Abstract

Reduced graphene oxide (rGO) exhibits strong anisotropic light absorption and high compatibility with photonic integrated chips, making it a promising material for implementing high-performance on-chip polarization-selective devices. This paper proposes and demonstrates a machine learning framework based on fully connected neural networks (FCNNs) to map the dependence of the polarizer figure of merit (FOM) on the waveguide geometry. Trained on a small dataset of low-resolution mode simulation results, the FCNN rapidly predicts FOM values across a large structural parameter space with high resolution, reducing computing time by more than 4 orders of magnitude and achieving an average deviation below 0.05.

Abstract

Reduced graphene oxide (rGO) exhibits strong anisotropic light absorption and high compatibility with photonic integrated chips, making it a promising material for implementing high-performance on-chip polarization-selective devices. This paper proposes and demonstrates a machine learning framework based on fully connected neural networks (FCNNs) to map the dependence of the polarizer figure of merit (FOM) on the waveguide geometry. Trained on a small dataset of low-resolution mode simulation results, the FCNN rapidly predicts FOM values across a large structural parameter space with high resolution, reducing computing time by more than 4 orders of magnitude and achieving an average deviation below 0.05.

Abstract

Reduced graphene oxide (rGO) exhibits strong anisotropic light absorption and high compatibility with photonic integrated chips, making it a promising material for implementing high-performance on-chip polarization-selective devices. This paper proposes and demonstrates a machine learning framework based on fully connected neural networks (FCNNs) to map the dependence of the polarizer figure of merit (FOM) on the waveguide geometry. Trained on a small dataset of low-resolution mode simulation results, the FCNN rapidly predicts FOM values across a large structural parameter space with high resolution, reducing computing time by more than 4 orders of magnitude and achieving an average deviation below 0.05.

Abstract

Reduced graphene oxide (rGO) exhibits strong anisotropic light absorption and high compatibility with photonic integrated chips, making it a promising material for implementing high-performance on-chip polarization-selective devices. This paper proposes and demonstrates a machine learning framework based on fully connected neural networks (FCNNs) to map the dependence of the polarizer figure of merit (FOM) on the waveguide geometry. Trained on a small dataset of low-resolution mode simulation results, the FCNN rapidly predicts FOM values across a large structural parameter space with high resolution, reducing computing time by more than 4 orders of magnitude and achieving an average deviation below 0.05.