报告题目：Accelerating the Design of Freeform and Reconfigurable Flat Optics using Deep Learning Techniques
报告人：Asst. Prof. Sensong An (University of North Texas)
摘要：Metasurfaces offer a unique combination of functionality and compactness in shaping optical wavefronts, outperforming traditional bulky geometric optics devices. However, designing meta-atoms to meet specific electromagnetic responses can be challenging and time-consuming due to the added complexity of freeform shapes and reconfigurable materials. In this presentation, we propose the use of deep neural networks to streamline the forward modeling and inverse design of meta-atoms with a large number of design degrees of freedom. The neural network frameworks are capable of generating a variety of designs for both freeform and reconfigurable metasurfaces, as demonstrated by several successful design examples.
报告人简介：Dr. An was born in Tianjin, China and received his undergraduate and graduate degrees from USTC in 2013 and 2016. He went on to receive his PhD in 2021 from the University of Massachusetts Lowell. After completing his PhD, Sensong worked as a postdoctoral researcher in the Material Science Department at MIT before joining META Reality Labs as a research scientist. He’s currently an Assistant Professor with the Department of Electrical Engineering at the University of North Texas. Dr. An’s research interests are in applied electromagnetics, specifically in the areas of passive microwave circuit and component design, meta-optic design, and inverse design approaches enabled by deep learning techniques.