[18a-D104-4~11] 10.3 Spin devices, magnetic memories and storages

▲ [18a-D104-10] Macro-magnetic simulation of reservoir computing utilizing spin-dynamics in magnetic tunnel junctions

〇(M1)Taishi Furuta¹, Fujii Keisuke², Nakajima Kohei³, Tsunegi Sumito⁴, Kubota Hitoshi⁴, Goto Minori^1,5, Suzuki Yoshishige^1,5, Miwa Shinji^1,5 (1.Osaka Univ., 2.Kyoto Univ., 3.The Univ. of Tokyo, 4.AIST, 5.CSRN-Osaka)

Keywords:Neural network, Magnetic tunnel junction

Reservoir computing is a kind of the neural network which is a mathematical model of machine learning for emulating a human brain. In the reservoir computing, we only optimize a linear transform matrix to obtain output from the non-linear elements, therefore it is easy to conduct learning. In this study, we conducted macro-magnetic simulation for the reservoir computing in magnetic tunnel junction to clarify the figure-of-merit.