Analysis of the High-Frequency Magnetization Process Through Machine Learning and Topological Data Techniques

The study of the high-frequency magnetization process is crucial for advancing technologies like electric vehicles and 6G communication systems. We employ Topological Data Analysis (TDA) and machine learning to explore the magnetization process in materials at frequencies up to 100 KHz. Our approach involves analyzing the microstructures affecting magnetization dynamics, aiming to enhance power device efficiency. Utilizing a high-resolution magneto-optical Kerr effect (MOKE) microscope, we visualize magnetic domains at varying frequencies. Persistent homology, a TDA method, is used to convert topological features into vectors, which are then processed using Principal Component Analysis (PCA) to capture the significant information. This study also addresses the challenge of machine learning explainability in material science, correlating algorithm outputs with physical phenomena.