[19a-Z32-1~10] 23.1 Joint Session N "Informatics"

[19a-Z32-10] Search for the hyperparameters of Bayesian optimization for materials synthesis

〇Ryo Nakayama¹, Ryota Shimizu^1,2, Taishi Haga¹, Takeshi Kimura¹, Ando Yasunobu³, Nobuaki Yasuo¹, Masakazu Sekijima¹, Taro Hitosugi¹ (1.Tokyo Tech, 2.JST-PRESTO, 3.AIST)

Keywords:Bayesian optimization

In the synthesis of new materials, it is necessary to optimize multiple synthesis conditions such as material composition, temperature, and pressure. Bayesian optimization is a powerful method for the optimization of synthesis condition by setting the hyperparameters at appropriate values. However, there is little discussion on the appropriate values of hyperparameters for material synthesis. In this study, we applied Bayesian optimization to a one-dimensional model function that mimics material synthesis, and optimized the hyperparameter.