Inverse Design of Metamaterials and Photonic Crystals Using Machine Learning

We present an overview of our work on inverse design of metamaterials and photonic crystals. In the past few years, we have developed neural networks that can provide new designs for nanophotonic structures with predefined optical response. In particular, we have developed a CGAN network that can provide lithographic masks for a meta-atom with desired transmission and reflection properties. We illustrate our machine-learning-based inverse design with examples of metasurfaces with refractive properties, metasurfaces with meta-atoms with interdependent properties, nonlinear photonic-crystal waveguides, and photonic-crystal membranes for optomechanical resonators.