[S-TT37] 最先端ベイズ統計学が拓く地震ビッグデータ解析

[STT37-04] ベイズ的動的モード分解

★招待講演

*日野 英逸¹ (1.大学共同利用機関法人情報・システム研究機構　統計数理研究所)

キーワード：ベイズモデリング、モード分解

Dynamic mode decomposition (DMD) is a data-driven deterministic method that has substantially contributed to our understanding of dynamical systems. In this work, a Bayesian formulation of DMD is proposed. It first determines the subspace of observables, and then compute the modes on that subspace. Variational matrix factorization makes it possible to realize a fully-Bayesian scheme of DMD. The proposed Bayesian DMD is capable of dealing with incomplete or missing data, which demonstrates the advantage of probabilistic modeling. Finally, both nonlinear simulated and real-world datasets are used to illustrate the potential of the proposed method. This is joint work with Mr. Takahiro Kawashima and Dr. Hayaru Shouno.