Inverse design of unparametrized nanostructures by generating images from spectra

Abstract

Recently, there has been an increasing number of studies applying machine learning techniques for the design of nanostructures. Most of these studies train a deep neural network (DNN) to approximate the highly nonlinear function of the underlying physical mapping between spectra and nanostructures. At the end of training, the DNN allows an on-demand design of nanostructures, i.e., the model can infer nanostructure geometries for desired spectra. While these approaches have presented a new paradigm, they are limited in the complexity of the structures proposed, often bound to parametric geometries. Here we introduce spectra2pix, which is a DNN trained to generate 2D images of the target nanostructures. By predicting an image, our model architecture is not limited to a closed set of nanostructure shapes, and can be trained for the design of a much wider space of geometries. We show, for the first time, to the best of our knowledge, a successful generalization ability, by designing completely unseen shapes of geometries. We attribute the successful generalization to the ability of a pixel-wise architecture to learn local properties of the meta-material, therefore mimicking faithfully the underlying physical process. Importantly, beyond synthetical data, we show our model generalization capability on real experimental data.

Deep learning in optics—a tutorial

“…For the parameterized case, since we have some vector with a set of parameters as output, the FC network is the obvious choice [160][161][162]. For the unparametrized case, where we have an image of the unit cell, convolutional layers will be required [155,163,164]. Both choices can cause design challenges.…”

“…Parameterized designs can greatly limit the versatility of your training set while unparametrized designs are limited by the spatial sampling frequency of the grid of choice, and there is no clear-cut guideline to define the right approach. In figure 9 we can see an example of such a dilemma, both 9(a) [154] and 9(b) [155] deal with the same inverse design problem but the first uses an FC network, whereas the latter uses a decoder that outputs the image of the unit-cell. In our view, the solution to this dilemma lies in the choice of the nanostructure.…”

Intelligent metaphotonics empowered by machine learning

“…Such techniques extend the expressivity of this approach, significantly extending the range of possible cell geometries. Conversely, the DL approach can also be used to restore a unit cell geometry from a given spectra [56]. The design procedure may also incorporate some transfer learning between meta-atoms of various shapes [57].…”

“…DL also opens the route for the design of biology-inspired devices such as moth-eye structures [73] with the designed average absorption reaching 90% in the range from 400 nm to 1600 nm. Scattering properties are a subject of many design procedures [56,[74][75][76][77][78], including those devoted to the development of anisotropic [79] and bianisotropic [80] metasurfaces, as well as switchable reflectors [59]. DL was also exploited for achieving electromagnetically-induced transparency [81][82][83].…”

Deep Learning for Inverse Design of Broadband Quasi-Yagi Antenna

“…Sahedian et al used CNN to collect spatial information from the images for plasmonic structures design [20]. Malkiel et al published a series of researches for inverse design of the nanophotonic structures [21][22][23]. By continuously improving the network, their CNN approach could generate 2D images of the target nanostructures with desired spectra.…”