# Two-Stage Deformable-Convolutional Inverse Design of Nanophotonic Absorbers from Optical Spectra

arXiv: https://arxiv.org/abs/2608.11860 · HTML: https://mlsanity.com/paper/2608.11860/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 21 references that are in the map.
- First posted on arXiv 12 Aug 2026.

## Builds on (most cited first, 9 of 9)

- [Deformable Convolutional Networks](https://mlsanity.com/paper/1703.06211/) (2017)
- [Least Squares Generative Adversarial Networks](https://mlsanity.com/paper/1611.04076/) (2016)
- [Deformable ConvNets v2: More Deformable, Better Results](https://mlsanity.com/paper/1811.11168/) (2018)
- [Dynamic Convolution: Attention over Convolution Kernels](https://mlsanity.com/paper/1912.03458/) (2019)
- [Involution: Inverting the Inherence of Convolution for Visual Recognition](https://mlsanity.com/paper/2103.06255/) (2021)
- [Deep neural networks for the evaluation and design of photonic devices](https://mlsanity.com/paper/2007.00084/) (2020)
- [Designing nanophotonic structures using conditional-deep convolutional generative adversarial networks](https://mlsanity.com/paper/1903.08432/) (2019)
- [Inverse Design of Diffractive Metasurfaces Using Diffusion Models](https://mlsanity.com/paper/2506.21748/) (2025)
- [Optimizing Spectral Prediction in MXene-Based Metasurfaces Through Multi-Channel Spectral Refinement and Savitzky-Golay Smoothing](https://mlsanity.com/paper/2602.08406/) (2026)

## Abstract (arXiv)

Data-driven inverse design enables efficient generation of nanophotonic structures with prescribed optical responses, but spectrum-to-geometry mapping remains challenging due to non-uniqueness and fine geometric features. This work presents a two-stage deformable-convolutional framework for reconstructing metal--insulator--metal resonator geometries from 80-dimensional absorption spectra. The spectrum is projected to a 150×4×4 latent representation and decoded into a 64×64 resonator mask. Training combines supervised reconstruction with least-squares adversarial refinement initialized from the best supervised checkpoint. A three-run ablation compares deformable convolution with plain convolution, involution, Dynamic Conv, and ODConv under the same architecture. The proposed model achieves 20.79±0.31~dB PSNR and 0.8501±0.0082 SSIM, improving over plain convolution by 2.16~dB and 0.0831, respectively. It further achieves Dice 0.9623±0.0027, IoU 0.9342±0.0038, and boundary F-score 0.9550±0.0027. Spectral consistency evaluated using a frozen forward surrogate yields RMSE 0.0805±0.0013 and R2=0.7923±0.0065. Learned offsets show stronger adaptive sampling at coarse and intermediate decoder stages. Overall, deformable sampling with supervised initialization and adversarial refinement improves spectrum-conditioned geometry reconstruction.