Title:  Building Meta-optic System through Machine Learning

Committee: 

Dr. Cai, Advisor   

Dr. Naeemi, Chair

Dr. Adibi

Abstract: The objective of the proposed research is to open up an entirely new vistas for the design and implementation of meta-optics, where deep learning enabled frameworks are developed and exploited for the inverse design of photonic structures in response to arbitrary defined, on-demand optical properties. The deep learning methodology consolidates the state-of-the-art deep learning techniques, traditional optimization approaches, and advanced pattern generation schemes in a concerted manner. I have utilized the deep learning framework in designing of functional metasurface for polarization conversion and beam deflection, as well as multifunctional meta-optic systems for beam shaping and holographic imaging. I will further explore deep-learning designed unconventional dielectric meta-optics, tunable meta-optics and meta-optic systems for all-optical computing.