Predictive Modelling of Optical Beams From Grating Structure Using Deep Neural Network
Yu Dian Lim, Peng Zhao, L. Guidoni, J.-P. Likforman, Chuan Seng Tan
Abstract
Integrated grating structure has been widely used in the optical addressing of trapped ion qubits in quantum computing. For accurate optical addressing, the optical properties of light beam coupled out from the grating should be thoroughly understood. In this study, deep neural network (DNN) modeling is used to predict the optical properties of light from silicon nitride (SiN) grating. DNN models with various number of layers (L) and nodes per layer (N) are attempted and optimized. Both overfitted and well-fitted L/N combinations are addressed. The APE values of the overfitted DNNs can reach as low as 5.2%, while the APE values of the well-fitted DNN reaches as low as 7.2%.
Topics & Concepts
GratingArtificial neural networkOpticsComputer scienceDiffraction gratingMaterials scienceOptoelectronicsArtificial intelligencePhysicsNeural Networks and Reservoir ComputingPhotonic and Optical DevicesAdvanced Optical Sensing Technologies