magneto-optic diffractive deep neural network (MO-D2NN) using magneto-optic effects and reported on its ability to operate in visible light. In the present study, we investigated the fabrication conditions of bismuth-substituted magnetic garnets optimal for MO-D2NN and the conditions for writing magnetic domains.