[22a-P07-1~6] 7 Beam Technology and Nanofabrication(Poster)

[22a-P07-1] Development of hyperparameter automatic tuning application on Bayes spectroscopy and application for HAXPES analysis

〇Masashi Yoshimura^1,2, Shingo Takeda^1,2, Yuichi Yokoyama³, Masaichiro Mizumaki³ (1.Hyogo Sci. and Tech. Assoc. SR Center, 2.SPring-8 Service, 3.JASRI)

Keywords:Bayesian inference, Automatic tuning, Model selection

Bayesian spectroscopy that estimates function model of experimental data by Bayesian inference have been proposed.To obtain good estimation, transition probability in searching model function parameter have to be controlled by tunning hyper parameter.We developed an application automatic tuning hyper parameters using transition probability.We analyzed laboratory-based HAXPES data using the developed application and repot the result of hyper parameter tuning and model selection.