Deep Learning of Input/Label Images for a Convolutional Neural Network (CNN) Utilizing Plasmonic Simulation Response of a Nanoparticle Sensor Substrate

G.R. Gallegos
New Mexico Highlands University,
United States

Keywords: Deep Learning, Materials Science, Plasmonics, Computational Physics, Nanoparticles


The current field of material science has opened its doors to innovative methods in computational physics, AI and machine learning to enhance both discovery and design of new materials and devices that utilize new materials. To this end, nanoparticles have entered the realm of materials which lend themselves to a subset of sensors particularly adept at measuring low levels of environmental contaminants, blood products and low-level electronic signals. Having said this, it is of interest to characterize and design sensors that take advantage of the unique properties found in nanoparticles. It is possible to employ novel methods in AI, machine learning and computational physics that augment the current experimental methods and improve the success of characterizing a wide variety of nanoparticle sensors as well as tightening the design loop time. This research employs methods in AI, machine learning and computational physics to reduce the time required to characterize/design plasmonic sensors. Utilizing experimental and simulated data, these methods are able to improve the traditional methods of experimental work and provide a more streamlined approach to the whole process.