Machine-Learning Algorithms for Nano-Material Characterization

The finite difference-time domain (FDTD) models for clusters of known substrate nano-particle distributions were developed. Additional models utilizing open source finite element analysis (FEA) software (ELMER) and commercial FEA software (COMSOL) were also developed. These models will be used to validate the FDTD models. The interface provides for a much more robust visualization of the output for the nano-particle light/matter interactions.

Novel techniques in the use of autoencoders (AE) for deep learning of salient features in large datasets of images were developed.