Time：2024-5-9 14:00-15:00

Location：ZJUI B407

Recently, research in deep learning techniques has attracted much attention. It hasalso been applied to electromagnetic engineering. Data-driven approaches allow machines to “learn” from a large amount of data and “master” the physical laws in certain controlled boundary conditions. However, this technique also faces many challenges, such as inaccuracy, limited generalization ability, etc.
In electromagnetic engineering, physical laws, i.e., Maxwell’s equations, set major guidelines in research and development. They discover the nature of electromagnetic fields and waves and are universal across various scenarios. Incorporating physical principles into the deep learning framework significantly improves the learning capacity and generalization ability of deep neural networks, hence increases the accuracy and reliability of deep learning techniques in modeling electromagnetic phenomena. In this talk, we will study several techniques to embed physical simulation into deep learning to model electromagnetic wave propagation. With the help of both physical simulation and deep learning, we can improve the accuracy and computational efficiency of electromagnetic modeling and data inversion. Hybridizing fundamental physical principles with “knowledge” from big data could help electromagnetic technologies be more automatic, accurate, and reliable.