|Title||Machine Learning for Exotic Metasurfaces|
|Publication Type||Conference Paper|
|Year of Publication||2020|
|Authors||Y Deng, S Ren, K Fan, JM Malof, and WJ Padilla|
|Conference Name||International Conference on Infrared, Millimeter, and Terahertz Waves, Irmmw Thz|
We introduce one deep neural network and an inverse method called neural adjoint method to solve high-dimensional metasurface inverse design problem. We find that, even when the solution is outside the geometry space, the neural adjoint method is still able to find a metasurface design with low error.