Wave Interpolation Neural Operator: Interpolated Prediction of Electric Fields Across Untrained Wavelengths
Abstract
Existing surrogate solvers are limited to performing inference at fixed simulation conditions, such as wavelengths, and require retraining for different conditions. To address this, we propose Wave Interpolation Neural Operator (WINO), a novel surrogate solver enabling simulation condition interpolation across a continuous spectrum of broadband wavelengths. WINO introduces the Fourier Group Convolution Shuffling operator and a new conditioning method to efficiently predict electric fields from both trained and untrained wavelength data, achieving significant improvements in parameter efficiency and spectral interpolation performance.
Cite
Text
Seo et al. "Wave Interpolation Neural Operator: Interpolated Prediction of Electric Fields Across Untrained Wavelengths." NeurIPS 2024 Workshops: D3S3, 2024.Markdown
[Seo et al. "Wave Interpolation Neural Operator: Interpolated Prediction of Electric Fields Across Untrained Wavelengths." NeurIPS 2024 Workshops: D3S3, 2024.](https://mlanthology.org/neuripsw/2024/seo2024neuripsw-wave/)BibTeX
@inproceedings{seo2024neuripsw-wave,
title = {{Wave Interpolation Neural Operator: Interpolated Prediction of Electric Fields Across Untrained Wavelengths}},
author = {Seo, Joonhyuk and Kang, Chanik and Seo, Dongjin and Chung, Haejun},
booktitle = {NeurIPS 2024 Workshops: D3S3},
year = {2024},
url = {https://mlanthology.org/neuripsw/2024/seo2024neuripsw-wave/}
}