[15p-P06-1~15] 23 Joint Session N "Informatics" (Poster)

[15p-P06-10] Development of a physics-informed conversion model of magnetic structure and energy using explainable AI

〇(B)Kanji Tsubouchi¹, Ryunosuke Nagaoka¹, Michiki Taniwaki¹, Yotaro Machida¹, Alexandre Lira Foggiatto¹, Masato Kotsugi¹ (1.Tokyo Univ. of Sci.)

Keywords:

magnetic material,spin texture,deep learning

Spin textures are a crucial source of information that governs macroscopic magnetic properties. Advancing magnetic devices requires an understanding of the physical mechanisms underlying magnetic structures. Thus, analyzing causal relationships is important for both fundamental and applied research. However, due to the complex shapes of magnetic structures, their quantification remains challenging. To address this, this study aims to develop a quantitative and interpretable model by leveraging explainable AI. In particular, to focus on the energy underlying the functionality, we construct a learning model using AutoEncoder.